Everyone remembers the last meeting slightly differently, so the first ten minutes negotiate a shared version of the past. Here is the page that removes that, assembled without anyone writing it.
Most recurring meetings begin with ten minutes of reconstruction. Where did we leave this, what was agreed, who was going to do the thing. Everyone remembers slightly differently and the meeting negotiates a shared version of the past before it can discuss the future.
What a preparation pack is, and is not
It is one page, assembled automatically, delivered the evening before: who is in the room, what was agreed last time, what is still open and who owes it, and the decision this meeting needs to reach.
It is not a briefing document, a deck, or a research summary. The moment it becomes something somebody has to write, it stops being produced during busy weeks, which are the weeks it matters most.
If producing the pack requires effort, it will not exist when you need it. Everything in it must come from somewhere that already holds the data.
How to build it
1. Pick the recurring meetings only
Weekly leadership, client check-ins, one-to-ones. Do not attempt this for ad hoc meetings; the value is in continuity across a series, and one-off meetings have no last time to refer to.
2. Fix where decisions live first
This is the prerequisite everything else depends on. If decisions are not recorded anywhere consistent, there is nothing to assemble. A decision log with what was decided, who owns it and by when is the source this automation reads.
3. Assemble from systems, never from prose
Attendees from the calendar. Open items from the decision log or tracker. Last meeting's agreed points from the previous pack. If a field cannot be pulled automatically, leave it out rather than asking someone to fill it.
4. Deliver the evening before, not the morning of
People read the evening before. On the morning of a meeting they are in the previous meeting. This single timing choice affects whether it gets read more than the content does.
5. Put the decision needed at the top
Not the history. The one line stating what this meeting must resolve. Everything else is context for that line, and putting it last means it gets read last.
6. Add a light AI layer only if you keep the gate
Summarising the previous discussion is a genuine use for a model. It must not decide what is still open, because a transcript contains commitments, suggestions and thinking aloud that sound alike. Pull open items from the log, not from the conversation.
Tools and what they cost
| Option | What it costs | Honest trade-off |
|---|---|---|
| Apps Script pulling calendar, docs and tracker | Free with Google Workspace. | Reads all three sources natively and writes a formatted document. Best fit if you are already on Workspace. |
| Meeting assistants with recaps (Fathom, Granola, Fireflies) | Free tiers exist; paid typically $10 to $30 per user monthly. | Handles the previous-meeting half automatically. Open action items from your tracker usually still need joining. |
| Native tracker reporting plus a scheduled email | Included in most paid tiers. | Covers open items well. Attendee context and last-meeting recall usually missing. |
| Connector platform assembling a doc | Per task; low volume here, so cost is modest. | Reasonable fit given a handful of recurring meetings. Document formatting is usually the weak point. |
What it is actually worth
The arithmetic is direct. Count your recurring meetings per week, times attendees, times the minutes currently spent reconstructing at the start. A weekly leadership meeting of six people losing ten minutes is an hour of collective time every week.
The larger effect is on follow-through. When last meeting's open items appear in front of everyone at the start of the next one, the discomfort of reporting the same item unfinished three weeks running does more for completion than any reminder system. That is the same mechanism the decision log relies on, and this automation is what puts it in front of people.
For context on the scale of meeting time at senior levels: Porter and Nohria's Harvard Business Review study tracked 27 chief executives across roughly 60,000 hours over thirteen weeks and found they spent about 72 percent of work time in meetings, averaging 37 meetings a week. That describes chief executives specifically rather than everyone, and it is a small sample, which I state because it is usually quoted without either caveat.
How it breaks
The decision log is not maintained, so the pack is empty. The pack is downstream of the log. Fix the log first or this automation will publish blank sections and quietly discredit itself.
It becomes a document somebody writes. The moment a human has to compose part of it, it stops appearing in busy weeks. Keep everything machine-assembled.
It arrives too late to read. Morning-of delivery is the same as no delivery.
It grows into a briefing pack. One page. If it needs scrolling, it will be skimmed, and the decision line at the top is what gets skipped.
How to tell whether it worked
Minutes spent on recap at the start of recurring meetings, measured honestly for a fortnight before and after. Then the share of open action items closed by the following meeting, which is the follow-through effect and the more valuable one.