wrhn

kennmangum

Joined 8 months ago

Karma
1
Submissions
7
Member since
Feb 27
Building SIX (https://six-sov.com) — cryptographic receipts for AI inference. Background: CFO/finance, now focused on verifiable AI infrastructure for regulated industries. [email protected]
View on Hacker News

Recent submissions

Comment · 7 months ago
We built a receipt system for AI inference — every response gets a SHA-256 hash, Schnorr signature, and optional XRPL anchor.

The problem: enterprise AI deployments need audit trails that can't be edited. Database logs don't cut it for regulated industries.

Try it: npx six-verifier verify <hash>

Looking for feedback from folks building compliance-sensitive AI applications.

Comment · 8 months ago
Agents get durable memory across sessions. Every write produces a cryptographic receipt. You can verify, export, or delete anytime. No vendor lock-in.

Happy to answer questions on architecture, crypto choices, or why "agentic memory" is a real category.

Comment · 8 months ago
Verify an inference in 30 seconds: npx six-verifier sample-receipt.json The receipt proves: who ran it, what model, when, and that the output wasn't tampered with. Not "trust our logs" — actual cryptographic proof. Signed receipts + Merkle attestations. More at: https://six-sov.com Happy to answer technical questions — threat model, crypto choices, performance tradeoffs.

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton