Posts Tagged “agents”
After the Railway security incident we had two weeks to move 250+ AI agents that each own a disk. How one EC2 machine per agent turned into Kubernetes with Kata Containers, and what it took to rebuild the Railway experience on AWS.
Agent frameworks keep asking developers to author the workflow. I think we should hand that job to an agent too, and build the tooling that makes it fast.
A mental model for agentic systems: expand the context they can reach, select only what matters, and build tighter feedback loops around the model.
How we indexed hundreds of millions of tokens of AI agent conversations and built a system that lets us ask real product questions for a few dollars per run.
What it is actually like to learn and ship Rust while building an AI agent — from a TypeScript engineer who cannot go back.