Prompt Lab

The Prompt Lab gives you a starting point for learning, everyday work, choosing tools, and planning responsible AI use. Pick the task closest to yours, fill in the details, and make the boundaries clear. Each page includes a complete template, a worked example, and specific checks for the result. Use the builder to shape your own brief or adapt a template. Before you share the output, compare it with your source material and the job you actually asked the assistant to do.

Copilot field guide adds 72 templates for presentations, documents and formatting, and work you’re getting ready to share, with a longer guide to shaping and reviewing a draft.

Your brief

Build a prompt around your actual task. Add context, boundaries, and the format you want.

YOUR PROMPT
Take it to your chosen AI tool.

Built here, run elsewhere. This builder creates text in your browser; it does not submit your brief to an AI model or execute code.

Adapt a template

Choose a starting point, then edit it to fit your task.

Fill in the blanks with the form, or edit the text below.

Read this template’s guide
View the original template

Fill in your template

Add your details below. When you’re ready, copy the prompt or add it to the editor.

Your prompt

Learn and understand

Explain it three ways

Get three explanations of one concept and see where each simplification stops helping.

Open the prompt

Build a learning path

Turn a skill goal and your available time into a practical sequence of learning sessions.

Open the prompt

Research with receipts

Build a source-backed answer with a clear distinction between facts, inferences, and open questions.

Open the prompt

Challenge my thinking

Expose the assumptions behind a claim and identify evidence that could change the conclusion.

Open the prompt

Everyday work

Make the email clearer

Clarify one email while preserving its facts, commitments, and intended meaning.

Open the prompt

Turn notes into next steps

Separate agreed decisions and actions from suggestions, missing owners, and unresolved dates.

Open the prompt

Compare real options

Compare supplied options using the same criteria and make missing decision evidence visible.

Open the prompt

Choose and evaluate tools

Compare tools against a real task

Compare candidate tools against one task, with evidence for capabilities and account requirements.

Open the prompt

Review a plugin or connector

Review an extension's capabilities, permissions, data destinations, and revocation before connecting it.

Open the prompt

Build a task-specific evaluation plan

Define representative test cases and evidence for one assistant, workflow, or agent.

Open the prompt

Build workflows and agents

Map a workflow before connecting apps

Map a process into steps, human checkpoints, and failure paths before connecting apps.

Open the prompt

Write a bounded agent brief

Specify an agent's job, tools, access boundaries, stopping rules, and human checkpoint.

Open the prompt

Review a code change

Review a supplied code change for located correctness, security, and failure-handling concerns.

Open the prompt

Organize a security investigation

Turn sanitized security observations into reversible checks and a clear escalation path.

Open the prompt

Decide responsibly and look ahead

Plan an accountable AI pilot

Plan an AI pilot with named responsibilities, evidence, controls, and conditions for stopping.

Open the prompt

Trace the benefits and harms

Trace who benefits from an AI proposal, who bears its risks, and which safeguards matter.

Open the prompt

Find a useful AI starting point

Find a bounded AI starting point by mapping process pain points to reviewable proposals.

Open the prompt

Explore futures without predicting them

Explore three horizons for a topic while keeping possibilities separate from forecasts.

Open the prompt

Sources

Share this page