Complete guide to Flowise

Getting started

Install Flowise locally with "npm install -g flowise" and "flowise start" (or use Docker, or its managed cloud plan), and access the dashboard at localhost:3000. Go to Chatflows → Add New to open a blank canvas, and build your flow by dragging and connecting nodes: language model, vector database, tools, and conversation logic, without writing code.

Use cases

Advanced setup

1. Chatflow vs Agentflow
   Chatflow is a linear, sequential flow (ideal for Q&A bots); Agentflow lets the model autonomously decide which tool to use at each step.
2. Supported components
   100+ language models, embedding systems, and vector databases (Pinecone, Chroma, Postgres, Weaviate...) available as canvas nodes.
3. Deployment
   It can run locally via npm, in Docker for your own production setup, or you can use the managed cloud plan (from $35/month) if you don't want to maintain your own infrastructure.

Tips

• Start with a simple Q&A Chatflow before moving to Agentflow, which is more powerful but also harder to debug.
• Test several model and vector database combinations locally before deciding which one to take to production — switching providers is as easy as swapping a node.
• Use Docker mode for production instead of the local npm startup, which is intended more for development and testing.
• Document the purpose of each canvas node if the flow grows large — complex diagrams become hard to maintain without that discipline.

FAQ