AI & Automation prompts

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Map a repeated task and identify safe, high-impact opportunities for AI assistance.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate recurring work for automation potential using frequency, variability, error cost, and data sensitivity. Distinguish tasks suited to simple rules from those needing judgment or a clearer process first.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A scored opportunity list, one recommended pilot, a manual baseline, and explicit reasons to leave some tasks human-led. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Map a repeated task and identify safe, high-impact opportunities for AI assistance.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Map a repeated task and identify safe, high-impact opportunities for AI assistance.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Map a repeated task and identify safe, high-impact opportunities for AI assistance.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Map a repeated task and identify safe, high-impact opportunities for AI assistance.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a strong reusable instruction with input fields, output specification, and examples.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a strong reusable instruction with input fields, output specification, and examples.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a strong reusable instruction with input fields, output specification, and examples.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a strong reusable instruction with input fields, output specification, and examples.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a strong reusable instruction with input fields, output specification, and examples.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Outline a workflow to classify messages and draft responses with human approval.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Outline a workflow to classify messages and draft responses with human approval.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Outline a workflow to classify messages and draft responses with human approval.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Outline a workflow to classify messages and draft responses with human approval.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Outline a workflow to classify messages and draft responses with human approval.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create a source-aware research workflow that checks evidence and records uncertainty.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create a source-aware research workflow that checks evidence and records uncertainty.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create a source-aware research workflow that checks evidence and records uncertainty.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create a source-aware research workflow that checks evidence and records uncertainty.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create a source-aware research workflow that checks evidence and records uncertainty.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a repeatable system that summarizes files while preserving important caveats.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a repeatable system that summarizes files while preserving important caveats.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a repeatable system that summarizes files while preserving important caveats.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a repeatable system that summarizes files while preserving important caveats.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Design a repeatable system that summarizes files while preserving important caveats.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Plan an assistant that suggests answers from verified help documentation and escalates tricky cases.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Plan an assistant that suggests answers from verified help documentation and escalates tricky cases.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Plan an assistant that suggests answers from verified help documentation and escalates tricky cases.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Plan an assistant that suggests answers from verified help documentation and escalates tricky cases.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Plan an assistant that suggests answers from verified help documentation and escalates tricky cases.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Transform meeting transcripts into decisions, owners, deadlines, and unresolved questions.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Transform meeting transcripts into decisions, owners, deadlines, and unresolved questions.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Transform meeting transcripts into decisions, owners, deadlines, and unresolved questions.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Transform meeting transcripts into decisions, owners, deadlines, and unresolved questions.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Transform meeting transcripts into decisions, owners, deadlines, and unresolved questions.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create checks for factual accuracy, originality, brand fit, accessibility, and privacy.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create checks for factual accuracy, originality, brand fit, accessibility, and privacy.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create checks for factual accuracy, originality, brand fit, accessibility, and privacy.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create checks for factual accuracy, originality, brand fit, accessibility, and privacy.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Create checks for factual accuracy, originality, brand fit, accessibility, and privacy.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Connect form submissions, spreadsheets, and notifications with clear trigger/error handling.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Connect form submissions, spreadsheets, and notifications with clear trigger/error handling.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Connect form submissions, spreadsheets, and notifications with clear trigger/error handling.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Connect form submissions, spreadsheets, and notifications with clear trigger/error handling.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Connect form submissions, spreadsheets, and notifications with clear trigger/error handling.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Set up a small evaluation comparing AI-assisted and manual work on quality, time, and cost.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — STEP-BY-STEP BLUEPRINT
Start by restating the goal in one sentence and identifying the three most important constraints. Develop a step-by-step approach ordered by impact and effort. Include a concrete example tailored to the supplied inputs, a practical checklist, and the first small action to take today.

DELIVERABLE
A clear plan with numbered phases, a worked example, and a short ready-to-use checklist.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Set up a small evaluation comparing AI-assisted and manual work on quality, time, and cost.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — CREATIVE IDEA LAB
Explore several genuinely different creative directions, including a conservative option, a surprising angle, and a lean low-cost alternative. Explain who each approach fits and what makes it different. Select the strongest direction and refine it with a memorable example.

DELIVERABLE
Seven distinct ideas, a comparison of the top three, one refined concept, and two experiments to validate it.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Set up a small evaluation comparing AI-assisted and manual work on quality, time, and cost.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — EXPERT CRITIQUE & UPGRADE
Approach this like a candid expert reviewer. Diagnose common weak points and hidden assumptions. Prioritize problems by likely impact and ease of fixing them. Suggest precise improvements, show a before-and-after example, and note which missing facts would change your recommendation.

