Guide
Anatomy of a perfect meeting write-up (and how to automate it)
Decisions, actions, owners, deadlines: the ingredients of minutes people actually use. And why AI does it better than taking notes by hand.

Everyone knows what meeting minutes are. Few people write good ones. Fewer still write them consistently. Here is why, and how to automate the production of minutes people genuinely use.
The five components of a perfect write-up
Usable minutes are not a summary of what was said. They are an action document. They contain:
1. The context
Date, participants, agenda. It sounds basic, yet 40% of handwritten minutes leave out at least one participant. Gilbert identifies every speaker automatically, by voice.
2. The decisions taken
The heart of the document. Not "we discussed the budget" but "Decision: the Q2 marketing budget is approved at €180k, owned by the CFO". A decision means an action verb, an owner, and an amount or deadline where applicable.
3. The actions to take
Every action needs an owner, a deadline and an expected output. "Peter should send the report" is not an action. "Peter sends the audit report to the CFO before Friday 14 February, in PDF" is one.
4. The open points
Unresolved questions, postponed topics. This is the team's institutional memory. Without it, the same discussions come round again and again.
5. The executive summary
Three to five lines at the top for decision-makers who do not have time to read everything. What someone who was not there absolutely needs to know.
Why taking notes by hand fails
The problem is not the note-taker's ability. It is structural:
- Selection bias: you write down what you think matters, not what actually does
- Loss of context: you cannot listen actively and write at the same time
- Delay: the minutes arrive 24 to 48 hours later, once memory has already reshaped the facts
- Inconsistency: the format changes with whoever is writing
How Gilbert automates the perfect write-up
Gilbert produces structured minutes two minutes after the conversation ends. The process:
- Transcription and diarisation: who said what, at 98.2% accuracy
- Semantic extraction: decisions, actions, questions and key figures identified automatically
- Structuring to your template: you define the format, Gilbert follows it
- Automatic export to Slack, Notion, email, your CRM or any other tool you use
In closing
Good minutes are not a luxury, they are the basis of team productivity. But writing them by hand is a waste of time nobody should still accept. Let the AI do the work. Concentrate on the conversation.




