News Hub

LangSmith Engine v2 adds red teaming and automated testing capabilities for agent systems.

LangSmith Trajectories provides conversational view of agent sessions for easier debugging.

Managed Deep Agents 0.8 adds user credentials, memory, HTTP channels, and web search tools.

LangChain released LangSmith Fine-Tuning and SmithTune CLI for model post-training without manual data pipelines.

LangSmith Custom Apps enables building custom interfaces around agent data without hosting setup.

LangSmith provides clinical review tools for healthcare AI teams to create evaluators and release gates.

Jev is now available as a judge model in LangSmith Evals for evaluating agent traces with structured feedback.

Researchers compared Jev against LLM judges on accuracy, repeatability, latency, and cost for agent evaluation.

TypeSafe AI's System One model makes fast, structured decisions and integrates with LangChain agent loops.

LangChain released Deep Life Sci, an open source agentic assistant accessing 600K+ clinical trials, 29M PubMed abstracts, and 12M full-text articles for scientists.

Included Health built Dot, a federated healthcare navigation agent using Deep Agents, LangGraph, and LangSmith with human handoff capabilities.

Healthcare AI agents from Madrigal, Abridge, and Vizient compress manual review hours into minutes and integrate data across multiple systems.

LangChain built a GTM agent that increased lead conversion by 250% and saved sales reps 40 hours monthly.

Credit Genie uses OpenWiki to automate repository documentation and provide searchable codebase context.

LangChain's Connections feature manages credentials and per-caller identity for Managed Deep Agents using OAuth.

LangChain's deepagents context modes let subagents inherit or isolate from supervisor context for efficient multi-agent coordination.

LangChain added MCP support in langchain.mcp module using FastMCP spec with elicitation as LangGraph interrupt and cached tool lists.

Schneider Electric, Vodafone, and monday.com share approaches to scaling AI agents including shared platforms, LLMOps practices, and multi-agent architectures.

LangChain released Managed Deep Agents and LLM Gateway to public beta along with Deep Agents v0.7 and other platform updates.

Podium achieved 98% agent F1 response quality and reduced engineering intervention by 90% using LangSmith.