Visual canvas, not config files
ReactFlow-powered editor with undo/redo, JSON import/export, and a sidebar of pre-built node types.
Weavy is a node-based editor for building AI workflows. Compose text and image inputs, wire them into Gemini or OpenAI, and run them — no backend code, no glue scripts.
How it works
Pick from text, image, and LLM nodes in the sidebar. Drop them anywhere — position is yours to keep.
Wire outputs into inputs. The graph captures the dependency order so each node runs with everything it needs.
Hit Run. Weavy executes the graph in topological order, calling Gemini or OpenAI and streaming results back into each node.
What you get
ReactFlow-powered editor with undo/redo, JSON import/export, and a sidebar of pre-built node types.
Mix Google and OpenAI models in the same graph. Swap providers per node without rewiring.
No free tier, no proxy. Paste a Gemini or OpenAI key once — it lives in your browser and goes straight to the provider.
Persist graphs to your own database via Prisma — your data stays on your infra.
Skip the blank canvas. Start with prebuilt graphs like the Product Listing Generator.
MIT licensed. Fork it, run it locally, or self-host on your own infra.
Templates
Drop a product photo and a one-line description. Get back a structured eBay-style listing with a title, key features, and a persuasive description.
Outline-to-draft pipeline: feed in a topic and reference links, branch into section nodes, merge into a final draft.
Upload a source image and a style reference, then chain through a generation node to produce stylised variants.