← back to journal
Published 15 May 2026 · by Thomas Kolmans

What tacitylab is, and what it isn't

A short note on why we're inverting the usual "AI-for-work" stack: capture the undocumented first, layer the documents on top.

The premise

Every "AI for work" tool we've seen starts the same way: point it at your wiki, your docs, your Slack history. Then it answers questions about what's already written down.

That misses the part that matters. The most valuable thing your team knows isn't in the wiki. It's in the head of the person who knows why the wiki says what it says.

What tacitylab does differently

Three moves, in order:

  1. Capture the undocumented. Structured interviews. Short video prompts. A questionnaire engine tuned to draw out the why, not just the what.
  2. Ingest the documented. Read-only connectors to the tools you already use. Indexed inside your tenant. Embeddings only.
  3. Surface at the moment of work. One sentence. One source. Inside the inbox, the IDE, the support desk, the editor — wherever the question is actually being asked.

What tacitylab is not

  • It isn't a chatbot. It will refuse to answer when the sources are weak.
  • It isn't a wiki replacement. Your wiki is fine. It's just incomplete.
  • It isn't a meeting recorder. Recording the meeting doesn't capture the reason the meeting reached the conclusion it did.

Where we are

Private beta opens Q3 2026. We onboard one team a week, founders involved. The waitlist is open.

If you want to be early — and you want the kind of attention that only a small team can give — join the list.
Also inNederlands
MS-∞ · waitlist

Make what your team
knows, findable.

Private beta opens Q3 2026. Founders involved in every onboarding.