Skip to main content

​
What are Chatflows?

Chatflows are visual workflows that allow you to build conversational AI applications using a drag-and-drop interface. They represent the traditional approach to building AI applications in Flowise, where you connect nodes in a sequential manner to create chat-based experiences.
Chatflows are ideal for building chatbots, Q&A systems, document retrieval applications, and other conversational interfaces that follow a predictable flow.

​
Chatflow Types

Flowise supports multiple chatflow types, each designed for specific use cases:

CHATFLOW

Standard conversational flows with chains and memory

MULTIAGENT

Coordinate multiple AI agents working together

ASSISTANT

OpenAI Assistant API-powered chatflows

​
Core Components

A chatflow is composed of several key elements:

​
Flow Data Structure

Each chatflow contains:
  • Nodes: Individual components that perform specific tasks
  • Edges: Connections between nodes that define the flow of data
  • Viewport: The visual canvas configuration (position, zoom)
interface IChatFlow {
  id: string
  name: string
  flowData: string // Serialized JSON of nodes and edges
  type: 'CHATFLOW' | 'MULTIAGENT' | 'ASSISTANT' | 'AGENTFLOW'
  deployed: boolean
  isPublic: boolean
  workspaceId: string
}

​
Chatflow Properties

Chatflows are stored as database entities with the following properties:
  • Name: Human-readable identifier for your chatflow
  • Flow Data: JSON representation of the visual canvas (nodes, edges, connections)
  • Deployed: Whether the chatflow is active and accessible via API
  • Public: Controls public accessibility without authentication
  • API Config: Configuration for API endpoints and responses
  • Chatbot Config: Settings for embedded chat widgets

​
Configuration Options

Chatflows support extensive configuration:

​
Analytics

Track and monitor chatflow performance:
  • Message analytics
  • User interactions
  • Response times
  • Token usage

​
Speech Integration

Enhance your chatflows with voice capabilities:
  • Speech-to-Text: Convert user voice input to text
  • Text-to-Speech: Generate voice responses from AI output

​
Follow-Up Prompts

Guide users with suggested questions and next steps after each response.

​
Categories

Organize chatflows by use case or domain (e.g., “Customer Support”, “Sales”, “Internal Tools”).

​
Building a Chatflow

1

Start with a Model

Add a Chat Model node (e.g., ChatOpenAI, ChatAnthropic) as the foundation
2

Add Components

Connect supporting nodes like:
  • Document Loaders for knowledge bases
  • Memory for conversation context
  • Embeddings for semantic search
  • Vector Stores for retrieval
3

Configure Chains

Use Chain nodes to orchestrate the flow (e.g., Conversational Retrieval Chain)
4

Test & Deploy

Test your chatflow in the interface, then deploy it to make it accessible via API

​
Chatflow vs Agentflow

Choosing between Chatflows and Agentflows depends on your use case:
FeatureChatflowAgentflow
StructureSequential, predictableDynamic, autonomous
Decision MakingPre-defined pathsAI-driven routing
Best ForQ&A, RAG, simple botsComplex workflows, multi-step reasoning
ExecutionFollows fixed chainAdapts based on context

​
Deployment

Once your chatflow is ready:
  1. Test Locally: Use the built-in chat interface to validate responses
  2. Deploy: Toggle the deployment switch to activate the API endpoint
  3. Integrate: Use the generated API endpoint in your applications
  4. Embed: Add the chatbot widget to your website
Deployed chatflows are accessible via:
  • REST API endpoints at /api/v1/prediction/{chatflowId}
  • Embedded chat widgets
  • Webhook integrations

​
Example Use Cases

Customer Support Bot
  • Document Loader → Vector Store → Conversational Retrieval Chain
  • Includes chat memory for context retention
Document Q&A
  • PDF Loader → OpenAI Embeddings → Pinecone → RetrievalQA Chain
  • Answers questions from uploaded documents
Conversational Assistant
  • Chat Model + Conversation Memory + Tools
  • Maintains context across multiple interactions

​
Best Practices

  • Keep it Simple: Start with basic flows and add complexity incrementally
  • Test Thoroughly: Validate edge cases before deployment
  • Monitor Performance: Use analytics to track token usage and response quality
  • Version Control: Save different versions as you iterate on designs
  • Agentflows - For more dynamic, autonomous workflows
  • Nodes - Building blocks of chatflows
  • Credentials - Managing API keys and authentication

Build docs developers (and LLMs) love

Get started for freeTalk to us