- OpenAI can’t win just by building better models or bigger data centers. It has to move up the stack and become the default AI operating layer where workflows, developers, and enterprises get locked in. https://www.ben-evans.com/benedictevans/2026/2/19/how-will-openai-compete-nkg2x
- When AI can write most of the code, the leverage shifts from implementing logic to designing the system around the AI. What starts to matter more is how you define constraints, connect tools, structure feedback loops, and create an environment where the agent can read, plan, execute, and iterate reliably. Codex makes this clear. The real power is not just the model, but the harness and surfaces that turn intelligence into execution. Engineering does not go away. It moves up a level, from writing code line by line to shaping how code gets written and shipped. https://openai.com/index/harness-engineering/
- In mid-December, AWS had a 13-hour outage affecting one of its customer systems after engineers allowed its Kiro AI coding tool to make changes autonomously. The tool determined that the best course of action was to delete and recreate the environment, which ended up causing the disruption. The incident has led some employees to question how useful AI tools are for most of their work, given the risk of errors. At the same time, the company has set a goal for 80 percent of developers to use AI for coding tasks at least once a week and is closely tracking adoption. https://archive.is/WX2a2
- AI made coding more enjoyable; very much enjoying this phase. https://weberdominik.com/blog/ai-coding-enjoyable
- Some foods we still eat today like ginkgo nuts have stayed basically unchanged for over 100 million years, so we’re kind of eating the same things that existed back when dinosaurs were around. https://borischerny.com/food/2022/01/17/Dinosaur-food.html
What I’ve been reading (W8, 2026)
AI strategy, the shift from writing code to designing how code gets written, and foods older than the dinosaurs