11 posts tagged with "Building"
A frontier model went from running my whole agent fleet to reviewing it. The cost cliff that forced the change and the early benchmark data point the same direction, but only one of those two is proven.
Getting an AI agent to act is the easy half. Leaving one running for months, unattended and accountable, is the hard half. Here is the discipline, the mechanism it turns on, and the system I run it on.
A viral paper says transformers fail a 90-year-old attention test as the list gets longer. I reproduced it on the 2026 models. The deficit is real. The cliff just moved out.
A community leaderboard ranks me among the heaviest Claude Code users it has seen. The rank is the least interesting part.
Every AI session starts from scratch. I built a system that fixes that: a vector database, multi-machine sync, and a core memory injected before the session even starts.
How I used Claude Code to turn a dreaded weekend of Japanese tax filing into a 3-hour Friday evening
I structured my AI assistant like an organization: a chief of staff, a council of advisors, and domain experts. It sounds absurd. It works.
Most projects get harder over time. Compound engineering inverts this by encoding every lesson back into the system. The same discipline works for code, AI agents, and leadership.
How to add AI features to applications where the data can't leave your machine but the LLM lives in the cloud
Most business leaders my age read about AI. I decided to build.
In enterprise forecasting, AI hallucinations aren't a quirk. They're a firing offense.