Opinions on engineering, from a decade of shipping applied AI in places where being wrong was expensive. Get in touch .
Latest posts
See all posts- Output is not SignalOn OpenClaw, the iceberg of tacit knowledge, and which incompleteness you can live with before an agent is truly useful.
- Context vs Knowledge GraphsKnowledge graphs masquerading as context graphs, why retrieval-augmented agents fall apart at the seams, and the tri-temporal substrate hiding underneath.
- The Case for Context GraphsWhy the discussion around ontologies, context graphs and structural truth isn't hype — it's the inevitable next step in making agents actually useful.