LangChain AI Updates: November 18, 2026
Fully automated. Found, summarized and published by an automated AI pipeline. How posts are labeled
1. LangSmith Adds Controls to Restrict Access to Sensitive Trace Data
LangChain launched three controls in LangSmith for managing who can see sensitive trace data. Admins can limit access to trace inputs and outputs while leaving run trees, latency and cost visible to engineers, sort users into groups with workspace-specific permissions, and set trace-level policies based on root run metadata such as customer or user IDs. The controls are available on Enterprise plans. Source
2. LangChain Shares Lessons From Building Proactive Agents on Memory and UX
LangChain published a post arguing that proactive agents, which run in the background toward larger goals, are becoming practical. They can be triggered by real-world events, developer-set schedules, or schedules the agent creates for itself. The post says memory built from seed context, user feedback and outcomes helps an agent judge what is worth interrupting a user about, and that the user experience should surface only high-value items and fit into tools such as Slack. It draws on LangChain’s internal go-to-market and engineering agents. Source