{"slug": "openwiki-brains-is-langchains-memory-enhancer-for-llms", "title": "OpenWiki Brains is LangChain’s Memory Enhancer for LLMs", "summary": "LangChain released OpenWiki Brains, a memory enhancement for large language models that builds a live-updating wiki from external sources such as Gmail, Notion, and X, nine days after the initial OpenWiki launch. The open-source tool gives LLMs a persistent, self-updating knowledge base to retain context across sessions, with connectors for Gmail, Notion, X, Hacker News, and web search, and support for any self-reasoning LLM.", "body_md": "As projects evolve, the agents working on them should keep pace.\n\nLangChain has expanded its OpenWiki ingestion engine just 9 days after the software’s initial launch to give LLMs a place to hold additional context and live updates about the tasks they are undertaking.\n\nIn other words, OpenWiki has been given a brain that updates itself with new information from other user apps, such as Gmail or the X social media service.\n\nThe [open source AI software provider](https://www.langchain.com/about) initially released OpenWiki to help agents better understand the codebases they were working on. The CLI-based tool generates a wiki that summarizes a codebase for a model that needs to work on the project. If the codebase changes, the software updates its wiki.\n\nThe new package, called [OpenWiki Brains](https://www.langchain.com/blog/introducing-openwiki-brains-general-purpose-wiki-memory-for-agents), extends the live updating concept with additional outside material, which could come from e-mail, note-taking apps, other repos, social media and other news sites.\n\nOpenWiki Brains works with any self-reasoning LLM. The user installs the software on their own machine (“*npm install -g openwiki*”), giving it the API key for the preferred LLM (or pointing it to the localhost).\n\nThe software creates a directory (*~/.openwiki/wiki/*) and stores the material in markdown files. Finally, the user gives it a set of sources to monitor and draw data from, by using pre-written connectors.\n\n**Proactive Memory**\n\nToday’s LLMs operate largely from guesswork. They are fed explicit information from the user, either by prompt or through agent instructions. From there, they can gather additional information for their tasks, though everything is lost (or must be rebuilt from scratch) when the session ends.\n\nOpenWiki Brains gives the LLM a place to store operational data.\n\nOf course, agents already have their own memory, but their limitation is, according to the LangChain, that these memory stores are reactive in nature. Give it some information once, and that data doesn’t usually get updated.\n\nProjects evolve, however. New objectives arrive by e-mail or chat. The user becomes the bottleneck: With OpenWiki Brains, the looping agent could get started before the developer even begins work for the day.\n\n**A Personal Brain **\n\nThe original OpenWiki only kept tabs on a git-based code repository. It made detailed notes on the file relationships, the history of the repository, the developer’s coding conventions, and so on. The secret sauce is that as the model or user updated the repository, OpenWiki would capture those updates for future sessions.\n\nThe “Personal Brain,” as this new mode is called, expands this concept by acting as a virtual assistant of sorts. The user specifies an area of interest and, from there, OpenWiki builds a live knowledge base of what the user is working on, collecting data through e-mail, chat sessions and news articles of interest.\n\nThe Personal Brain gathers this info through connectors. LangChain has posted the first set for Gmail, Notion, the X social media site, the Hacker News site, and general Web search. The user describes what they are looking for specifically from these sources, and these instructions get passed along to the LLM, which periodically checks the sources for new material that is relevant to the task at hand.\n\nLangChain is working on additional connectors–Slack being one of the most needed connectors. It hasn’t formally announced a plug-in program where users could contribute their own.\n\n**The Power of the Wiki**\n\nLangChain is not alone in its use of the Wiki. It turns out that the open directory Wiki format, which allows anyone to make changes to files, is a very handy scratchpad for LLMs, especially when used in conjunction with the Markdown format.\n\nEarlier this year, AI pioneer [Andrej Karpathy](https://techstrong.ai/features/karpathy-agent-portends-ai-final-boss-battle/) proposed using the wiki format to capture everything that agents learn in the course of doing their work. He called it the “[Second Brain](https://techstrong.ai/features/karpathys-instructions-for-building-an-ai-driven-second-brain/).” His approach was then formalized by Google, which offered a standardized template for using a Wiki and Markdown, [called the Open Knowledge Format](https://techstrong.ai/articles/google-launches-a-universal-format-for-karpathys-llm-wiki/).", "url": "https://wpnews.pro/news/openwiki-brains-is-langchains-memory-enhancer-for-llms", "canonical_source": "https://techstrong.ai/articles/openwiki-brains-is-langchains-memory-enhancer-for-llms/", "published_at": "2026-07-21 00:15:14+00:00", "updated_at": "2026-07-21 00:37:55.153391+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "developer-tools", "ai-products"], "entities": ["LangChain", "OpenWiki Brains", "OpenWiki", "Gmail", "Notion", "X", "Hacker News", "Andrej Karpathy"], "alternates": {"html": "https://wpnews.pro/news/openwiki-brains-is-langchains-memory-enhancer-for-llms", "markdown": "https://wpnews.pro/news/openwiki-brains-is-langchains-memory-enhancer-for-llms.md", "text": "https://wpnews.pro/news/openwiki-brains-is-langchains-memory-enhancer-for-llms.txt", "jsonld": "https://wpnews.pro/news/openwiki-brains-is-langchains-memory-enhancer-for-llms.jsonld"}}