Hanzo Flow

Hanzo Flow is a visual AI workflow builder (based on Langflow) for designing, prototyping, and deploying LLM-powered applications. Build complex AI pipelines by connecting components visually — from prompts and chains to RAG systems and multi-agent architectures.

Features

Visual Builder: Drag-and-drop canvas for composing LLM pipelines
100+ Components: LLMs, embeddings, vector stores, tools, agents, chains
Multi-Model Support: OpenAI, Anthropic, Google, Ollama, HuggingFace, and more
RAG Pipelines: Build retrieval-augmented generation workflows visually
Agent Systems: Design multi-agent architectures with tool use
API Export: Every flow becomes a REST API endpoint automatically
Multi-Tenant: Isolated workspaces per organization via Hanzo IAM
Python Extensibility: Add custom components in Python

Architecture

┌─────────────────────────────────────────────────────────────────────────┐
│                           HANZO FLOW                                     │
├─────────────────────────────────────────────────────────────────────────┤
│                                                                          │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐  ┌─────────────┐ │
│  │  React UI    │  │  FastAPI     │  │  LangChain   │  │  Component  │ │
│  │  (@xyflow)   │  │  Backend     │  │  Runtime     │  │  Registry   │ │
│  │              │  │              │  │              │  │             │ │
│  │  - Canvas    │  │  - REST API  │  │  - Chains    │  │  - LLMs     │ │
│  │  - Sidebar   │  │  - WebSocket │  │  - Agents    │  │  - Vectors  │ │
│  │  - Chat      │  │  - Auth      │  │  - RAG       │  │  - Tools    │ │
│  │  - Logs      │  │  - Storage   │  │  - Memory    │  │  - Custom   │ │
│  └──────────────┘  └──────────────┘  └──────────────┘  └─────────────┘ │
│                                                                          │
│  ┌──────────────────────────────────────────────────────────────────┐   │
│  │              Provider Integrations                                │   │
│  │  OpenAI · Anthropic · Google · Ollama · HuggingFace · Cohere    │   │
│  │  Pinecone · Chroma · Weaviate · pgvector · Qdrant               │   │
│  └──────────────────────────────────────────────────────────────────┘   │
│                                                                          │
└─────────────────────────────────────────────────────────────────────────┘

Endpoints

Environment	URL
Production	`https://flow.hanzo.ai`
API Base	`https://flow.hanzo.ai/api/v1`

Quick Start

Build an AI Flow

Navigate to flow.hanzo.ai
Sign in with your Hanzo account
Click New Flow or start from a template
Drag components from the sidebar onto the canvas
Connect components by dragging between ports
Click Run to test your flow
Use the API tab to get your flow's REST endpoint

Example: RAG Chatbot

Build a retrieval-augmented chatbot in minutes:

Add components: File Loader → Text Splitter → Embeddings → Vector Store → Retriever → LLM → Chat Output
Configure: Set your OpenAI/Anthropic API key, upload documents
Connect: Wire components together visually
Test: Use the built-in chat interface
Deploy: Get your API endpoint and integrate

API Access

# Run a flow
curl -X POST https://flow.hanzo.ai/api/v1/run/:flow_id \
  -H "Authorization: Bearer $HANZO_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "input_value": "What is Hanzo AI?",
    "output_type": "chat",
    "input_type": "chat"
  }'

# List flows
curl https://flow.hanzo.ai/api/v1/flows \
  -H "Authorization: Bearer $HANZO_TOKEN"

# Get flow details
curl https://flow.hanzo.ai/api/v1/flows/:flow_id \
  -H "Authorization: Bearer $HANZO_TOKEN"

Component Categories

Category	Examples	Description
LLMs	OpenAI, Anthropic, Ollama, HuggingFace	Language model providers
Embeddings	OpenAI, Cohere, HuggingFace, Sentence Transformers	Text embedding models
Vector Stores	Pinecone, Chroma, Weaviate, pgvector, Qdrant	Vector databases
Document Loaders	PDF, CSV, Web, GitHub, Notion, Confluence	Data ingestion
Text Splitters	Recursive, Character, Token, Semantic	Document chunking
Chains	LLM Chain, Sequential, Router, Conversation	LangChain chains
Agents	ReAct, OpenAI Functions, Plan-and-Execute	Autonomous agents
Tools	Search, Calculator, Python REPL, API Call	Agent tools
Memory	Buffer, Summary, Entity, Conversation	Context management
Prompts	Template, Few-shot, Chat, System	Prompt engineering

Multi-Tenant Configuration

Hanzo Flow supports multi-tenant isolation:

Per-org workspaces: Each organization sees only their flows
Shared components: Admins can publish org-wide custom components
API key isolation: Each org manages their own provider API keys
Usage tracking: Per-org token usage and cost monitoring

Environment Variables

# Core
LANGFLOW_HOST=0.0.0.0
LANGFLOW_PORT=7860
LANGFLOW_AUTO_LOGIN=false
LANGFLOW_CONFIG_DIR=/app/langflow

# Database
LANGFLOW_DATABASE_URL=postgresql://hanzo:[email protected]:5432/langflow

# Auth
LANGFLOW_SUPERUSER=admin
LANGFLOW_SUPERUSER_PASSWORD=<from-kms>

# Store
LANGFLOW_STORE=true
LANGFLOW_STORE_ENVIRONMENT_VARIABLES=true

Development

# Install (Python)
pip install langflow

# Or with uv
uv pip install langflow

# Run locally
langflow run --host 0.0.0.0 --port 7860

# Custom components
# Place .py files in LANGFLOW_CONFIG_DIR/components/

Hanzo Auto vs Hanzo Flow

Feature	Hanzo Auto	Hanzo Flow
Purpose	Service-to-service automation	AI/LLM pipeline building
Upstream	ActivePieces	Langflow
Language	TypeScript/Node.js	Python
Focus	Triggers, webhooks, integrations	LLM chains, RAG, agents
Integrations	600+ SaaS connectors	100+ AI/ML components
Use Case	"When X happens, do Y"	"Build an AI that does Z"

Use Auto for connecting business services (Slack + Jira + Salesforce). Use Flow for building AI applications (RAG chatbot, agent pipelines, document processing).

Hanzo Auto — Workflow automation (ActivePieces fork)
Hanzo Gateway — LLM API gateway
Hanzo MCP — Model Context Protocol tools
Langflow Docs — Upstream documentation

Hanzo Flow

On this page