Asked whether the whole personal-brain architecture is just RAG, Tan concedes the framing and dismisses its force in one line: “Sure, and Postgres is just B-trees.” Retrieval is the primitive, not the product. The hard part is everything around it — what gets written down in the first place, how it gets enriched and linked, what gets promoted to hot memory versus filed as cold reference, and who arbitrates when two facts disagree. His summary is the useful inversion: “Retrieval is easy. Being worth retrieving from is the product.”
He pairs it with the caveat that makes the claim falsifiable rather than triumphant: “a brain nobody curates is a garbage dump with great search.” Retrieval will surface a stale fact with total confidence, and a bad skill file encodes a bad process forever, so the primitive he actually names is memory plus hygiene — provenance on every fact, contradiction checks when new information collides with old, and a librarian whose real job is pruning. Treat the brain like production infrastructure and it compounds; treat it as a dumping ground and you get a confident agent wrong in ways nobody can trace. This is the same failure surface as Once skill files accumulate, provenance becomes the binding constraint and it sharpens Don't mistake document ingestion for memory — learning from interactions is a different and harder system by naming curation, not ingestion, as the scarce work. It also explains why A better model makes your library worth more, not obsolete — a smarter reader extracts more from the same books is conditional — a smarter reader extracts more from good books and more confident nonsense from bad ones.