Any company can use AI. The ones that see lasting value will own their intelligence: the model, the harness, and the context that turn a generic model into one that knows your business.
At Interrupt NYC and London, we shipped new features to enable teams to own their intelligence, giving them more control over how they build, understand, and improve their agents in production.
New in LangSmith
🔎 LangSmith AI Search: Find conversations based on a semantic understanding of your traces, with questions like “did the user get frustrated?” or “did PII leak?” without relying on traditional filters.
🤖 Managed Deep Agents v0.9: Build dynamic, personalized agents that are tailored to each run's context, react in Slack, have user-level memory and tool use, schedule tasks, and more. Plus, built-in web search for free.
🛡️ LangSmith Engine v2: Catch issues before they reach production with Red Teaming. Engine v2 also detects harder-to-spot issues like inefficient agent work and metric trends, and automatically tests and validates its proposed fixes.
🧭 LangSmith Trajectories: Navigate trace data and follow your agent’s path with a chronological view of each session across the main agent and its subagents.
🧩 LangSmith Custom Apps: Build, publish, and run custom interfaces directly in LangSmith, turning recurring workflows into shared tools built on your LangSmith data.
🎛️ LangSmith Fine-Tuning: Turn production trajectories into training data, fine-tune models with Baseten or Fireworks, then evaluate and serve the tuned models.
Open source
🔧 LangChain v1.4.0: LangChain now supports FastMCP through the langchain.mcp namespace.
⚡ Building production agents with Jev and LangGraph: See how LangGraph orchestrates Jev, TypeSafe AI’s decision model, to build agents that are faster and cheaper to run.
The Agentic Operating Model: How teams are running production AI at scale
The Agentic Operating Model looks at how leading teams bring people, processes, and technology together to scale production AI. It covers rethinking the agent development lifecycle, building continuous evaluation practices, and embedding governance: what it takes to move beyond experimentation and run agents reliably at scale. Join an upcoming virtual session to explore the framework and what it looks like in practice.
🌍 EMEA · Oct 2910:00 GMT / 11:00 CETRegister →
🌎 North America · Nov 510:00 AM PT / 1:00 PM ETRegister →
Speak the Lang
Live workshops on the fundamentals of building reliable agents.
🧪 Agent Evals 101: What are agent evals?Wed, Oct 28 · 11:00 AM PT · VirtualLearn how to measure agent performance by assessing its execution trajectory: the sequence of messages and tool calls it produces.Register now →
🧠 Agent Memory 101: Context engineeringTue, Nov 3 · 10:00 AM PT · VirtualA live workshop on how agents remember. See how Deep Agents, our open source agent harness, helps keep context windows lean.Register now →
Upcoming events
Hosted by LangChain
🇩🇪 Munich, Germany · Oct 27 · 17:00–21:00 CETLangChain x Data Reply Hackathon: Building with agent harnessesRegister →
🤖 Chicago, IL · Oct 27 · 10:00 AM CTLangSmith Roadshow: Building agents with agentsRegister →
💻 Virtual · Oct 28 · 11:00 AM PTGTM Engineering Live: How we built our design agentRegister →
🏛️ Vienna, VA · Oct 29 · 6:00 PM ETDMV Agent Builders MeetupRegister →
🍁 Toronto, ON · Nov 4 · 5:15 PM ETOwn your intelligence with open‑weight modelsRegister →
Hosted by the LangChain community
🇲🇽 Mexico City · Oct 29 · 5:00 PM CTAgents in productionLearn more →
See all events →
What teams are building with LangChain
Before, adding a feature to an early version of Zip’s chatbot or agent took weeks, and its in-house traces pipeline was missing details the team needed. Now, LangSmith and LangGraph make shipping new features faster.
See all customer stories →
How can you follow along with the Lang Latest? Check out the LangChain blog, Changelog, and YouTube channel for more product and content updates.
