So far Scriven Labs helps one engineer at a time. The next phases connect those personal agents live, during real conversations and meetings, so the same grounded context reaches the whole room.

Pairing

A lightweight desktop app captures each person's own microphone, not the room's system audio. The apps pair into a shared session using device pairing tokens. Everyone keeps their own private memory and their own private nudges; the session is what lets those nudges become shared context when it helps.

The Coordinator

A Main Coordinator listens across all the paired apps and publishes grounded mediation cards: the options on the table, the evidence, and where every claim came from. No card publishes without a verifiable artifact citation. No citation, no publish.

Shared cards re-check access against the intersection of everyone who will see them, and performance-adjacent facts stay blame-neutral by design. The goal is to settle a stalled debate with a source, never to put someone on the spot. This is what we're building toward next.