Langflow

Langflow offers a canvas-based interface where users drag and drop nodes (LLMs, prompts, retrievers, tools) and connect them to rapidly build complex AI applications like RAG systems, multi-turn chatbots, and intelligent agents. A built-in Playground enables real-time testing, and flows can be exported as JSON for versioning and team collaboration. A Python runtime orchestrates components, data flow, and state management.

It dramatically lowers the barrier to AI development, supporting mainstream LLMs (OpenAI, Claude, Gemini, DeepSeek, etc.) and integrating local models via Ollama for privacy. A rich component library and custom Python nodes allow easy connection to external APIs or private vector stores like Milvus. Finished workflows can be deployed as APIs and integrated into multi-platform bots (WeChat, QQ, Slack). Visual debugging and step-by-step execution enable fast iteration.

Despite low-code, understanding LLM parameters and component logic is still required; installation can be tricky with dependency issues, and the open-source version has weak multi-user access control (e.g., public space risks).