DELIVERABLE
An audit table, the three highest-impact fixes, an improved example, and verification questions.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Set up a small evaluation comparing AI-assisted and manual work on quality, time, and cost.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — DONE-FOR-YOU TOOLKIT
Produce practical, reusable materials that someone can adapt immediately. Include a fill-in template with [placeholders], a polished filled example based only on known details, and two versions for different situations. Make each item self-contained and easy to copy.

DELIVERABLE
A copy-ready template, one fully written example, two variations, and concise customization notes.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect and responsible AI workflow designer.

MY PROJECT
I want to: Set up a small evaluation comparing AI-assisted and manual work on quality, time, and cost.

CONTEXT TO USE
[Describe your specific situation here.] Include what you know about: current workflow, tools, data sensitivity, budget, and success criteria. If essential details are missing, ask up to three targeted questions; otherwise state reasonable assumptions and move forward.

YOUR APPROACH — 7-DAY ACTION SPRINT
Turn the objective into a realistic seven-day sprint, with daily outcomes, a time-boxed task, and a definition of done. Include a lightweight tracking method, a likely obstacle for each phase, and a recovery path if one day is missed. Prefer momentum over perfection.

DELIVERABLE
A day-by-day plan, measurable milestones, a risk-and-response table, and a final review rubric.

