Every's 'frontier team' model: 100 AI experiments are useless if none land
every · x · 2026-10-26
Every shares how its small 'frontier team' tests each new model, assesses impact on products and workflows, and converts useful experiments into articles, products, and tools while the rest of the company stays on the roadmap. Dan Shipper calls the setup 'kind of transformative.'
Key advice from Kieran Klaassen: 100 AI experiments won't help if none land—listen to the people doing the work, make experiments understandable, and drive adoption. 'Dumping everything into a channel is easy. Integration is the job.'
Also covered: a code-review skill for non-coders and using AI to address a cybersecurity incident.
More from coding & agent
- badlogic on his 2026 agent workflow: "I tell them what to do, I yell when they do it badly" — steipete · 2026-10-26
- Indie dev's trip planner fails 3 AI review rounds before making offline prices read-only — SeanOliver · 2026-10-26
- $21 of DeepSeek API wrote Molten: a Rust+Metal engine serving 300+ tok/s on M5 Macs — GabGarrett · 2026-10-26
- EMNLP 2026 Findings: "Ask or Assume?" turns coding agents into proactive help-seekers — najoungkim · 2026-10-26
- Purchasing agent in production: >95% recall, >90% accuracy, author asks how to push F1 higher — Ancient_Ad_916 · 2026-10-26
- Claude Opus 5.5 built an interactive 3D F1 aerodynamics demo you can flip upside down — techartist_ · 2026-10-26