flowise-builder

Overview

Flowise is an open-source visual tool for building LLM workflows. It provides a drag-drop interface to connect LLMs, document loaders, vector stores, tools, and chains — then deploy as API endpoints.

Capabilities

• Build chatflows visually with drag-drop nodes
• Connect to OpenAI, Anthropic, Ollama, and local models
• Add document loaders (PDF, web, CSV, Notion)
• Integrate vector stores (Pinecone, FAISS, Chroma, Qdrant)
• Add tools (web search, calculator, API calls)
• Deploy as REST API with streaming support
• Embed chatbot widget in websites

When to Use

Trigger phrases:

• "flowise builder"

• "Flowise visual LLM workflow builder — drag-drop chatflows, API endpoints, docume"

• Building LLM apps without writing code

• Prototyping RAG chatbots quickly

• Needing visual workflow design for AI pipelines

• Deploying AI chatbots as APIs or website widgets

• Self-hosting AI infrastructure

When NOT to Use

• Task requires custom AI model training (use ML tools)
• You need complex AI agent logic (use LangChain directly)
• Task is about data processing, not AI app building
• You don't have Flowise instance running
• Task requires real-time AI inference (use dedicated AI services)
• You need to build a custom AI application (use development tools)

Pseudo Code

Implementation patterns for common use cases with this skill.

Installation

# npm
npm install -g flowise
npx flowise start

# Docker
docker run -d -p 3000:3000 flowiseai/flowise

# Access at http://localhost:3000

Chatflow Architecture

Document Loader → Text Splitter → Embedding → Vector Store
                                                    ↓
User Question → Embedding → Vector Store Retriever → LLM Chain → Response

API Usage

# Prediction
curl -X POST http://localhost:3000/api/v1/prediction/{chatflow-id} \
  -H "Content-Type: application/json" \
  -d '{"question": "What is the return policy?", "overrideConfig": {}}'

# Streaming
curl -X POST http://localhost:3000/api/v1/prediction/{chatflow-id} \
  -H "Content-Type: application/json" \
  -d '{"question": "Hello", "streaming": true}'