LangChain trains custom agent models for LangSmith Engine with Baseten Loops
baseten · x · 2026-09-16
LangChain is using Baseten's Loops training infrastructure to build custom models for LangSmith Engine, its in-platform agent that autonomously debugs and improves users' agents.
- Approach: fine-tuning open-weight models on agent traces to specialize them for agent-specific tasks
- Example task: reading a connected GitHub repo to diagnose why an issue occurs, then drafting prompts or code changes
- The goal is for LangChain to own the intelligence underpinning its products rather than relying fully on general-purpose models
A useful case study for teams turning their own agent traces into specialized vertical models.
Related event: LangChain Trains Custom Agent Models on Baseten Loops(2 posts)→
More from coding & agent
- Dev slams agent sandbox pricing as 20x+ the cost of a $6/month always-on VPS — Aryvyo · 2026-09-16
- Abacus.AI says Smaug Flash fixes open-source models' tool-call hangs in production — bindureddy · 2026-09-16
- Dev lets AI agents reverse engineer Zapier and build his own automation app — NathanWilbanks_ · 2026-09-16
- 60fps launches MCP giving coding agents real UI motion references — 60fpsdesign · 2026-09-16
- Compound Engineering v3.26 ships multi-model design bake-offs and annotatable web prototypes — kieranklaassen · 2026-09-16
- Voice mode breaking up, Codex erroring: reliability still far from AGI-ready — koltregaskes · 2026-09-16