Article
CommunityJev Guides & Articles
ThinkThen: why people are excited about Jev
Ian Maurer's ThinkThen analysis argues Jev's faster and cheaper are measurable today while its better is a bet on RLCD: unlike RLHF, which pleases humans, and RLVR, which optimizes benchmarks, Jev is trained to be reliable for programmatic use, in Diogo Almeida's own framing.
- Format
- Article
- Reading time
- 3 min
- Published by
- Community
- Author
- Ian Maurer
- Last verified
- Oct 2026
Highlights
- Maurer's framing: people are excited because Jev is faster and cheaper, and potentially better; faster and cheaper are measurable today, better is a bet.
- Almeida, quoted from the podcast interviews: RLCD is make it reliable for programmatic use, while RLHF is please humans and RLVR is optimized benchmarks.
- Almeida on calibration: when Jev says 0.8 it should be right about 8 times in 10, and he wants people to program against Jev without making example queries.
- Tool-choice example: search gets 0.84 against a 0.7 threshold, so code calls search; if no tool clears 0.7 the code stops and asks a person.
- Maurer's verdict: calibration is still a claim until you check it on your own cases, so label real examples and measure before picking a threshold.
Watch out
Copy reviewed against the published post on 2026-10-17. The piece is a quote-driven blog analysis assembled with ThinkThen, and its claims are attributed to Maurer's own talk and to Almeida on the Latent Space and a16z podcasts; Maurer states that nobody has proven the RLCD approach wins.
Questions about this entry
Short answers drawn from the linked project or page.
That training a model for decisions rather than pleasing people produces a genuinely better tool. Faster and cheaper can be measured today, Maurer writes, but better is a bet on TypeSafe's RLCD method, which nobody has yet proven wins.
Using Almeida's own three-word summaries: RLHF is please humans, RLVR is optimized benchmarks, and RLCD, TypeSafe's method for Jev, is make it reliable for programmatic use.
Almeida says a downside of RLHF is mode dropping: the model stops giving rare answers, like GANs that drop minority classes. Its sense of how sure it should be is gone, which he calls total poison for the probability distributions.
If Jev says 0.8, it should be right about 8 times in 10 across a population of predictions. That lets code set a threshold, branch on it, and send the cases Jev is not sure about to a person.
An agent must pick among tools at each step. If the search tool gets 0.84 and the threshold is 0.7, the code calls search; if no tool clears 0.7, the code stops and asks a person. Maurer says that check takes milliseconds and costs a fraction of a cent.
He says the highest honor of reliability would be getting to the point where people can program against Jev without making example queries.
Calibration is still a claim until you check it on your own cases. Maurer advises labeling real examples and measuring before picking a threshold, and points to ThinkThen's how-to articles on testing before you trust it.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Hacker News: Simon Willison's overview of Noul, Choice, and Score, and where a decision model fits.
benwerd53 pointsJev introduces a new shape of LLMRead the thread on Hacker NewsHacker News: swyx posts the Latent Space interview with Diogo Almeida.
swyx4 pointsJev: System One Models for Prod, Not God – With Diogo Almeida, CEO, TypeSafe AIRead the thread on Hacker NewsX: Diogo Almeida in five minutes
X: LLMs vs. Jev, clearly explained
More Jev resources like this
A BIG CHANGE analysis of the Diogo Almeida Latent Space interview: Jev's most consequential proposition is the unit of automation, a bounded judgment inside an existing process, and the launch claim that Jev can't hallucinate needs a narrow reading because schema matching does not establish truth.
Jev Guides & ArticlesArticle#community#press#explainer
Community
TWIML episode 779 features TypeSafe co-founder and CEO Diogo Almeida on Jev, the company's model for bringing fast, reliable intelligence directly into software, and why text-generating models are poorly suited to many real-world automation decisions.
Jev Guides & ArticlesArticle#community#press#podcast
Community
Pillar Security analyzes Jev and System One models: a smart-home demo answered three typed questions in about 200 ms with no language model in the loop, while prompt scanning, output filtering and chain-of-thought monitoring all have nothing to inspect. TypeSafe's docs say state is not treated as hostile by default.
Jev Guides & ArticlesArticle#community#press#security
Community
Rogue Security's post uses Jev and System One models to frame agentic security as a decision problem: the same send_file action should be allowed, blocked or reviewed depending on who requested it and where the data goes, illustrated with policy case files rather than product results.
Jev Guides & ArticlesArticle#community#press#security
Community
A podcast episode on Jev as a decision model rather than a chat model: hosts Luis Calderon and Carina discuss roughly 20x faster Coach testing, why that preliminary figure differs from TypeSafe's vendor benchmarks, and why type safety does not guarantee a correct decision.
Jev Guides & ArticlesArticle#community#press#podcast
Community
On The MAD Podcast, Matt Turck talks with TypeSafe CEO Diogo Almeida about why intelligence per dollar matters more than enterprise adoption, and why RLVR benchmark gains still leave models 95% Einstein, 5% Mr. Bean.
Jev Guides & ArticlesArticle#community#press#podcast
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.