Module 4: Meetings, Notes, and Synthesis

Why this matters

Meetings produce two kinds of work, preparation before, and follow-up after. Both are text-heavy, both are repetitive in shape, and both reward structure. AI turns meeting prep from “I’ll wing it” into “I walked in ready” without adding hours to your week. Synthesis, pulling key points out of long material, is the same skill, applied to reports, articles, and threads instead of meetings.

This is also where the “second perspective, not oracle” frame from Module 1 starts paying off. Meetings and decisions are full of viewpoints; AI’s job is to surface the ones you might miss, not to hand you a verdict.

Pre-meeting prep

A solid prep brief takes 5 minutes if you have AI; 30 if you don’t.

You're helping me prep for a meeting. Here's the context:
- Meeting: [title]
- Attendees: [roles]
- My role in the meeting: [presenter / participant / decision-maker]
- Goal: [what I want to walk out with]
- Background: [paste agenda, prior notes, or one paragraph of context]

Give me:
1. A clear agenda I could share (or refine if one exists)
2. The 3 most likely questions I'll get asked
3. The 1 thing I should make sure I say
4. Anything I should bring or send ahead

The “1 thing I should make sure I say” prompt is the unlock. It forces you (and AI) to decide what matters before the meeting starts.

Notes → action items

After the meeting, dump your raw notes into AI and let it pull the structure out.

Below are my raw meeting notes. Extract:
1. Decisions made
2. Action items — owner and deadline if mentioned
3. Open questions / unresolved items
4. Anything that should be communicated to people who weren't in the room

[paste raw notes]

This is the difference between “I took notes” and “I left the meeting with a plan.” It also makes you the person who reliably drives follow-up, a potentially career-defining habit.

Multi-source research synthesis

When you need to make a decision from a stack of inputs, a few articles, a long report, a Slack thread, a transcript, AI is dramatically faster than reading everything yourself, if you ask the right questions.

I'm trying to make a decision about [decision]. Below are [N] sources.
For each source, give me:
1. The 3 most relevant points to my decision
2. Any claim that contradicts the others
3. Any obvious bias or limit in the source

Then give me an overall summary: where the sources agree, where they disagree,
and what I should still verify before deciding.

[paste sources, separated by --- or labeled clearly]

Critical: the verification step. AI will sometimes synthesize confidently across sources that disagree, smoothing over real conflict. The “where they disagree” prompt forces it to surface that, that’s the second-perspective frame in action.

The “explain like I’m reviewing this” prompt

Use this when you’re walking into a meeting in 10 minutes about a doc you got 5 minutes ago, or any time you need a fast, defensible read on long material:

Read the material below. Then answer:
1. What is this actually about? (one sentence)
2. What's the main claim or recommendation?
3. What's the strongest argument for it?
4. What's the strongest argument against?
5. What should I follow up on?

[paste material]

This is faster than reading and gives you a defensible position to walk into a meeting with.

Try It At Work: Prep Your Next Real Meeting

Time: 10 min

You’ll need: A real meeting on your calendar in the next 48 hours.

Do this:

1. Open the meeting invite. Copy the agenda (or write a sentence about what it’s about).
2. Run the prep prompt above.
3. Identify the “1 thing I should make sure I say.”
4. Save the brief somewhere you’ll see before the meeting.

Done when: You have a 1-page brief and you know your one key message before walking in.

Try It At Work: Extract Action Items from Real Notes

Time: 10 min

You’ll need: Notes from a meeting you had this week (any format, bullets, scribbles, transcript, screenshots).

Do this:

1. Paste the notes (or screenshots of them) into your model.
2. Run the action-items extraction prompt.
3. Review the output. Did it catch everything? Did it invent anything that wasn’t there?
4. Send the cleaned action items to whoever needs them.

Done when: You’ve sent (or saved) a clean action-item list and you noticed at least one thing AI got wrong or missed. That noticing is the skill.

Try It At Work: Synthesize a Real Decision

Time: 15 min

You’ll need: A real decision you’re working through, plus 2–4 inputs (articles, docs, a thread, internal notes, whatever you’d actually consult). Confirm safe to paste, or redact first.

Do this:

1. Paste each source labeled clearly.
2. Run the multi-source synthesis prompt.
3. Note where AI says the sources disagree. Verify at least one of those claims by checking the source yourself.

Done when: You’ve used the synthesis to advance a real decision and you’ve verified at least one disagreement claim AI surfaced.

Key takeaways

• Pre-meeting briefs take 5 minutes with AI. Always include “the 1 thing I should make sure I say.”
• Notes → action items is the highest-leverage post-meeting move. It makes you the follow-up driver.
• Multi-source synthesis is fast but risky, always force the model to surface disagreement and verify.
• The “second perspective, not oracle” frame is what makes synthesis safe, you’re using AI to reveal disagreement, not to declare a winner.

Quick Check