TAG: LLM (3)
Scaffolding vs. Intelligence: Why AI Evaluation Harnesses Fake 70-Point Leaps
Why AI evaluation harness architectures create massive score spreads on benchmarks like ARC-AGI-3, and how to establish strict benchmark provenance.
The Phantom Endpoint: Why Your RAG Assistant Hallucinates APIs (And How to Fix It)
A deep dive into RAG hallucinations, same-domain brand substitution, why similarity thresholds fail, and how deterministic scope gating prevents fake API generations.
Durable Memory: Why Vector Databases Aren't Enough
Why durable memory is more than persistent storage, and why the real architectural problem begins at the write boundary.