Simon Willison introduces “Deep Blue,” a term coined on the Oxide and Friends podcast for the psychological ennui and creeping existential dread many developers feel as generative AI encroaches on their craft. It’s a useful frame for talking about the emotional impact of coding agents and the changes that have swept through the industry so quickly. It’s the other side of the exhileration I sometimes feel when I can solve problems I never would have bothered with incredibly quickly.

Kurt Mackey argues that ephemeral sandboxes are the wrong approach for AI agents. Instead, agents work better with persistent environments where they can maintain context across sessions and avoid rebuilding their workspace repeatedly.

Simon Willison’s 2025 recap maps the year’s biggest LLM shifts, from inference-scaled reasoning models and tool-using agents to the breakout of coding CLIs like Claude Code. It’s a dense, opinionated timeline linking major releases to what actually changed for developers and everyday workflows.

A deep dive into how LLM prompt caching works under the hood, focusing on the transformer attention mechanism and the exact data providers reuse between requests. This is also one of the most accessible explanations of how LLMs work that I’ve encountered. The visuals are really clear, and the step by step walkthrough is incredibly clear. Via Simon Willison.

Via kottke.org, Robin Sloan describes himself as the programming equivalent of a home cook. I’ve been working in professional kitchens for a really long time, but lately I’ve rediscovered the joy of home cooking myself.

Simon Willison released a Python CLI tool that converts Claude Code sessions into shareable HTML pages with more detail than Claude Code itself provides, including hidden thinking traces.

A compelling reframing of Kubernetes as ‘a runtime for declarative infrastructure with a type system’ rather than just a container orchestrator.

A reflection on Sander Schulhoff’s “AI security crisis” discussion, framing LLM and agent threats as social engineering at scale. It explains why simple-sounding mitigations like restricting access and capabilities are hard to implement in practice—and why agentic tools (especially browsers) demand extra caution.

A practical walkthrough of adding semantic search to a Hugo site hosted on Cloudflare Pages while spending as close to $0 as possible. It covers the core pieces—embeddings, indexing, a vector database, and a small backend wrapper—plus notes from building it with Claude Code.

Welcome to my tech blog.