How To Build A Harness With Jev | A LangChain x TypeSafe Conversation

Hear from Sydney Runkle, Product Manager, OSS at LangChain, Allie Laabs, Founding Developer Advocate at TypeSafe AI, and Hunter Lovell, Technical Lead at LangChain, for a live conversation on building an agent harness with Jev, a new System One model from TypeSafe AI. How To Build A Harness With Jev | A LangChain x TypeSafe Conversation 0:00 Cold open: why agents need a new kind of model 1:44 Welcome and introductions: Sydney, Allie, and Hunter 3:12 What is Jev, and why is everyone excited about it? 5:20 What does it mean that Jev is a System 1 model? 6:34 Do traditional LLMs count as System 2 thinking? 7:53 Why is Jev a code first model, and what is it good at? 12:13 The three question types: Choice, Score, and Noul 17:47 How does Jev fit into an agent harness? 19:57 Live demo: safety classification and model routing 30:53 Is Jev's context length really capped at 32K tokens? 31:27 Any plans for longer context or multimodal support? 32:23 What other Jev use cases are showing up in production? 36:25 What else is on the roadmap? 37:45 Community Q&A: is Jev trained on the same data as an LLM? 38:55 How should you set confidence and probability thresholds? 43:45 Does a large state input degrade Jev's accuracy? 46:41 Closing thoughts: context engineering for Jev Resources: → Building a Harness with Jev: https://www.langchain.com/blog/building-a-harness-with-jev → Jev-as-a-Judge in LangSmith Evals: https://www.langchain.com/blog/jev-is-now-available-in-langsmith-evals → Jev-as-a-Judge for Agent Evals: https://www.langchain.com/blog/jev-agent-evals-langsmith → LangSmith: https://www.langchain.com/langsmith → LangChain Academy: https://academy.langchain.com