All work
Search across our own archive
A shared archive of notes, decisions and records serves as memory for people and AI agents, with full-text search and automatic sync.
The situation
Agents had to read whole folders at the start of every session. That costs time and compute and does not scale. The rule to document work was regularly forgotten.
The idea
There is exactly one search index that every machine queries. What used to be a rule is now automated, such as the journal entry after every working session.
How it is built
- Full-text search with filters for type, folder and date, with machine-readable output for agents.
- The index deliberately lives outside the archive, so no generated files end up in version control.
- The server saves every two minutes; every change is versioned.
- Overview pages renew themselves.
- Sessions with real work automatically produce a journal entry, plain conversations do not.
Documented figures
- 1,641notes in the search index (22 October 2026)
- 2 minsync interval
Status
Sync in daily use since April 2026, search since August 2026.
A similar problem?
Tell us in one sentence what it is about. We will tell you honestly whether we can help.
Write an email