Share understanding as the team grows.
Start with a shared workspace and who-reviewed-what. GitHub-connected intelligence is next, and stays clearly labeled Pilot until it ships.
Share a workspace. See who reviewed what.
Founding Teams today is a shared workspace with invites and a who-reviewed-what activity feed. Full explanation text stays on each person’s device. GitHub-connected intelligence is Pilot — not shipped yet.
- Shared workspaceCreate a team, invite seats, and work in one place.
- Who reviewed whatSee who understood which file, concept, and project.
- GitHub intelligencePR walkthroughs and knowledge-risk dashboards stay Pilot.
understood auth middleware
reviewed billing plans.ts
needs review on sync.ts
Illustrative activity feed. PR dashboards and knowledge-risk scores are not current claims.
One useful layer at a time.
The personal product remains the developer wedge. Teams starts with shared activity, then adds GitHub intelligence only when it is real — never as a silent claim.
Share the platform
Create a team workspace, invite teammates, and see who reviewed or understood which file, concept, and project. Explanation text stays on each device.
- Team workspace and invites
- Who-reviewed-what activity feed
- Authorship on shared metadata
Connect GitHub next
Repository connection, PR walkthroughs, and transparent knowledge-risk signals are labeled Pilot. They are not included in the founding workspace today.
- Organization and repository connection
- PR understanding
- Coverage, freshness, and Understanding Gap
Act on the next gap
Project-native recommendations that point to a real PR, architecture path, concept, or repository walkthrough — after the GitHub layer ships.
- Engineering onboarding
- Weekly knowledge brief
- Ask Engineering and architecture context
Understanding coverage is not employee performance.
Commit activity can establish familiarity, but it never proves comprehension. When GitHub intelligence ships, scores will use explicit evidence, show why a number moved, and say “Not enough data” when the evidence is thin.
- Expose its contributing evidence and weights.
- Separate source evidence from AI inference.
- Avoid leaderboards and “good/bad engineer” labels.
- Handle loading, missing-data, error, and insufficient-evidence states.
Governance comes after the useful foundation.
Organization intelligence, policies, permissions, audit events, and retention controls are planned. SSO/SAML, SCIM, BYOK, VPC deployment, data residency, and SLA support are later-stage capabilities — not current claims.