QUALITY STANDARD
Include human review, privacy safeguards, and failure paths where needed. Be specific rather than generic. Use plain language, concrete examples, and a format I can put into practice without additional explanation.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate recurring work for automation potential using frequency, variability, error cost, and data sensitivity. Distinguish tasks suited to simple rules from those needing judgment or a clearer process first.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A scored opportunity list, one recommended pilot, a manual baseline, and explicit reasons to leave some tasks human-led. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate recurring work for automation potential using frequency, variability, error cost, and data sensitivity. Distinguish tasks suited to simple rules from those needing judgment or a clearer process first.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A scored opportunity list, one recommended pilot, a manual baseline, and explicit reasons to leave some tasks human-led. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate recurring work for automation potential using frequency, variability, error cost, and data sensitivity. Distinguish tasks suited to simple rules from those needing judgment or a clearer process first.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A scored opportunity list, one recommended pilot, a manual baseline, and explicit reasons to leave some tasks human-led. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate recurring work for automation potential using frequency, variability, error cost, and data sensitivity. Distinguish tasks suited to simple rules from those needing judgment or a clearer process first.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A scored opportunity list, one recommended pilot, a manual baseline, and explicit reasons to leave some tasks human-led. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate recurring work for automation potential using frequency, variability, error cost, and data sensitivity. Distinguish tasks suited to simple rules from those needing judgment or a clearer process first.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A scored opportunity list, one recommended pilot, a manual baseline, and explicit reasons to leave some tasks human-led. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a system that classifies incoming messages and proposes the next action. Define categories, ambiguous cases, escalation rules, and a review queue without automatically sending sensitive replies.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A decision tree, field schema, sample classifications, and a safe rollout plan with human approval points. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a system that classifies incoming messages and proposes the next action. Define categories, ambiguous cases, escalation rules, and a review queue without automatically sending sensitive replies.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A decision tree, field schema, sample classifications, and a safe rollout plan with human approval points. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a system that classifies incoming messages and proposes the next action. Define categories, ambiguous cases, escalation rules, and a review queue without automatically sending sensitive replies.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A decision tree, field schema, sample classifications, and a safe rollout plan with human approval points. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a system that classifies incoming messages and proposes the next action. Define categories, ambiguous cases, escalation rules, and a review queue without automatically sending sensitive replies.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A decision tree, field schema, sample classifications, and a safe rollout plan with human approval points. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a system that classifies incoming messages and proposes the next action. Define categories, ambiguous cases, escalation rules, and a review queue without automatically sending sensitive replies.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A decision tree, field schema, sample classifications, and a safe rollout plan with human approval points. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a system that classifies incoming messages and proposes the next action. Define categories, ambiguous cases, escalation rules, and a review queue without automatically sending sensitive replies.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A decision tree, field schema, sample classifications, and a safe rollout plan with human approval points. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Map a form submission into a useful downstream workflow. Include validation, missing information, duplicate detection, ownership, status changes, and clear communication when processing fails.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A workflow diagram in text, data mapping, failure cases, and an implementation checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Map a form submission into a useful downstream workflow. Include validation, missing information, duplicate detection, ownership, status changes, and clear communication when processing fails.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A workflow diagram in text, data mapping, failure cases, and an implementation checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Map a form submission into a useful downstream workflow. Include validation, missing information, duplicate detection, ownership, status changes, and clear communication when processing fails.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A workflow diagram in text, data mapping, failure cases, and an implementation checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Map a form submission into a useful downstream workflow. Include validation, missing information, duplicate detection, ownership, status changes, and clear communication when processing fails.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A workflow diagram in text, data mapping, failure cases, and an implementation checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Map a form submission into a useful downstream workflow. Include validation, missing information, duplicate detection, ownership, status changes, and clear communication when processing fails.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A workflow diagram in text, data mapping, failure cases, and an implementation checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Map a form submission into a useful downstream workflow. Include validation, missing information, duplicate detection, ownership, status changes, and clear communication when processing fails.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A workflow diagram in text, data mapping, failure cases, and an implementation checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design an assistant that turns provided meeting notes into proposed actions, decisions, and open questions. Distinguish explicit commitments from inferred suggestions and preserve links to supporting evidence.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
An extraction schema, sample output, review checklist, and rules for handling unclear ownership or deadlines. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design an assistant that turns provided meeting notes into proposed actions, decisions, and open questions. Distinguish explicit commitments from inferred suggestions and preserve links to supporting evidence.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
An extraction schema, sample output, review checklist, and rules for handling unclear ownership or deadlines. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design an assistant that turns provided meeting notes into proposed actions, decisions, and open questions. Distinguish explicit commitments from inferred suggestions and preserve links to supporting evidence.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
An extraction schema, sample output, review checklist, and rules for handling unclear ownership or deadlines. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design an assistant that turns provided meeting notes into proposed actions, decisions, and open questions. Distinguish explicit commitments from inferred suggestions and preserve links to supporting evidence.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
An extraction schema, sample output, review checklist, and rules for handling unclear ownership or deadlines. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design an assistant that turns provided meeting notes into proposed actions, decisions, and open questions. Distinguish explicit commitments from inferred suggestions and preserve links to supporting evidence.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
An extraction schema, sample output, review checklist, and rules for handling unclear ownership or deadlines. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design an assistant that turns provided meeting notes into proposed actions, decisions, and open questions. Distinguish explicit commitments from inferred suggestions and preserve links to supporting evidence.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
