Define the goal, context, constraints, examples and output format, then evaluate the result against explicit quality criteria.
Guided workflow25 minute route
I want more useful AI results
Improve AI results by choosing an appropriate task, defining success, supplying relevant context and reviewing output through a repeatable workflow.
Quick answer
The safest way to start
Define the goal, context, constraints, examples and output format, then evaluate the result against explicit quality criteria.
Capture goal, audience, context, constraints, output format and evaluation criteria in a few lines.
Do not rely on role-play phrases alone or accept a fluent first draft without checking whether it meets the actual task criteria.
Do not use an unapproved AI service for confidential business data, credentials, private customer information or decisions that require qualified professional judgment.
The output meets explicit audience, format and quality criteria.
1 · Triage
Quick checks before changing anything
Use these checks to narrow the problem and avoid applying an unrelated fix.
Decide whether AI fits the task
AI is useful for drafting, transformation, classification and exploration; it is less reliable as the final authority for current, high-stakes or private decisions.
Define the outcome
State who the output is for, what action it should support and how you will judge whether it is good.
Gather safe context
Provide relevant notes, examples and constraints while removing confidential, personal or unnecessary information.
2 · Stabilize
Safe first actions
Mark actions as you complete them. Progress is stored only in this browser.
3 · Diagnose and resolve
Step-by-step route
Continue only when the result of the current step supports the next one.
- 1
Choose a bounded task
Break a large request into research, outline, draft and review stages.
Expected resultEach AI step has a clear input and output. - 2
Prompt with context and constraints
Provide the source material, audience, exclusions and format.
Expected resultThe first result is relevant enough to evaluate. - 3
Critique and iterate
Point to specific omissions or failures and ask for a targeted revision, not a complete random rewrite.
Expected resultQuality improves while useful parts are preserved. - 4
Verify and sign off
Check facts, sources, privacy, bias and final usability before publication or action.
Expected resultThe human owner remains accountable for the result.
Stop and escalate when
- Do not use an unapproved AI service for confidential business data, credentials, private customer information or decisions that require qualified professional judgment.
4 · Verify
How to know the problem is resolved
- The output meets explicit audience, format and quality criteria.
- Important claims and sources have been independently checked.
- A human owner has approved any consequential action.
5 · Build the skill
Recommended learning route
Open these lessons in order when you need more detail or want to prevent the issue from returning.
Choose suitable AI tasks
Use AI for drafting, organizing, transforming and exploring—not unquestioned final decisions.
AI & Productivity · AI FoundationsWrite context and a clear goal
Explain the situation, audience, source material and desired result.
AI & Productivity · Prompting for Useful ResultsAdd constraints and output format
Specify length, structure, exclusions and quality criteria.
AI & Productivity · Prompting for Useful ResultsImprove prompts through iteration
Review the first result, identify the gap and ask for a targeted revision.
AI & Productivity · Prompting for Useful ResultsKeep human sign-off
Make the responsible person review and own the final decision.
AI & Productivity · Privacy and Responsible Use6 · Practice and reference
TipsMake tools and guides
Related routes
Continue with a connected problem
Common questions
FAQ
Does making a prompt longer always improve the answer?
No. Relevant context and clear constraints help; unrelated detail can distract the model. Include information that changes the task, decision or evaluation.
Should I ask AI to “act as an expert”?
A role can set perspective, but it does not create expertise or guarantee accuracy. Concrete context, source material, constraints and verification are more important.