Scriven Labs connects your team's work, decisions, and history and organizes it so context is never lost. That is the Context Layer, and it is the foundation everything else stands on. It shipped first, on purpose, because a meeting assistant is only as trustworthy as the memory underneath it.

What it does

It connects the tools where work actually happens: GitHub, Slack, Gmail, Docs. From each, it mines the work you own and drops the rest. Pull requests, issues, and decisions become a searchable memory you can question in plain language, and every answer comes back with its source attached.

The two guarantees we proved first

Before any of it touched a real team, two load-bearing guarantees were proven in CI against real Postgres. Tenant isolation at the database layer, so one team's context can never leak into another's. And provenance-gated retrieval, so every claim resolves to a real, access-checked artifact ID. If a claim can't be traced to a retrieved source, it is dropped, not hedged.

Both guarantees are enforced in code, not in a prompt. OAuth tokens live in a dedicated secrets manager, never the application database. Raw audio is never stored beyond the transcription stream. These aren't features you toggle; they're the floor.

With the Context Layer shipped and the connect, sync, retrieve loop proven end to end, the memory is real. Everything Scriven Labs does in a meeting draws from here.