The Hidden Cost of AI Code Review
Three months after introducing an AI code review tool that caught 40% more issues than our human reviewers, I noticed something troubling. We had optimised away the conversations that make engineers better.
WRITING
All Posts
Most AI Agents Fail in Production. Here is What We Are Getting Wrong
AI agents look impressive in demos. But real-world systems are messy. Here is why most agents fail in production and how system thinking changes everything.
Cursor Just Changed How AI Agents Work in Development
Cursor introduced cloud-based agents that can onboard to your repo, spin up a Linux VM, write code, run tests, push a branch, and open a PR, all without touching your local machine.
