TypeSafe's Jev: Agents Building Fast, Cheap Classifiers as Their System One
vesko_st · x · 2026-10-09
The article discusses Jev, released last month by TypeSafe as its first "System One" model, inspired by Kahneman's Thinking, Fast and Slow: System 2 is slow and deliberate, while System 1 is fast and automatic.
Key argument: most agentic systems today never make that hand-off—one frontier model plans, calls tools, and also makes every repetitive judgement (routing messages, deciding whether an email needs a reply) one item at a time. That is expensive, slow, and inconsistent. A better architecture lets the agent act as System 2: figure out how a task should be done, then build, check, and update a fast classifier (System 1) that takes it over.
Traditional classifiers are painful to train, monitor, and keep in sync; an agent can take over that burden: given labels, its natural-language classifiers rival fine-tuning; without them, it can interview the user instead.
More from coding & agent
- Hot reload plus GPT-6.1 ultrafast: edit SwiftUI apps by talking — Dimillian · 2026-10-09
- GPT-6.1 Sol ultrafast demo: 8x faster SwiftUI edits with hot reload — Dimillian · 2026-10-09
- Dev builds his own Gmail AI reply agent that learns from every correction — Al_Grigor · 2026-10-09
- Hone launches Engines: AI that owns business outcomes, with simulation for self-improving harnesses — graceisford · 2026-10-09
- LangChain Podcast Digs Into Decision Models as OpenAI and Databricks Enter the Space — LangChain · 2026-10-09
- Persistent Self-Improvement: Why Having Memory Isn't the Same as Learning — anirudhg9119 · 2026-10-09