Unit 4 ·

Summaries and drafts

Read this first. Every question after it comes from these cards.

Learn · 1/3

Say who it's for

The same document needs a different summary for every reader.

A summary is most useful when you say who will read it and what they need to decide.

For long documents, ask for key points with the section each comes from, so checking is easy.

The whole lesson as text

Cards and key takeaways

Emails and documents that need little fixing.

Key takeaways

  • Name the reader and their decision; ask for section references.
  • You decide the message; AI drafts; you edit.
  • Audience, purpose, length, tone.

Say who it's for

The same document needs a different summary for every reader.

A summary is most useful when you say who will read it and what they need to decide.

For long documents, ask for key points with the section each comes from, so checking is easy.

Remember: Name the reader and their decision; ask for section references.

Decide the message first

Let AI write the email, but never let it decide what you're saying.

Before AI drafts an email, decide the main message yourself.

Then give context and key points, ask for the right length and tone, and edit the draft into your own voice.

Good request

Reply to this email. Accept the meeting, propose Tuesday or Wednesday afternoon, under 80 words, match the sender's tone.

Remember: You decide the message; AI drafts; you edit.

Four things to set

Four settings turn a generic draft into the one you need.

Audience: who reads it. Purpose: what they should know or do.

Length: how much space it gets. Tone: how it should sound.

Set all four and the draft fits its reader.

Remember: Audience, purpose, length, tone.

Try it now

Paste this into any AI assistant (ChatGPT, Claude, Gemini, Copilot). Fill in the parts in square brackets.

Summarise this text for a busy manager: 3 bullets, then one line starting "Decision needed:". Text: [paste]

What to check in the answer: Every bullet must trace back to the original. Delete anything you cannot find in the source.

Want to go deeper?

This unit continues in a full course: AI Operating Systems & Work Environments (OpenClaw, Claude Cowork). One of its lessons is free to try: “Why single-tool setups fail”.