Team memory, compiled daily
Stop losing decisions in Slack.
Compound turns email and Slack into a living wiki your team can actually use — updated every day, linked, and cited back to the source. Export anytime.
The problem
You already decided. Nobody can find it.
Your team lives in Slack and email. Chat tools answer once. Wikis go stale the week everyone gets busy.
Threads bury the call
The real decision was in #product at 6pm. Good luck next quarter.
Chat forgets by morning
AI answers once. The thread dies. Nothing compounds.
Wikis go stale
Asking humans to “update the KB” fails the week you get busy.
The model
A real wiki. Maintained for you.
Not a chatbot with amnesia — three layers that keep quality honest.
Kept raw
Email, Slack, uploads — hashed, provenance kept, never silently rewritten.
Linked pages
Decisions, people, products, customers — cross-linked, indexed, with a change log.
Your schema
What deserves a page, how tags work, when contradictions get flagged for a human.
How it works
One stream in. A clearer team out.
Connect
One email label or a few Slack channels — not the whole company dump.
Ingest
Daily pull, deduped and filtered so bots and noise stay out.
Compile
Pages created and updated with links back to the original source.
Use
Ask with citations — or open the morning digest of what changed.
Contrast
Chat finds. Compound keeps.
| Other tools | Compound |
|---|---|
| Search or chat over raw piles | Compiles durable wiki pages |
| Answer lives in one thread | Knowledge stays in the vault |
| Stale KB — or no KB | Daily maintain + change log |
| Hard to leave | Markdown export, Obsidian-friendly |
| Enterprise rollout or solo chat | Built for founder-led teams in Slack |
vs chat over notes
Notion AI and ChatGPT answer once. We pull your streams and keep the pages.
vs hand-written KBs
Guru and Confluence need writers. We compile from work that already happened.
vs DIY agents
Same compounding idea — hosted, isolated, connectors, daily cadence.
Trust
Private. Portable. Start small.
- Workspace isolation — your vault isn’t mixed with anyone else’s context.
- Encryption — TLS in transit; encryption at rest on stored content.
- No training on your data — your sources aren’t a model improvement feed.
- Full export & delete — markdown zip out; hard-delete when you say so.
- Start narrow — one low-risk label or channel before you widen scope.
- Human review hooks — contested claims surface instead of silently becoming “truth.”
Who it’s for
Built for the person who is the company wiki.
Primary: founders and PMs on small teams. Same product also fits solo research vaults and client-isolated agencies.
Founders & PMs · 5–30 people
Get out of the “let me find that thread” loop. Onboard against living pages — not Slack archaeology or oral history.
Investors & researchers
Thesis and deal vaults that update as sources arrive.
Agencies
Per-client isolation; export cleanly at offboarding.
Design partners — $99/mo
Limited seats. Real email or Slack, daily compile, founder access, price lock when we GA. Apply below as design partner.
Waitlist
Tell us where context dies.
Waitlist open. Design partners (paid) go first if you’ll connect a real source.
FAQ
Straight answers
Is this just another AI chat?
No. Chat answers once and the thread dies. Compound maintains wiki pages from what you allowlist, then answers from that layer — with citations back to source.
What do you read from Slack or email?
Only what you allowlist (label, address, or channels). Filters cut bots and noise so they don’t poison the vault.
What about Notion or Confluence?
Keep them. Compound is for decisions and context that never made it into the doc — especially Slack and email.
Can I export?
Yes. Markdown vault export is first-class — Obsidian-friendly on purpose.
Will you train on my data?
No. Customer content is not used to train models.
Who should be a design partner?
Early paid users ($99/mo) with real pain who’ll connect one source, give product feedback, and get founder access plus a price lock toward GA.