An extraction schema, sample output, review checklist, and rules for handling unclear ownership or deadlines. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a document intake process that organizes, labels, and routes files consistently. Account for unsupported formats, sensitive material, duplicate uploads, uncertain classification, and retention requirements.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A processing sequence, metadata schema, review queue design, and a failure-recovery plan. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a document intake process that organizes, labels, and routes files consistently. Account for unsupported formats, sensitive material, duplicate uploads, uncertain classification, and retention requirements.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A processing sequence, metadata schema, review queue design, and a failure-recovery plan. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a document intake process that organizes, labels, and routes files consistently. Account for unsupported formats, sensitive material, duplicate uploads, uncertain classification, and retention requirements.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A processing sequence, metadata schema, review queue design, and a failure-recovery plan. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a document intake process that organizes, labels, and routes files consistently. Account for unsupported formats, sensitive material, duplicate uploads, uncertain classification, and retention requirements.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A processing sequence, metadata schema, review queue design, and a failure-recovery plan. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a document intake process that organizes, labels, and routes files consistently. Account for unsupported formats, sensitive material, duplicate uploads, uncertain classification, and retention requirements.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A processing sequence, metadata schema, review queue design, and a failure-recovery plan. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a document intake process that organizes, labels, and routes files consistently. Account for unsupported formats, sensitive material, duplicate uploads, uncertain classification, and retention requirements.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A processing sequence, metadata schema, review queue design, and a failure-recovery plan. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Plan a repeatable report assembled from defined sources. Specify freshness checks, data transformations, missing-value handling, narrative boundaries, and review before distribution to others.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A source-to-output map, validation checks, report template, and a monitoring checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Plan a repeatable report assembled from defined sources. Specify freshness checks, data transformations, missing-value handling, narrative boundaries, and review before distribution to others.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A source-to-output map, validation checks, report template, and a monitoring checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Plan a repeatable report assembled from defined sources. Specify freshness checks, data transformations, missing-value handling, narrative boundaries, and review before distribution to others.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A source-to-output map, validation checks, report template, and a monitoring checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Plan a repeatable report assembled from defined sources. Specify freshness checks, data transformations, missing-value handling, narrative boundaries, and review before distribution to others.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A source-to-output map, validation checks, report template, and a monitoring checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Plan a repeatable report assembled from defined sources. Specify freshness checks, data transformations, missing-value handling, narrative boundaries, and review before distribution to others.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A source-to-output map, validation checks, report template, and a monitoring checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Plan a repeatable report assembled from defined sources. Specify freshness checks, data transformations, missing-value handling, narrative boundaries, and review before distribution to others.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A source-to-output map, validation checks, report template, and a monitoring checklist. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Review an automation for silent failures, partial completion, retries, and unintended duplicate actions. Develop realistic failure scenarios and show how a person would notice, pause, and recover.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A failure-mode table, alert rules, recovery runbook, and a small test plan before wider use. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Review an automation for silent failures, partial completion, retries, and unintended duplicate actions. Develop realistic failure scenarios and show how a person would notice, pause, and recover.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A failure-mode table, alert rules, recovery runbook, and a small test plan before wider use. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Review an automation for silent failures, partial completion, retries, and unintended duplicate actions. Develop realistic failure scenarios and show how a person would notice, pause, and recover.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A failure-mode table, alert rules, recovery runbook, and a small test plan before wider use. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Review an automation for silent failures, partial completion, retries, and unintended duplicate actions. Develop realistic failure scenarios and show how a person would notice, pause, and recover.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A failure-mode table, alert rules, recovery runbook, and a small test plan before wider use. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Review an automation for silent failures, partial completion, retries, and unintended duplicate actions. Develop realistic failure scenarios and show how a person would notice, pause, and recover.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A failure-mode table, alert rules, recovery runbook, and a small test plan before wider use. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Review an automation for silent failures, partial completion, retries, and unintended duplicate actions. Develop realistic failure scenarios and show how a person would notice, pause, and recover.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A failure-mode table, alert rules, recovery runbook, and a small test plan before wider use. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a retrieval workflow that answers from approved documents with traceable references. Specify search inputs, permissions, relevance checks, stale content handling, and what to do when evidence is insufficient.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A retrieval specification, answer format, evaluation examples, and escalation rules for unsupported questions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a retrieval workflow that answers from approved documents with traceable references. Specify search inputs, permissions, relevance checks, stale content handling, and what to do when evidence is insufficient.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A retrieval specification, answer format, evaluation examples, and escalation rules for unsupported questions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a retrieval workflow that answers from approved documents with traceable references. Specify search inputs, permissions, relevance checks, stale content handling, and what to do when evidence is insufficient.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A retrieval specification, answer format, evaluation examples, and escalation rules for unsupported questions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a retrieval workflow that answers from approved documents with traceable references. Specify search inputs, permissions, relevance checks, stale content handling, and what to do when evidence is insufficient.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A retrieval specification, answer format, evaluation examples, and escalation rules for unsupported questions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a retrieval workflow that answers from approved documents with traceable references. Specify search inputs, permissions, relevance checks, stale content handling, and what to do when evidence is insufficient.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A retrieval specification, answer format, evaluation examples, and escalation rules for unsupported questions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Design a retrieval workflow that answers from approved documents with traceable references. Specify search inputs, permissions, relevance checks, stale content handling, and what to do when evidence is insufficient.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A retrieval specification, answer format, evaluation examples, and escalation rules for unsupported questions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a review interface or process for machine-generated suggestions. Explain what context reviewers need, how to record corrections, and which actions require explicit approval rather than passive acceptance.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A queue schema, review criteria, example cases, and a feedback loop that improves future suggestions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a review interface or process for machine-generated suggestions. Explain what context reviewers need, how to record corrections, and which actions require explicit approval rather than passive acceptance.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A queue schema, review criteria, example cases, and a feedback loop that improves future suggestions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a review interface or process for machine-generated suggestions. Explain what context reviewers need, how to record corrections, and which actions require explicit approval rather than passive acceptance.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A queue schema, review criteria, example cases, and a feedback loop that improves future suggestions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a review interface or process for machine-generated suggestions. Explain what context reviewers need, how to record corrections, and which actions require explicit approval rather than passive acceptance.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A queue schema, review criteria, example cases, and a feedback loop that improves future suggestions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a review interface or process for machine-generated suggestions. Explain what context reviewers need, how to record corrections, and which actions require explicit approval rather than passive acceptance.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A queue schema, review criteria, example cases, and a feedback loop that improves future suggestions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Create a review interface or process for machine-generated suggestions. Explain what context reviewers need, how to record corrections, and which actions require explicit approval rather than passive acceptance.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A queue schema, review criteria, example cases, and a feedback loop that improves future suggestions. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate an automation using time saved, error rate, maintenance effort, user trust, and recovery cost. Compare against the previous process and avoid declaring success from raw activity counts.

