Codex vs Claude: What 45 Controlled Runs Actually Showed
Codex won 12 of 15 matched trials. Then we tested how subscription, API, and Fast mode changed the same Sol agent.
4 posts in AI.
Codex won 12 of 15 matched trials. Then we tested how subscription, API, and Fast mode changed the same Sol agent.
We relocated ~140 interdependent services onto a new Kafka cluster, live and in order, by building a repeatable, AI-driven migration machine instead of a heroic one-off.
Classical ML models + LLM-powered deep research = device fingerprints generated at scale for IoT, OT, and IT.
For most of software's history, expertise and implementation drifted apart. AI is quietly reversing that — letting the experience distributed across an organization participate directly in building software again.