The Recap Workflow.

The most universally useful AI workflow for senior professionals is also the most boring one. It is the recap.

Take scattered notes from a call, a meeting, or a long-running matter. Produce a clean, client-ready memo. Twelve to fifteen minutes, including the edit.

Is Claude a good AI note taker for meetings?

People search for an "AI note taker" expecting a bot that joins the call and hands back a transcript. Those tools exist, and they are fine at capturing words. The problem is what you get back: a wall of text that still needs a human to decide what actually mattered.

That second step is where Claude earns its place. Whether your notes came from a transcription tool, a colleague, or three lines you scribbled yourself, Claude turns raw meeting notes into the thing you actually need: the decisions made, who owns what, and a clean recap you can send. You are not handing over your judgment about what matters. You are skipping the slow part of writing it up.

So the honest answer: for capturing audio, use a transcription tool if you like one. For turning any set of meeting notes into a decision log, action items, and a follow-up email, the AI note taker worth having is the general assistant you can direct, not a single-purpose app. The workflow below works with notes from any source, and the Claude email workflow handles the follow-ups that come out of the meeting.

The structure of a useful recap

Every good recap has the same four parts, in this order:

• What we covered. Two to four sentences. Neutral tone. No editorialising.
• What was decided. A short list. Each decision, one line.
• What is next. A short list. Each action, with an owner and a date.
• Open questions. A short list. Anything not resolved.

That is the whole shape. You can deliver it as an email or a one-page memo. The structure does not change.

The prompt

Below are my raw notes from [type of meeting] with [Client A / Partner B / Counterparty C]. Produce a recap memo with these four sections, in this exact order:
1. What we covered (2–4 sentences, neutral)
2. What was decided (bullet list)
3. What is next (bullet list with owner + date)
4. Open questions (bullet list)
Tone: calm, senior, professional. Length: about 250 words. Use placeholders where I have used them. Do not invent details.

Notes:
[paste]

Sanitisation, first

You sanitise the notes before you paste them. Client name becomes [Client A]. Real dollar figures become bands: [mid-seven-figure]. Counterparty names become [Counterparty C]. The model produces the memo with the placeholders intact. You substitute the real names back in during your edit, in your own document, locally.

This is the same discipline as the Never Upload List. Same reason.

The edit, not the draft

The output you get back will be 70–85 percent of the memo you would have written yourself. Your job is the remaining 15–30 percent:

• The two sentences about tone, posture, or politics that the model could not have known
• The one decision that is more important than the others, which the model treated as one bullet among many
• The open question that you know from context is actually the central question
• The name corrections, the date corrections, the figure corrections

If your edit is bigger than that, the input notes were not detailed enough. The fix is on the input side.

What this buys you

About thirty minutes back, per recap. Probably four or five recaps a week, for a senior professional with a normal book. That is two hours. Compounded over a year, that is a working month.

The clients also notice. Recaps that arrive same-day, in clean shape, with clear next steps, change the way the client experiences the relationship. The system is small. The compounding is not.

Filed under Workflow · The Leveraged Years · Vol. I, Issue 13.