shteg.ai documents its build in public because the doctrine demands it. De-theatricalization and honest status registers only mean something if the work behind them is legible — so we write up the real commits, the real dated milestones, and the bugs the discipline caught. Every post below is grounded in git history and specs at HEAD. No invented releases, no fabricated volume, no live-money claims we have not earned.
RSS feedNewest first. Each write-up is anchored to git history and specs at HEAD — the discipline made public.
We migrated the claim, remittance, and wallet spine off floating-point dollars and onto integer cents — and the compiler-invisible unit change surfaced three latent 100x bugs that were already in the tree.
Read the postOur EMR sync had a callback that could exchange a code for tokens — but nothing that ever sent the clinician to the vendor's consent screen. We built the authorize leg, and made it refuse rather than fake a connection.
Read the postAn OIG-excluded or unlicensed provider's claim should never originate money. So provider legitimacy isn't a report you run alongside billing — it's a fail-closed condition inside the one gate that authorizes every advance.
Read the postA solo operator running a fleet of AI agents rebuilt this system in a six-day, 310-commit sprint. The reason the output didn't collapse into fiction is the doctrine the codebase enforces on itself.
Read the post'The code exists and is tested' is a different claim than 'it runs in production moving real money.' We keep those two claims separate on purpose — and we publish which is which.
Read the postNo real dollar has moved. No real PHI has flowed.
What is real: the architecture — hash-chained ledger, settlement gate, and reconciliation waterfall, tested in code at HEAD — the Type-2 organizational NPI and registered HIPAA clearinghouse identity, and the doctrine. We register capability as capability, never as traction.