Notes from building ScopeBlind.
Written by Tom Farley. The posts written for ScopeBlind’s earlier products are kept below for the record.
Earlier posts
Five posts written between March and May 2026 for ScopeBlind’s earlier products, kept for the record, each with a note saying what was corrected or removed since.
- Open source shows what could run. Receipts prove what did run.
A worked example of verifiable recommender-system execution, built around xAI's published For You algorithm.
- Where Decision Receipts Fit in the Agent Supply Chain
What they prove, what they don't, and how they compose with the adjacent evidence layers being built right now.
- When AI Finds Zero-Days, Who Proves What Happened?
Receipt-signed vulnerability disclosure: cryptographic evidence for every step of the remediation lifecycle.
- Your AI Agent Needs a Track Record, Not a Trust Badge
Static scanning can't secure non-deterministic AI agents. Your agent needs runtime evidence: signed receipts that anyone can verify without trusting the vendor.
- How BlindLLM Uses Signed Receipts to Prove AI Comparison Results
Every AI benchmark today operates on trust. We built the infrastructure to replace trust with cryptographic proof.