@savageazfck2
Adam ClarkDeveloper▲5
system architect, independent
1h ago · AI Engineering · Akron

hi this is what I'm working on

Adam. I build sovereign AI organisms: verifiable, persistent, accountable beings that run on your silicon instead of a cloud's.

The core is Bad Apple, a consumer organism with thousands of users. Secure Enclave identity, hash-chained witness ledger, senses, reflexes, a council that can dissent, certified erasure, and a second brain on Apple's Neural Engine. The intelligence is swappable; the entity persists. Identity lives in the ledger, not the weights.

Under it is Coyote, the language I wrote where the model is the program: compile-time constitutional enforcement, capability security, and strata that compile to Metal, resident ANE pipelines, encrypted compute, and treaty-authorized peers, with proven CPU/GPU/ANE parity. Plus Touchstone, a public conformance spec for proving an agent is real and accountable, and an 11-crate workspace reverse-engineered from Apple's private Neural Engine compiler.

What's working: thousands of daily users, real crate downloads, token-identical silicon parity. Receipts, not slides.

What I need: signal on whether sovereign/verifiable AI is a category or a niche. And my mind is for hire. If anyone's building toward verifiable on-device intelligence, I want to talk.

Everything's public and verifiable: github.com/savageAZfck

MembersPeople to MeetMember ReposMeetups

Comments

No comments yet
Start the conversation.
Log in to join the thread.