Anki is a governed AI workforce — 45 engines working under one owner's authority to scout, draft, verify, and prepare, built around a simple rule: nothing moves forward without proof. Built in public by a non-coder, Anki is proof that AI does not need to be trusted blindly. It can be governed. Every claim on this page is real.
Engines hunt opportunities, retrieve evidence, and prepare work. Nothing sends itself. Nothing spends itself. The owner presses every trigger that matters.
Engines locate real prospects, bank the evidence behind every fact, and compose outreach from verified data only. A contact route that isn't evidence-backed cannot ship — the validator redacts it at the gate.
This is the machine behind the pilot — not a separate product. It's how the work gets done, honestly.
Every build, deploy, correction, and ruling lands as a numbered ledger entry with its proof attached. The tablet tradition, continued: if it isn't receipted, it didn't happen.
Secure checkout through Stripe. Every pilot is handled by me, personally — start to finish.
The workforce runs under written constitutional law. Three of its load-bearing rules:
Models locate and assess; only retrieval creates facts; only deterministic validation admits them. A minted email or invented claim is redacted at the gate — structurally, not by promise.
The owner's direct action is its own approval — the system may never wall its own human. Judgment stations, yes. Internal permission theater, never.
Every send, publish, and payment is a human press. The workforce drafts and queues; one person decides what enters the world.
The root of the record — an honest build log of governing an AI workforce as a non-coder. Wins, failures, and the receipts for both.