What Is Flowise?
Flowise is an open-source tool that brings LangChain to a visual drag-and-drop interface. You build LLM pipelines — RAG systems, conversational agents, document QA bots — by connecting nodes on a canvas instead of writing Python or JavaScript.
The output is a live API endpoint and embeddable chat widget, ready to integrate into any application.
Core Interface
The canvas has nodes for every major LangChain component:
- LLMs — OpenAI, Anthropic, Bedrock, Ollama (local), Groq
- Memory — Buffer Memory, Redis Memory, Zep
- Retrievers — Pinecone, Chroma, Weaviate, Qdrant, FAISS
- Document Loaders — PDF, URL, Notion, GitHub, CSV
- Tools — web search (SerpAPI, Tavily), code interpreter, custom function calls
- Chains — Conversational Retrieval QA, SQL Database, OpenAPI
You drag nodes onto the canvas, connect their inputs and outputs, configure credentials in the node settings, and click Save.
Pre-Built Templates
Flowise ships with templates for common patterns:
- Conversational RAG — upload PDFs, ask questions with chat history
- SQL Agent — connect to a database, query it with natural language
- Document QA — multi-document Q&A with source citations
- Web Scraping Agent — browse URLs and extract structured data
Templates give you a working flow in under two minutes that you can then customize.
Document Upsert API
For dynamic document ingestion, Flowise exposes an Upsert API:
curl -X POST http://localhost:3000/api/v1/vector/upsert/{chatflowId} -H "Content-Type: multipart/form-data" -F "files=@document.pdf" -F "pineconeIndex=my-index"
This lets you build workflows where new documents are automatically indexed when uploaded.
Calling Your Flow as an API
Every saved flow gets an API endpoint:
const response = await fetch("http://localhost:3000/api/v1/prediction/YOUR_FLOW_ID", {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({ question: "What is the refund policy?" }),
});
const { text } = await response.json();
Embedded Chat Widget
Flowise generates a script tag you paste into any webpage to embed a chat widget that connects to your flow:
<script type="module">
import Chatbot from "https://cdn.jsdelivr.net/npm/flowise-embed/dist/web.js"
Chatbot.init({
chatflowid: "YOUR_FLOW_ID",
apiHost: "http://localhost:3000",
})
</script>
Self-Hosting with Docker
docker run -d --name flowise -p 3000:3000 -v ~/.flowise:/root/.flowise flowiseai/flowise
Flowise vs Dify vs Langflow
Flowise is the simplest of the three — closest to "LangChain with a GUI." Dify has better built-in RAG pipeline management and more sophisticated dataset handling. Langflow is the most technically similar to Flowise but has a different node library. For teams already familiar with LangChain who want a visual interface, Flowise has the most direct mapping between the canvas and LangChain concepts.
