FlowGraph vs Mermaid

Mermaid is diagrams-as-code, and it earned its popularity honestly: you write a few lines of text, and a flowchart, sequence diagram, or graph renders. It lives in your Markdown, versions in your repo, and — crucially — large language models emit it by default, so every ChatGPT or Claude answer with a diagram tends to hand you Mermaid. Mermaid Chart adds a commercial editor on top. FlowGraph respects this so much that we built a free tool to import it: paste Mermaid, get a living graph.

The difference is what the output is. Mermaid renders your text into a static picture — accurate, version-controllable, and perfect for documentation. FlowGraph turns that same structure into a living graph: the nodes and edges become real data you can query, link to files and models, extend with AI you verify, and expose to agents through a governed write path. Mermaid is the best way to keep a diagram in your source tree. FlowGraph is for when that diagram needs to become something you interrogate and act on.

What FlowGraph does that rendered text does not

It is interactive data, not an image. A Mermaid diagram is the rendered result of your text — you read it, but you do not query it, link a node to a document, or attach a building element to it. In FlowGraph every node is a real object: filter the graph, follow relationships as live links, pin a model element, and ask questions against the structure.

AI proposes, you verify. With Mermaid, you (or an LLM) write the text and it renders exactly what was written — there is no review loop, because there is nothing to govern. FlowGraph's AI proposes changes as a reviewable patch; you approve or reject, and every accepted change keeps its provenance. That governed loop is what lets AI build the graph with you rather than just print one.

Models and documents. FlowGraph opens IFC models in the browser with no upload and Revit and Navisworks through Autodesk Platform Services, and ties elements to process steps. Mermaid describes diagrams in text; it does not read your model.

Agents can act on it. FlowGraph's MCP server lets an agent read the graph and propose writes through the same reviewed path a human uses. Rendered Mermaid is an output; a governed graph is a place agents can work.

Where the two overlap — and where Mermaid wins

Both are comfortable expressing flowcharts and graphs, and both are local-friendly: Mermaid text lives wherever you keep it, and FlowGraph keeps your vault on your device. If your need is diagrams that live in your repo, render in your docs, and stay diffable in version control, Mermaid is exactly right and FlowGraph is not trying to replace it. In fact, because LLMs speak Mermaid natively, it is often the perfect on-ramp — which is why our importer exists.

The best of both: keep authoring in Mermaid where that fits, and when a diagram needs to become a graph you can interrogate, paste it into FlowGraph's free Mermaid importer and keep going. FlowGraph is free to open and explore with no account, and your vault stays on your device until you decide otherwise.

At a glance