Xmind launches a mind-map AI agent and connects maps to ChatGPT and Claude
The release lets users develop maps inside Xmind or work through outside assistants. Agent access spans every plan, but usage draws on AI credits.
Loading page…
The release lets users develop maps inside Xmind or work through outside assistants. Agent access spans every plan, but usage draws on AI credits.
Listen to this story
Xmind’s launch makes mind maps a working surface for AI-assisted research and planning: users can revise maps through Xmind Agent or connect ChatGPT and Claude to read and update the same files. That shared context could keep decisions and new findings together instead of leaving them in separate chat histories. Access is included across Xmind plans, but usage is credit-based, and the number of tasks covered will vary with their complexity.
Xmind Agent can start from an idea, file or Notion document, and can also research from the web or revise an existing map.
Xmind MCP supports Claude and compatible assistants; a separate ChatGPT plugin offers similar map-reading and editing functions.
Free and Pro users get a one-time 50-credit allotment; Premium users get 500 credits monthly, with tasks costing a few to several dozen credits.
Xmind launched an AI agent on October 29, 2026, that lets users build and revise mind maps through conversation. Alongside it, the company released connections for ChatGPT, Claude and other compatible assistants to create, read and update maps—giving users ways to develop the same visual document from inside or outside Xmind.
The company’s launch press release frames the problem as what happens after an assistant produces its answer. Research, summaries and lists still leave someone to sort the material and decide what to do. Xmind’s approach puts that work into a map that can change as users ask follow-up questions, compare options and refine their thinking.
Xmind Agent can start with an idea, a file or a Notion document, according to the company. It can also work on an existing map, reading what is already there rather than requiring a fresh start. Web research can supply additional material while the conversation continues.
The company says the organization changes with the task. Meeting notes become decisions and to-dos; a comparison lays out each option’s strengths and limitations. Those are the launch’s concrete examples of moving beyond a text summary toward material arranged for a decision.
Users can ask the agent to expand a topic, combine sections, reorder branches or change the styling. Manual editing remains available at any point, so conversation is not the only way to shape the result. Finished maps can then be turned into Notion documents.
We want Xmind to give people and AI a shared visual context to explore ideas, weigh possibilities, and build understanding together.
Mango Du, Xmind co-founder
The external connections take two forms. Xmind MCP uses the Model Context Protocol, an open standard for connecting AI assistants to other software, and supports Claude and compatible assistants. A separate ChatGPT plugin provides similar functions inside ChatGPT. Both connections are available now.
Through either route, an assistant can save a research conversation as an editable map. It can also read goals, requirements and decisions already recorded in a map before drafting a document, planning a project or writing code. New findings can go into the original map or onto another sheet in the same file.
Xmind’s proposed team workflow keeps the reasoning behind a plan in one file that colleagues and AI tools can open, rather than scattered across chat histories. For changes made through MCP, the company says version history preserves earlier map versions so users can restore them.
Xmind Agent is available on the web, desktop and Android. The company lists iOS as coming soon. Chat history syncs across devices, and the agent is offered on every Xmind plan, with work charged against AI credits.
The launch release says a task consumes anywhere from a few to several dozen credits, depending on complexity. The allowances therefore do not translate into a fixed number of tasks: the amount of work users can complete depends on what they ask the agent to do.
Loading discussion...
Join the conversation
Explain when a shared map would serve you better than a chat history.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.