TL;DR of my new article: WTF is Jev by @typesafeai, and the 9 things people are already building with it. The thesis: ๐ฎ ๐ฐ๐ผ-๐ฐ๐ฟ๐ฒ๐ฎ๐๐ผ๐ฟ ๐ผ๐ณ ๐๐ต๐ฎ๐๐๐ฃ๐ง ๐๐ฝ๐ฒ๐ป๐ ๐๐๐ผ ๐๐ฒ๐ฎ๐ฟ๐ ๐ถ๐ป ๐๐๐ฒ๐ฎ๐น๐๐ต ๐ผ๐ป ๐ฎ ๐บ๐ผ๐ฑ๐ฒ๐น ๐๐ต๐ฎ๐ ๐ฐ๐ฎ๐ป๐ป๐ผ๐ ๐๐ฟ๐ถ๐๐ฒ ๐ฎ ๐๐ฒ๐ป๐๐ฒ๐ป๐ฐ๐ฒ, ๐ฎ๐ป๐ฑ ๐ถ๐ป๐๐ถ๐ฑ๐ฒ ๐ณ๐ฎ ๐ต๐ผ๐๐ฟ๐ ๐ฑ๐ฒ๐๐ฒ๐น๐ผ๐ฝ๐ฒ๐ฟ๐ ๐๐ถ๐ฟ๐ฒ๐ฑ ๐ถ๐ ๐ถ๐ป๐๐ผ ๐ฒ๐๐ฒ๐ฟ๐ ๐ฐ๐ต๐ฒ๐ฎ๐ฝ ๐ท๐๐ฑ๐ด๐บ๐ฒ๐ป๐ ๐ฐ๐ฎ๐น๐น ๐ฎ๐ป ๐ฎ๐ด๐ฒ๐ป๐ ๐บ๐ฎ๐ธ๐ฒ๐.
Think AI multiple choice, not AI essay writing. It doesn't chat. You hand it app state plus a typed question, it hands back a decision with a probability attached. ๐ฏ๐ญ.๐ฐ๐ ๐๐ถ๐ฒ๐๐ on the launch post in two days (@CompleteSkeptic, who co-invented RLHF). I ran @slashlast30days on it 11 times, then checked every big post by hand.
๐ ๐ ๐๐ถ๐ป๐ ๐ผ๐ฝ๐ฒ๐ป ๐๐ผ๐๐ฟ๐ฐ๐ฒ ๐ฏ๐ฟ๐ผ๐๐๐ฒ๐ฟ ๐ฎ๐ด๐ฒ๐ป๐ ๐ณ๐ผ๐๐ป๐ฑ ๐ณ๐น๐ถ๐ด๐ต๐๐ ๐ถ๐ป ๐ณ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ๐ ๐ณ๐ผ๐ฟ $๐ฌ.๐ฌ๐ฌ๐ฏ๐ต. New action space every step, DOM as state, Jev picks the click, a small LLM only wakes up to type. The Browser Use founder built it (@gregpr07, 7.2K likes, 1.8M views) and had to note the video is 1x speed
๐งน The sleeper: instant compaction. Score every tool call, drop the junk, skip the summarization prompt entirely. "๐ช๐ฏ 2026, ๐ธ๐ฉ๐บ ๐ช๐ด ๐ค๐ฐ๐ฎ๐ฑ๐ข๐ค๐ต๐ช๐ฐ๐ฏ ๐ด๐ต๐ช๐ญ๐ญ ๐ข ๐ด๐ถ๐ฎ๐ฎ๐ข๐ณ๐ช๐ป๐ข๐ต๐ช๐ฐ๐ฏ ๐ฑ๐ณ๐ฐ๐ฎ๐ฑ๐ต?" asked @tamarajtran, 5K likes, then shipped the answer that afternoon. Run as a Claude plugin it took a session ๐ณ๐ฟ๐ผ๐บ ๐ญ๐ ๐๐ผ๐ธ๐ฒ๐ป๐ ๐๐ผ ๐ด๐ฒ๐ ๐ถ๐ป ๐ผ๐ป๐ฒ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ (@altryne). Diogo's reply: "๐ง๐ณ๐ฆ๐ฆ ๐ค๐ฐ๐ฅ๐ช๐ฏ๐จ ๐ข๐จ๐ฆ๐ฏ๐ต๐ด ๐ง๐ณ๐ฐ๐ฎ ๐ฅ๐ฆ๐ด๐ช๐จ๐ฏ๐ช๐ฏ๐จ ๐ข๐ณ๐ฐ๐ถ๐ฏ๐ฅ ๐ต๐ฉ๐ฆ ๐๐ ๐ค๐ข๐ค๐ฉ๐ฆ"
