LangChain AI Updates: July 16, 2026

1. LangSmith Fleet Adds One-Click Deployment of Agents Into Slack

LangChain. LangChain added a capability to LangSmith Fleet that lets teams deploy custom AI agents directly into Slack without writing code, so one person can build an agent the whole team can use inside their existing workflow. The feature supports natural language agent creation from templates, gives each agent its own Slack identity, and adds in-thread approval workflows for sensitive actions. Administrators retain control over permissions, credentials, and spending limits. Source

2. LangChain Introduces LangSmith Sandboxes for Isolated Agent Execution

LangChain. LangChain introduced LangSmith Sandboxes, isolated computing environments that let agents run code, install packages, and iterate without affecting host infrastructure. Each agent task gets a hardware-virtualized microVM with its own kernel, filesystem, and network boundary, plus snapshot-and-fork functionality for branching parallel execution paths. The service also provides credential injection through an auth proxy and sub-second provisioning aimed at coding, data analysis, and research agent workflows. Source