Every on Claude Fable 5.2: Human Thinking at the Start Matters More as AI Builds More
every · x · 2026-11-19
Every's team recaps their Claude Fable 5.2 camp, where they gave the model bigger projects than usual: Kieran Klaassen built Everycut, a video editor with an agent inside, and Jack Cheng turned a photo of a toddler's playroom into a racing game.
The core idea is Compound Engineering: as AI takes on more building, human thinking at the start matters even more — ask questions, develop strategy, push back on assumptions, and turn that thinking into work the model can execute.
They also share cost lessons: Fable 5.2 is slow and token-hungry, so they explain when to reach for it versus faster models, and how to set up a job the model can work on overnight. Full recording requires an Every subscription.
More from coding & agent
- Semantic caching LLM answers by embedding similarity is unsafe — 138-pair test shows why — Narrow_Side_7778 · 2026-11-19
- Can ComfyUI refine existing mocap animations? A motion-to-motion workflow question — _Tito5_ · 2026-11-19
- Mycelium: an open-source framework that turns analytical project knowledge into a living dataset — arjunrajlab · 2026-11-19
- harness-hub open-sources RL-trained LoRA adapters tuned per agent harness, recipe coming Friday — yb2698 · 2026-11-19
- Hermes adapter solves a SWE-Bench Pro task in 35 tool calls where the base Qwen3.8 model failed — yb2698 · 2026-11-19
- NousResearch's Hermes adapter lifts Qwen3.8-27B agent scores: SWE-Bench Pro 29.4%→43.1% at 4× fewer tokens — yb2698 · 2026-11-19