๐ก๏ธ Vercel put it in production as the safety reviewer in fx auto mode. ๐จ๐ฝ ๐๐ผ ๐ญ๐ด๐ ๐ณ๐ฎ๐๐๐ฒ๐ฟ ๐ฎ๐ ๐ฝ๐ต๐ฑ ๐ฎ๐ป๐ฑ ๐บ๐ผ๐ฟ๐ฒ ๐ฎ๐ฐ๐ฐ๐๐ฟ๐ฎ๐๐ฒ than the model it replaced, per @rauchg, 3.7K likes. LangChain open-sourced the same idea the next day as AutoModeMiddleware. The closed danger classifier inside every coding harness is now a 100ms primitive
๐ฆ Model routing as middleware instead of a paragraph in a system prompt. About a dozen lines, probabilities left in agent state so you can audit the choice. The LangChain writeup by @sydneyrunkle is the cleanest how-to-wire-it piece anyone has published
๐ RAG precision, solved the dumb way: retrieve as usual, run Jev on every chunk, delete the irrelevant ones. "๐ข๐ญ๐ด๐ฐ ๐ฅ๐ช๐ฅ ๐ข๐ฏ๐บ๐ฐ๐ฏ๐ฆ ๐ณ๐ฆ๐ข๐ญ๐ช๐ป๐ฆ ๐ซ๐ฆ๐ท ๐ด๐ฐ๐ญ๐ท๐ฆ๐ฅ ๐ฑ๐ณ๐ฆ๐ค๐ช๐ด๐ช๐ฐ๐ฏ ๐ช๐ฏ ๐๐๐?" (@kushbhuwalka, 416 likes)
๐ฎ Minecraft in real time: ๐๐ฒ๐ ๐ฟ๐ฒ๐ฎ๐ฐ๐๐, ๐๐ฃ๐ง-๐ฒ ๐๐๐๐ฟ๐ฎ ๐ฝ๐น๐ฎ๐ป๐, and they fight multiple zombies at once (@wuyang_zhou). A launcher that reads intent on every keystroke in about 100ms (@dabit3). TypeSafe's own demo is Doom at 10 decisions a second, roughly $7 an hour
๐ฌ Email triage at scale: 1,500 emails in batches of 100 with 8 workers, 60,996 views on the demo. "๐๐ฆ ๐ฐ๐ฏ๐ญ๐บ ๐ฉ๐ข๐ท๐ฆ ๐ข ๐ฃ๐ข๐ญ๐ข๐ฏ๐ค๐ฆ ๐ฐ๐ง $5 ๐ฅ๐ฐ๐ธ๐ฏ ๐ฉ๐ฆ๐ณ๐ฆ, ๐ธ๐ฉ๐ช๐ค๐ฉ ๐ซ๐ถ๐ด๐ต ๐ด๐ฉ๐ฐ๐ธ๐ด ๐ฉ๐ฐ๐ธ ๐ค๐ฉ๐ฆ๐ข๐ฑ ๐ต๐ฉ๐ช๐ด ๐ฎ๐ฐ๐ฅ๐ฆ๐ญ ๐ช๐ด"
๐๏ธ ๐ณ๐ณ๐ณ ๐ท๐๐ฑ๐ด๐บ๐ฒ๐ป๐๐ ๐ถ๐ป ๐๐ป๐ฑ๐ฒ๐ฟ ๐ฌ.๐ณ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ๐ ๐ณ๐ผ๐ฟ ๐ฎ ๐พ๐๐ฎ๐ฟ๐๐ฒ๐ฟ ๐ผ๐ณ ๐ฎ ๐ฐ๐ฒ๐ป๐. Every's head of evals asked 21 questions of 37 documents in one request, and that is what came back
