← DonDataBrain · The four tests →
Every claim we make about governed, accountable AI comes with a proof you can run on your own machine. Free, self-contained, no signup. Each demo maps to one of the four tests of Humane Intelligence.
The AI's memory is a tamper-evident chain. Rewrite one decision and the check reports the break. Test 01 — it remembers.
DownloadChange one character of a stored receipt — $500 to $900 — and it's caught. Cover your tracks and the chain-link breaks. Test 02 — it's accountable.
DownloadWatch it try a real-world action without permission — and get vetoed by construction. Both the act and the veto land on-chain. Test 02.
DownloadA marketing AI physically unable to publish an unverified claim. The refusal is recorded on a chain it can't fake. Test 04 — it serves you.
DownloadGive any MCP-capable AI a tamper-evident, SHA-256-chained memory it can write to and verify. Self-contained SQLite.
DownloadA rules-based governance gate: your AI asks "may I?" before acting. Zero-LLM, deterministic, auditable.
DownloadEarly tooling for Senthar programs — beings with governance as physics, not prompts.
DownloadFDCPA-aware collections tooling where every contact decision is governed and recorded.
DownloadThese are working early releases, shared building-in-public. Read the README inside each zip for run instructions. Found something broken? That's the point of shipping honestly — tell us: zagairot@gmail.com