Jev's creator argues software needs calibrated System 1 models

What's this about?

Diogo Almeida says Jev is optimised for fast, structured decisions consumed by code, with explicit confidence useful for thresholds and branching.

What exactly is a System 1 model?

The label borrows from fast, intuitive human judgement: classify or choose quickly instead of reasoning through a long chain. It is TypeSafe's product framing, not a universally agreed model category.

What's the community saying?

1. Builders explore classification, tool routing, game control and data extraction.
2. Critics ask how Jev differs from existing classifiers and whether speed hides lower capability.
3. Almeida argues workflow-specific evaluation should replace trust in public leaderboards.

Podcast deep summary

TLDR: Almeida argues that automation needs small, calibrated decisions which software can compose, while public benchmarks are gameable and deployment-specific evaluation matters more. His practical advice is to decompose workflows, choose thresholds from real examples and measure reliability continuously. Chapters: 00:02:50 System 1 models; 00:05:54 calibration and mode collapse; 00:24:59 RLCD versus RLHF/RLVR; 00:39:55 reliability; 00:54:04 API primitives; 00:58:28 building with Jev; 01:18:28 AI disappearing into software; 01:38:48 coding agents; 02:08:00 multi-agent future. Short quotes from Diogo Almeida: “the goal is for code to be the consumer” [00:03:52]; “public benchmarks are extremely gameable” [00:19:20]; “the solution is begging in a system message” [02:17:22].