Where humans and AI actually create together.
Chat, meet, write, analyze, and code — with Fab, the AI teammate who’s already read everything.
Product Launch
5Wait — what did we land on for the database choice?
@Fab can you drop a summary into the PRD?
Sure! You all decided on Postgres back on April 4 — schema flexibility and Convex compatibility were the main reasons over DynamoDB. I just wrote up the rationale in Q2 PRD §3.1.
Q2 Product PRD
Updated 1m ago · 4 contributors
3.1 Database Choice
The team chose Postgres on April 4, citing schema flexibility and Convex compatibility. The 47-message thread in Product Launch documented the trade-offs vs. DynamoDB.Fab
3.2 Schema Migration Strategy
Existing collections will migrate using the widen-migrate-narrow pattern — ship the new column nullable, backfill, then enforce…
A teammate, not a tab
Humans and AI, in the same room.
ChatGPT, Claude, Copilot, Gemini — all one human, one AI. With Cofab, your team, and Fab are in the same thread, the same doc, the same call.
Product Launch
5I think we should go with Postgres for the v2 schema
agreed — does Convex support it natively?
Yes — Convex has had Postgres adapters since v1.34. I dropped the docs in Engineering.
oh perfect
ship it
Write once, update everywhere
Stop switching tabs.
Chat, meetings, docs, spreadsheets, slides, code — all in one workspace, so you never lose context. Edit a spec doc, and the corresponding task, kickoff post, and meeting agenda all update instantly.
Q2 Product PRD
Updated just now · 4 contributors
Goals
Ship the v2 ingestion pipeline by EOM. Reduce p95 latency by 40%.Fab
Scope
v2 covers the high-volume tenant ingestion path, schema migration tooling, and observability hooks.
Non-goals
Multi-region failover and historical backfill — both deferred to Q3 unless a customer commitment forces the timeline.
for the kickoff —
Goals
Ship the v2 ingestion pipeline by EOM. Reduce p95 latency by 40%.
v2 launch
Goals
Ship the v2 ingestion pipeline by EOM. Reduce p95 latency by 40%.
Even code
Code lives where the work does.
Code is a container, like docs and slides — Reviews span code, schemas, and specs together. Fab pairs with your team right in the file diffs of every pull request.
// Fetch a userexport function getUser(id) { return db.users.find(id)} // TODO: add cachingPricing
Try it free. Then pay for what you use.
Starter
For individuals and small teams.
- All surfaces — chat, docs, databases, code, meetings
- 100 Fab requests/user/month, pooled
- Powered by Claude Opus
Pro
PopularFor teams getting serious about AI in their workflow.
- Everything in Starter
- 500 Fab requests/user/month, pooled
- Permissions, audit log, version history
Max
For engineering teams and AI power users.
- Everything in Pro
- 10,000 Fab requests/user/month, pooled
- Priority routing + SSO + SCIM via WorkOS