watch it happen
◐proven end-to-end on stage 2026-06-12You make exactly two judgment calls — sign in and point at a repo. Every click below is a consent screen, counted honestly. Everything else is the app's job: the workspace, the import, the linking, the first answer.
One button. Your org is provisioned the instant you sign in — no setup form, no “create a project” wizard, nothing to name.
↑ real stage capture (2026-06-12 walk) replaces this frame in production
GitHub's own consent screens — named as such, because that's what they are. You pick the repo; GitHub asks permission; that's the whole job.
↑ real stage capture (2026-06-12 walk) replaces this frame in production
Decisions and claims tick up per source, live. If we truncate, we say so on the page — and email you when first wisdom is ready, so nobody babysits a progress bar.
↑ real stage capture (2026-06-12 walk) replaces this frame in production
The workspace opens on a synthesis of what your repo argued about, every sentence cited. On a thin repo it shows the gap ledger instead — what you decided in practice but never wrote down — because a gap is honest wisdom and a restated index is not.
Split in 2024 after the double-write incident: audit writes require serializable transactions the shared pool couldn't guarantee under load.1 Re-merging was argued twice and ruled out both times — the compliance boundary is load-bearing.2
↑ real stage capture (2026-06-12 walk) replaces this frame in production
One real walk, receipts attached — not an average. A fresh identity did exactly this on 2026-06-12: connected a fork of mozilla/fxa, watched a ~50-minute import, and landed on 49 decisions · 974 claims · 11 auto-links — then asked why, and got citations. The frames above are placeholders for that walk's real captures.
two doors, one graph
Everyone else hands your AI a filing cabinet; lema hands it the verdicts — and the map to the cabinet, opened only on demand. The two doors share one graph and one rule of identity:
sign-in is for humans · lema_live_ keys are for agentsnothing local is ever crippled.
don't lose the why
Run a real “why” query on React, Kubernetes, Rust, Vue, or Go right now — no account, no import. Or run it local in ten minutes. Both doors lead to the same graph.