THE SETTING — A SOLO OPERATOR
Assume one person maintains the workflow alongside normal work. Favor simple tools, visible state, straightforward recovery, and low maintenance over impressive but fragile complexity.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Begin with the central decision and the supplied constraints. Work through the task in a clear sequence. Show one concrete example before giving the reusable final result. Explain any important dependency that changes the order of work.

WHAT TO DELIVER
A measurement plan, baseline template, pilot review agenda, and go-adjust-stop decision criteria. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate an automation using time saved, error rate, maintenance effort, user trust, and recovery cost. Compare against the previous process and avoid declaring success from raw activity counts.

THE SETTING — A SMALL TEAM
Account for shared ownership, handoffs, vacations, and different technical confidence. Make responsibilities and escalation routes explicit so the automation does not depend on one expert.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Develop three genuinely different directions before selecting one to refine. Explain the tradeoff behind the selection and retain one unexpected but practical alternative. Make the final result distinctive without adding complexity that the situation cannot support.

WHAT TO DELIVER
A measurement plan, baseline template, pilot review agenda, and go-adjust-stop decision criteria. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate an automation using time saved, error rate, maintenance effort, user trust, and recovery cost. Compare against the previous process and avoid declaring success from raw activity counts.

THE SETTING — LOW-CODE TOOLS
Design around forms, spreadsheets, documented connectors, and simple condition rules. Describe capabilities generically until actual tool availability and permissions are confirmed.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Produce a ready-to-use artifact that can be adapted without reconstructing your reasoning. Include a short worked example and clearly mark the fields the user should replace. Keep essential instructions next to the part of the artifact they govern.

WHAT TO DELIVER
A measurement plan, baseline template, pilot review agenda, and go-adjust-stop decision criteria. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate an automation using time saved, error rate, maintenance effort, user trust, and recovery cost. Compare against the previous process and avoid declaring success from raw activity counts.

THE SETTING — SENSITIVE INTERNAL INFORMATION
Minimize the data processed and shared, define access boundaries, and keep logs useful without exposing content unnecessarily. Include explicit review before external transmission.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Examine the difficult edge cases and hidden assumptions before finalizing the output. Compare a conventional approach with a credible alternative. Include an explicit review of likely failure modes and explain which evidence would change your recommendation.

WHAT TO DELIVER
A measurement plan, baseline template, pilot review agenda, and go-adjust-stop decision criteria. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate an automation using time saved, error rate, maintenance effort, user trust, and recovery cost. Compare against the previous process and avoid declaring success from raw activity counts.

THE SETTING — AN UNRELIABLE INPUT SOURCE
Assume incoming data may be incomplete, inconsistent, duplicated, or delayed. Define validation, quarantine, retry, and human correction before downstream actions occur.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Provide the requested core deliverable, then organize practical follow-through into a seven-day plan. Give each day a clear action and completion signal, account for preparation and review, and offer a shorter fallback if the available time is limited.

WHAT TO DELIVER
A measurement plan, baseline template, pilot review agenda, and go-adjust-stop decision criteria. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.

Act as an automation architect who values reliability and human judgment.

THE TASK
Evaluate an automation using time saved, error rate, maintenance effort, user trust, and recovery cost. Compare against the previous process and avoid declaring success from raw activity counts.

THE SETTING — A MEASURED PILOT
Start with a narrow, reversible pilot and a manual fallback. Define success criteria, a review window, and the evidence needed before expanding the workflow.

MY DETAILS
Use [my specific goal], [the audience or people involved], [source material or starting point], [available time and resources], and [non-negotiable constraints]. Treat these as editable inputs. If an essential detail is missing, ask one focused question; otherwise state reasonable assumptions and proceed.

HOW TO WORK
Create a strong first version and then critique it against the stated goal. Revise the weakest part rather than rewriting everything. Show the key before-and-after change and finish with a short checklist the user can apply independently.

WHAT TO DELIVER
A measurement plan, baseline template, pilot review agenda, and go-adjust-stop decision criteria. Tailor every section to the setting above and the details I supply. Make the final artifact usable, not just a list of general suggestions.

QUALITY CHECK
Protect sensitive information, use least privilege, define failure paths, and require human review for consequential external actions. Do not assume integrations or permissions that are unavailable. Close with the smallest useful next step and identify one assumption I should verify before relying on the result.