LangChain Launches OpenWiki Brains for Agent Memory

• OpenWiki Brains lets AI agents collect and update context from multiple sources automatically.
• Supports Gmail, Notion, git, Twitter/X, Hacker News and web search integration.
• Agents stay current on a user's projects and interests without the user copying context into every session.

LangChain released OpenWiki Brains on July 10, an open-source framework that lets AI agents pull context from tools like Gmail, Notion, git repositories, Hacker News and web search, then store it as plain Markdown files on the user's machine for the agent to read during a task. The goal is to avoid the copy-paste method of transferring context between projects and to automatically update that context. 

The release includes two modes: Personal Brain, which creates a wiki from connected sources tailored to a user's focus areas, and Code Brain, the framework's original use case, which runs inside a git repo and maintains documentation for the codebase. Because Personal Brain runs locally, OpenWiki keeps it current through a scheduled job that runs on whatever frequency the user sets.

Some connectors are straightforward: Gmail pulls recent emails, Hacker News pulls recent posts, git inspects recent commits. Others, like Notion and web search, use an agent at ingestion time to search based on the user's stated goals.

Slack support is coming. Future connectors for LangSmith traces and Claude Cowork sessions are also on the roadmap, according to the company.

AI Agents Need Persistent Memory to Move Beyond Pattern Matching

AI agents can process large amounts of data, but stay limited without memory of what came before. Persistent memory allows agents to move faster, better personalize responses and reduce hallucinations.

Context graphs and knowledge graphs have picked up attention for a related reason: they store not just data but the relationships and decisions behind it. Vendors pitch this as the shift from pattern-matching to reasoning, arguing that context-aware systems make fewer mistakes and produce work that fits what the business actually wants.

LangChain Background

LangChain, founded in 2023 and headquartered in San Francisco, provides AI infrastructure and developer tooling for teams building agent applications. The company offers open-source frameworks such as LangChain, LangGraph and Deep Agents, along with the commercial LangSmith platform. Its open-source frameworks have reportedly been downloaded more than 1 billion times.

Have a tip to share with our editorial team? Drop us a line: