Domain
Meetings
Each transcript compiled into one entry, including what I said I would do.
The most useful part of a meeting is often one sentence: something I agreed to do. It is also the part most easily lost. Meetings keeps the whole conversation and pulls out the commitments.
The transcript is the evidence; the entry is knowledge compiled from it, which a better model can compile again.
From meeting to entry
- 01
Recorded at home
An open-source recorder on my home server transcribes the meeting locally. Only the transcript is kept, never the audio.
- 02
Sealed as it arrives
Each transcript is sealed as the original the moment it lands, so nothing can change it later.
- 03
Compiled into one entry
A model turns each transcript into one entry: what was discussed, what was decided, and what I said I would do.
- 04
Actions I can tick
My own actions from a meeting appear in the console, and I tick them off as I do them.
The boundaries
- Speakers stay anonymous
- Other people in a meeting are labelled by number, not by name. Only my own words are attributed to me.
- Transcripts stay local
- Transcription happens on my home server, and the transcripts never leave it except through the domain's interface.
- Other people's actions stay as said
- I can mark my own commitments done. What other people said they would do is kept as said.
Meetings shows the same pattern as every domain: keep the source exactly as it was, let a model compile it, and keep the compiled part replaceable.