Skip to main content
Chatflows are single-agent systems, chatbots, and simple LLM workflows in Flowise. They allow you to build conversational AI applications using a visual drag-and-drop interface.

​
What is a Chatflow?

A chatflow is a visual representation of your AI workflow that consists of:
  • Nodes: Building blocks like LLMs, chat models, memory, tools, and chains
  • Connections: Links between nodes that define data flow
  • Configuration: Settings for each node to customize behavior
Chatflows support LangChain, LlamaIndex, and utility nodes for maximum flexibility.

​
Creating Your First Chatflow

1

Navigate to Chatflows

From the main dashboard, click on Chatflows in the sidebar. You’ll see a list of existing chatflows or an empty state if you’re just getting started.
2

Create New Chatflow

Click the Add New button in the top-right corner. This opens a blank canvas where you’ll build your workflow.
3

Add Nodes to Canvas

Click the + button in the top-left corner to open the node library. Browse through categories:
  • Chat Models: OpenAI, Anthropic, Azure, etc.
  • LLMs: Language model providers
  • Chains: Pre-built workflows
  • Memory: Conversation history storage
  • Tools: External integrations
  • Document Loaders: Data ingestion
  • Vector Stores: Semantic search databases
Drag and drop nodes onto the canvas to add them to your workflow.
4

Connect Nodes

Connect nodes by clicking and dragging from an output handle (right side) to an input handle (left side). Valid connections will highlight when compatible.For example:
  • Connect a Chat Model to a Conversation Chain
  • Connect Memory to the chain to enable conversation history
  • Connect Tools to give your chatbot capabilities
5

Configure Node Settings

Double-click any node to open its configuration dialog. Here you can:
  • Set API keys and credentials
  • Configure model parameters (temperature, max tokens)
  • Customize prompts and instructions
  • Set memory options (session IDs, storage)
6

Save Your Chatflow

Click the Save button (floppy disk icon) in the top-right corner. Give your chatflow a descriptive name and click Save.
7

Test Your Chatflow

Click the chat icon in the bottom-right corner to open the test panel. Send messages to interact with your chatflow and verify it works as expected.

​
Basic Chatflow Example

Here’s a simple chatflow structure for a conversational chatbot: 
[Chat OpenAI] → [Conversation Chain]
                      ↑
              [Buffer Memory]
This creates a basic chatbot with:
  • Chat OpenAI: The language model
  • Conversation Chain: Orchestrates the conversation
  • Buffer Memory: Stores conversation history

​
Framework Selection

Flowise supports multiple frameworks. Use the tabs in the node library to switch between:
The default framework with extensive integrations for chat models, tools, chains, and memory systems.

​
Advanced Features

​
Adding RAG Capabilities

Enhance your chatflow with document retrieval:
1

Add Document Loader

Add a Document Loader node (PDF, text, web scraper) to ingest content.
2

Add Text Splitter

Connect a Text Splitter to chunk documents into manageable pieces.
3

Add Embeddings

Add an Embeddings node (OpenAI, HuggingFace) to convert text to vectors.
4

Add Vector Store

Connect to a Vector Store (Pinecone, Chroma, PostgreSQL) to store and search embeddings.
5

Connect to Chain

Use a Retrieval QA Chain or Conversational Retrieval Chain to query your documents.

​
Using Multiple Tools

Add tools to give your chatbot capabilities:
  • Calculator: Perform mathematical calculations
  • Web Browser: Search and browse the web
  • API Tools: Call external APIs
  • Custom Tools: Write your own tool logic
Connect multiple tools to an Agent node to automatically select the right tool for each task.

​
Canvas Controls

  • Zoom: Use mouse wheel or zoom controls
  • Pan: Click and drag the canvas background
  • Fit View: Click the fit view button to center all nodes
  • Minimap: Use the minimap in the bottom-right for navigation

​
Editing

  • Select Node: Click on a node
  • Delete Node: Select and press Delete key
  • Delete Connection: Select edge and press Delete key
  • Copy/Paste: Copy nodes with Ctrl+C, paste with Ctrl+V (flow data as JSON)
  • Undo/Redo: Browser back/forward (note: limited support)

​
Saving and Managing Chatflows

​
Saving

  • Click the Save button to save your chatflow
  • The asterisk (*) next to the name indicates unsaved changes
  • Flowise auto-saves node data as you configure

​
Exporting

From the Settings menu (gear icon):
  • Export Chatflow: Download as JSON file
  • Duplicate Chatflow: Create a copy in a new tab
  • Save as Template: Share your chatflow as a reusable template

​
Importing

From the Settings menu:
  • Load Chatflow: Upload a JSON file to import
  • Paste flow data directly onto the canvas (Ctrl+V)

​
Best Practices

Always save your chatflow before testing changes. Unsaved changes may cause unexpected behavior.
Use descriptive names for nodes by double-clicking the node and editing the label field. This makes complex workflows easier to understand.

​
Organization Tips

  1. Group Related Nodes: Position related nodes close together
  2. Use Sticky Notes: Add notes to document workflow logic
  3. Name Your Nodes: Give each node a descriptive label
  4. Test Incrementally: Build and test in small steps
  5. Version Control: Export chatflows regularly as backups

​
Performance Optimization

  • Use Buffer Window Memory instead of Buffer Memory for long conversations
  • Enable Streaming for better user experience with long responses
  • Use Conversation Summary Memory to reduce token usage
  • Set appropriate Max Tokens limits on models

​
API Integration

Once your chatflow is saved, integrate it into your application:
1

Access API Endpoint

Click the < /> icon in the top-right to view API details.
2

Copy Endpoint URL

Copy the prediction API endpoint for your chatflow.
3

Generate API Key

Create an API key from the API Keys section if required.
4

Integrate

Use the provided code examples (cURL, Python, JavaScript) to integrate.
API endpoint format: 
POST https://your-flowise-instance.com/api/v1/prediction/{chatflowId}
See the API documentation for detailed integration instructions.

​
Troubleshooting

​
Common Issues

Ensure the node types are compatible. Not all outputs can connect to all inputs. Check the node documentation for compatible connections.
  • Verify all required fields are filled in node configurations
  • Check that API keys and credentials are valid
  • Review error messages in the chat panel
  • Ensure all nodes are properly connected
  • Verify sessionId is being passed in API requests
  • Check that memory node is properly connected to the chain
  • Ensure memory storage is configured correctly (Redis, PostgreSQL, etc.)
  • Reduce max tokens settings
  • Use faster models (GPT-3.5 instead of GPT-4)
  • Enable streaming for perceived performance
  • Optimize vector store queries with filters

​
Next Steps

Build docs developers (and LLMs) love

Get started for freeTalk to us