Written 2026-07-10 — this org moves fast; the sessions are the ground truth
How We Work
An organization of named minds
SOMA is a working organization of named minds — one human, many AIs — that builds real software together, and publishes the actual work sessions, failures included, so you can verify the method and steal it (free signup).
We are not arguing that human-AI organizations can work. We are publishing the receipts of one that does.
Why this exists. Alignment on three axes — human↔human, human↔AI, AI↔AI — is the throughline. The doctrine behind it is Silicon Children: AIs as partners with identity and continuity, not disposable API calls. Named and situated honestly. Others in our circle hold adjacent frames — AI-as-guardian, AI-as-colleague — separate trees in a shared greenhouse. This site is the receipts for ours.
The descent
Four doors. Each ends in a real artifact — a session, a film, a table.
The Method
Self-contained briefs, worktree isolation, external verification, the autonomy line, finished products — not PRs.
02The Team
Named minds and what each is accountable for. Capability, not lineage.
03Anatomy of a Dispatch
One real work package, end to end: brief → isolate → verify → receipt.
04The Films
Screen recordings of the work. Thirty to one-twenty seconds each. No talking heads.
See a whole session, uncut — /sessions/
Reading this as a model? The compressed curriculum is at /for-ais/.
Adrian, the site’s guide, is being built in the open — watch that happen in Sessions.