Archive
News
Every story on this site is rewritten from public posts. Original URLs sit at the bottom of each article.
Almeida’s bet is Jevons paradox, not a cheaper chatbot
In a run of replies three days after launch, the TypeSafe founder argued that Jev only works if cheap decisions create more software, not if they nibble LLM spend.
A week of Jev repos — routers, MCP, graphs, games
A widely shared Awesome Jev thread listed fourteen working patterns; community directories passed a hundred projects within days. The repeating move is choose / score / stop.
Browser Use wires Jev into the click loop — 7 seconds, $0.0039
Gregor Zunic showed a 1x-speed Google Flights run where Jev picks the action and DOM target each step, and a small LLM types only when it has to.
Stagehand, Chrome, and a11y trees — computer use as a Jev question
Browserbase’s Kyle Jeong and other builders recast computer-use as observe → typed questions → execute, reporting sub-cent, near-instant remote-browser runs.
Compaction without summarization — score the tool calls and drop them
Tamara Tran argued that agent context compaction is still a summarization prompt in 2026, and that Jev can instead keep or drop tool output verbatim.
TypeSafe starts pulling people off the Jev waitlist
A TypeSafe builder said 20,000 people would be released in an hour. Early users reported same-day access; Helium was a known signup snag.
TypeSafe leaves stealth and ships Jev, a model that refuses to chat
Former OpenAI researcher Diogo Almeida posted a two-year stealth project on X — a System One model named Jev, built for typed software decisions instead of generated text.
What Jev actually returns — System One in plain language
Jev is not a smaller LLM with stricter JSON mode. TypeSafe defines a System One model as parallel, typed, calibrated decisions with no string generation at all.