Langflow
Workflow automationTurn visual AI workflows into reusable applications and tools.
Langflow represents an AI application as a flow of configurable components. Builders can connect models, prompts, agents, tools, data stores, and custom Python components in the visual editor, test either a component or the complete flow, and then serve the result through an API or MCP interface.
- Compose AI application flows from configurable visual components
- Build agents and use flows or components as agent tools
- Serve flows through an API or expose them as an MCP server
What people use it for
Prototype an AI application visually
Connect model, prompt, retrieval, input, output, and agent components in the editor, run individual nodes to isolate problems, and use the playground to test the complete application before writing an integration.
Embed a tested flow in a product
Run a Langflow server and trigger a reviewed flow through its API so an application can reuse the same orchestration that was assembled and tested in the visual editor.
Connect agent tools through MCP
Expose a Langflow project as an MCP server, connect external MCP servers as clients, or turn components and complete flows into tools that an agent can call during a larger workflow.