๐งช Jev in your browser: Reflex, a Qwen model doing structured decisions on WebGPU, built at Shopify by @kshetrajna and passed around by @tobi. Three independent clones inside 72 hours. ๐ง๐ต๐ฒ ๐ถ๐ป๐๐ฒ๐ฟ๐ณ๐ฎ๐ฐ๐ฒ ๐ถ๐ ๐๐ต๐ฒ ๐ถ๐ป๐๐ฒ๐ป๐๐ถ๐ผ๐ป, ๐ป๐ผ๐ ๐๐ต๐ฒ ๐๐ฒ๐ถ๐ด๐ต๐๐
๐ Already behind the gateways you use: @vercel AI Gateway inside 48 hours (2,341 likes, the company's second-biggest post), Cloudflare, and @OpenRouter in beta
๐ธ ๐ฑ,๐ฌ๐ฌ๐ฌ ๐ฟ๐ฒ๐พ๐๐ฒ๐๐๐ ๐ณ๐ผ๐ฟ ๐ฎ๐ฏ๐ผ๐๐ $๐ฎ. That was one developer counting his bill on day one (@MichaelLee04, 3,060 likes). Input is $0.042 per million tokens. Output is free
๐งจ The honest part: Every's second test came out ๐ฎ๐ฑ๐ ๐ณ๐ฎ๐๐๐ฒ๐ฟ, ๐ป๐ผ๐ ๐ฎ๐ฌ๐ฌ๐ , and Jev caught 6 of 7 planted defects to Fable 5.1's 7. The HN launch thread (1,863 points) spent most of its length on "can't hallucinate." Top critical comment: "๐ช๐ต ๐ค๐ข๐ฏ'๐ต ๐ฆ๐ฎ๐ช๐ต ๐ข๐ฏ ๐ช๐ฏ๐ท๐ข๐ญ๐ช๐ฅ ๐ต๐บ๐ฑ๐ฆ, ๐ฃ๐ถ๐ต ๐ช๐ต ๐ค๐ข๐ฏ ๐ด๐ต๐ช๐ญ๐ญ ๐ฆ๐ฎ๐ช๐ต ๐ข ๐ค๐ฐ๐ฎ๐ฑ๐ญ๐ฆ๐ต๐ฆ๐ญ๐บ ๐ธ๐ณ๐ฐ๐ฏ๐จ ๐ท๐ข๐ญ๐ช๐ฅ ๐ท๐ข๐ญ๐ถ๐ฆ." Diogo called the "it's a zero-shot classifier" read "๐ท๐ฆ๐ณ๐บ ๐ข๐ค๐ค๐ถ๐ณ๐ข๐ต๐ฆ!" And the biggest Reddit thread is someone who open-sourced the same architecture a year ago, 1,568 upvotes. Top reply: "๐๐ถ๐ต ๐ฅ๐ช๐ฅ ๐บ๐ฐ๐ถ ๐ฑ๐ฐ๐ด๐ต ๐ช๐ต ๐ด๐ข๐บ๐ช๐ฏ๐จ ๐ช๐ต'๐ด ๐ต๐ฉ๐ฆ ๐ฏ๐ฆ๐น๐ต ๐ฃ๐ช๐จ ๐ต๐ฉ๐ช๐ฏ๐จ? ๐๐ฐ๐ฐ๐ฌ๐ช๐ฆ ๐ฎ๐ช๐ด๐ต๐ข๐ฌ๐ฆ"
Bonus: the name is not Kahneman. It's William Stanley Jevons, of Jevons paradox. Make a resource cheaper and people consume far more of it. Naming your decision model after that is a thesis statement.
๐๐ฒ๐ฒ๐ฝ ๐๐ต๐ฒ ๐ฏ๐ถ๐ด ๐บ๐ผ๐ฑ๐ฒ๐น ๐ณ๐ผ๐ฟ ๐๐ต๐ฒ ๐ต๐ฎ๐ฟ๐ฑ ๐๐ต๐ถ๐ป๐ธ๐ถ๐ป๐ด ๐ฎ๐ป๐ฑ ๐๐ฟ๐ถ๐๐ถ๐ป๐ด. ๐จ๐๐ฒ ๐๐ฒ๐ ๐ณ๐ผ๐ฟ ๐๐ต๐ฒ ๐ฟ๐ฎ๐ฝ๐ถ๐ฑ-๐ณ๐ถ๐ฟ๐ฒ ๐ฑ๐ฒ๐ฐ๐ถ๐๐ถ๐ผ๐ป๐ ๐ถ๐ป ๐ฏ๐ฒ๐๐๐ฒ๐ฒ๐ป. That's the whole article.
