What are system one models?

They are non-generative models that return structured choices, scores, or probabilities, and they are typically much faster and cheaper than reasoning LLMs, also called system two models.

How is Glean using Jev?

Glean uses system one models alongside LLMs and other classifiers, and Jev is especially suited for zero-shot flexibility, so the team does not have to fine-tune models again and again as the decision space changes.

What did UiPath demonstrate with Jev?

UiPath showed a test where Jev ranked document chunks in a tournament, picking the most relevant chunks from small groups and advancing winners to another round. The whole process took under a second, even with thousands of Jev calls.

Large Language ModelsAI Coding AssistantsLatent Space11h agoNov 12, 2026, 16:00 JST

TypeSafe AI's Diogo Almeida: Integrate Jev into code

Send on LINE

3 Key Points

What happened
At the AI Engineer Code Summit, TypeSafe AI CEO Diogo Almeida called Jev "reliable intelligence" and urged engineers to integrate system one models directly into application code.
Why it matters
Almeida argues that code is the language of automation, so making software smarter means calling these fast, non-generative models from inside application code rather than building ever more convoluted LLM pipelines, which he says are not always reliable.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Get these free every morning

Context & Analysis

Jev was released in September and quickly became a popular topic among AI engineers, with a Latent Space podcast episode about its launch becoming the show's most popular of the year. Almeida's keynote built on the reliability problem he discussed then: making LLMs the heart of software can produce inconsistent results, and he argues system one models offer a more dependable foundation for automation.

The summit presentations showed that message being put into practice. Glean's Chinmay Goyal said Jev is used alongside LLMs and other classifiers and is especially suited for zero-shot flexibility, so teams avoid constant fine-tuning as decision spaces change. UiPath's Chibi Vikram and Kevin Bernstein described Jev as a building block they can adjust, and their document-chunk ranking tournament completed in under a second even with thousands of Jev calls. Workday's Scott Roberts showed a prototype that fills in a time-off card while the user is still typing, which he described as going from predicting what you want to say to predicting what you want to do.

Almeida closed by comparing the moment to the early internet, predicting an explosion of what software can do as engineers explore new boundaries, not necessarily because it is valuable but because they can.

FAQ

What are system one models?: They are non-generative models that return structured choices, scores, or probabilities, and they are typically much faster and cheaper than reasoning LLMs, also called system two models.
How is Glean using Jev?: Glean uses system one models alongside LLMs and other classifiers, and Jev is especially suited for zero-shot flexibility, so the team does not have to fine-tune models again and again as the decision space changes.
What did UiPath demonstrate with Jev?: UiPath showed a test where Jev ranked document chunks in a tournament, picking the most relevant chunks from small groups and advancing winners to another round. The whole process took under a second, even with thousands of Jev calls.

Latent SpaceRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.