LangChain announced product updates at Interrupt NYC and London in its October 2026 newsletter, published October 15. The company said the features give teams more control over how they build, understand and improve agents in production, as part of its goal of letting companies own the model, harness and context that shape business-specific AI.

LangSmith AI Search lets users find conversations by semantic meaning in traces, with questions such as whether a user became frustrated or personal information leaked, without relying on traditional filters. Managed Deep Agents v0.9 supports agents tailored to each run’s context, with Slack interaction, user-level memory, tool use, scheduled tasks and built-in web search at no extra charge. LangSmith Engine v2 adds red teaming to catch issues before production, detects inefficient agent work and metric trends, and automatically tests and validates proposed fixes.

LangSmith Trajectories provides chronological views of sessions across a main agent and its subagents. LangSmith Custom Apps lets teams build, publish and run interfaces based on LangSmith data. LangSmith Fine-Tuning turns production trajectories into training data, then supports model fine-tuning with Baseten or Fireworks, followed by evaluation and serving.

LangChain v1.4.0 adds FastMCP support through the langchain.mcp namespace. A LangChain item describes how LangGraph orchestrates Jev, TypeSafe AI’s decision model, to build agents that are faster and cheaper to run. The newsletter also promotes an Agentic Operating Model session on scaling production AI through changes to the agent development lifecycle, continuous evaluation and governance. Sessions are listed for EMEA on October 29 and North America on November 5.

LangChain listed workshops on agent evaluation for October 28 and agent memory and context engineering for November 3, plus events in Munich, Chicago, virtual, Vienna, Toronto and Mexico City from October 27 to November 4. The newsletter says Zip previously took weeks to add features to an early chatbot or agent, while its in-house traces pipeline lacked needed details; LangSmith and LangGraph now make shipping features faster.

LangChain announced product updates at Interrupt NYC and London in its October 2026 newsletter, published October 15. The company said the features give teams more control over how they build, understand and improve agents in production, as part of its goal of letting companies own the model, harness and context that shape business-specific AI.

LangSmith AI Search lets users find conversations by semantic meaning in traces, with questions such as whether a user became frustrated or personal information leaked, without relying on traditional filters. Managed Deep Agents v0.9 supports agents tailored to each run’s context, with Slack interaction, user-level memory, tool use, scheduled tasks and built-in web search at no extra charge. LangSmith Engine v2 adds red teaming to catch issues before production, detects inefficient agent work and metric trends, and automatically tests and validates proposed fixes.

LangSmith Trajectories provides chronological views of sessions across a main agent and its subagents. LangSmith Custom Apps lets teams build, publish and run interfaces based on LangSmith data. LangSmith Fine-Tuning turns production trajectories into training data, then supports model fine-tuning with Baseten or Fireworks, followed by evaluation and serving.

LangChain v1.4.0 adds FastMCP support through the langchain.mcp namespace. A LangChain item describes how LangGraph orchestrates Jev, TypeSafe AI’s decision model, to build agents that are faster and cheaper to run. The newsletter also promotes an Agentic Operating Model session on scaling production AI through changes to the agent development lifecycle, continuous evaluation and governance. Sessions are listed for EMEA on October 29 and North America on November 5.

LangChain listed workshops on agent evaluation for October 28 and agent memory and context engineering for November 3, plus events in Munich, Chicago, virtual, Vienna, Toronto and Mexico City from October 27 to November 4. The newsletter says Zip previously took weeks to add features to an early chatbot or agent, while its in-house traces pipeline lacked needed details; LangSmith and LangGraph now make shipping features faster.