langflow

Overview

Langflow is a platform for building and deploying AI agents and workflows. You design flows in a visual editor and can export the resulting workflows as an API or an MCP server to plug into apps built with any framework.

It supports major LLMs, vector databases and AI tools as components, and each component can be modified in Python. You can test flows step by step in the Playground, use multi-agent setups and retrieval, and connect to observability tools such as LangSmith and LangFuse.

It suits developers who want to quickly prototype RAG apps or agents without writing much code. Automatic login is turned off in the official distribution packages, so you must set an admin password with LANGFLOW_SUPERUSER_PASSWORD before starting.

What it is used for

• Build RAG application flows with drag and drop.
• Deploy a flow as an API or export it as JSON to use in a Python app.
• Deploy a flow as an MCP server and use it as a tool in MCP clients.
• Combine multiple agents to handle conversation management and retrieval together.

Pricing and license

MIT; a free cloud account is also available

Getting started

1. Set up Python 3.10–3.14 and uv, or download Langflow Desktop for Windows or macOS.
2. In a new folder, install with uv pip install langflow -U.
3. Set LANGFLOW_SUPERUSER_PASSWORD, then run uv run langflow run.
4. Go to http://127.0.0.1:7860, log in with the username langflow and build a flow.

Links

Official site ↗Repository ↗

Related terms and companies

Agentic workflow RAG MCP Multi-agent Vector DB LLM observability

Sources

• langflow-ai/langflow README
• Langflow | Low-code AI builder for agentic and RAG applications

Compiled from public sources (official pages, READMEs). Pricing and features change, so check the official site.