Langflow
Drag-and-drop agent building, genuinely MIT
What it is
A visual canvas for composing LLMs, vector stores, tools, and APIs into agents and workflows, each servable as an API. Python-extensible, with MCP support on both the client and server side.
Why it's interesting
The leading visual agent builder that is actually MIT-licensed — a meaningful contrast with Dify and n8n, whose licenses carry conditions. Acquired into DataStax/IBM and still permissive.
Use cases
- Rapid agent and RAG prototyping
- Low-code internal AI tools
- Teaching how LLM pipelines fit together
Who it's for
Mixed technical and semi-technical teams; solution engineers
Setup
Easy. pip install or Docker; a desktop app is available; provider API keys
Limitations & cautions
A history of serious CVEs (including a 2025 pre-auth RCE) means you patch promptly and never expose an unauthenticated instance to the internet. Big flows get unwieldy.
Editorial takeaway
The visual builder to reach for when the license fine print on the alternatives gives you pause.