I Ran a Coding Agent on Two Consumer GPUs. Speed Wasn't the Bottleneck.
North Mini Code crossed the usefulness threshold under supervision, but not the trust threshold, on a local two-GPU coding-agent setup.
Blog
LLM Context Is Not a Database
Large tool outputs turn the prompt into an accidental data store; reliable agents keep growing data external, bounded, and retrievable.
The Canon Has Not Yet Formed
The engineering discipline of agentic systems is taking shape before our eyes: core computer science principles remain, but the canon of agent engineering does not yet exist.