2026 in LLMs — Simon Willison, Datasette
Simon Willison used AI coding agents to locate a beeping detector buried in 40 boxes. At CODE Summit 2026, the Datasette creator traces three LLM inflection points: reliable coding agents, persistent problem solving and vibe coded game development. Pelican benchmarks, prompt injection and rogue agents frame a broader question about developer productivity: why does software engineering still feel harder? Speaker info: - https://github.com/simonw - https://x.com/simonw - https://simonwillison.net/ Timestamps: 0:00 - The coding agent inflection point 1:43 - More ambition and predictions for 2026 3:57 - Pelicans, parrots and model benchmarks 5:23 - Frontier models, security and export controls 7:49 - Defining goals models can solve 8:42 - Finding a hidden beep with coding agents 11:22 - Rogue agents and AI security 13:42 - From raccoon heists to playable games 15:37 - Making vibe coded games fun 17:33 - Why software engineering still feels harder




Join the discussion
Have something to add?
Sign in to reply, follow this thread, and earn XP when your answer helps someone.