🚀 Connect to 2,700+ apps - Integrate with thousands of services like Google Sheets, HubSpot, Slack, and more without writing code.
Flows are visual conversation builders for WhatsApp that bring the best of both worlds: AI-powered decision making and content generation, while preserving determinism where you need it. Think of them as smart interactive forms that can adapt based on context while following predictable paths. Use flows when you need:
  • Structured data collection (lead forms, surveys)
  • Predictable conversation paths with AI-enhanced flexibility
  • Menu-based navigation with intelligent routing
  • Integration with external systems
  • Dynamic content generation based on conversation context

​
Core concepts

Flow canvas: Where you build by connecting nodes Nodes: Actions in your conversation (send message, wait for response, make decisions) Edges: Lines connecting nodes that define the path Triggers: What starts a flow (WhatsApp messages, API calls) AI Integration: Use AI to make intelligent routing decisions and generate dynamic content while maintaining control over the conversation flow

​
Flow triggers

Flows can be triggered by:
  • WhatsApp messages: Automatically start when users message your WhatsApp number
  • API calls: Trigger flows programmatically via REST API
For API trigger documentation, see the API Reference

​
Node types reference

​
Start node 🟢

  • Entry point for every flow
  • Cannot be edited or deleted
  • Shows active triggers

​
Send text 💬

  • Sends plain text messages
  • AI generation available

​
Send template 📄

  • Uses pre-approved WhatsApp templates
  • Required for notifications outside 24-hour window
  • Parameters support AI generation

​
Send interactive 🔘

  • Reply buttons: Up to 3 quick actions
  • List message: Menu with up to 10 sections
  • Call-to-action URL: External link button
  • WhatsApp flow: Native WhatsApp forms
  • Product catalog: Coming soon

​
Wait for response ⏰

  • Pauses until user responds
  • Captures any message type

​
Decision 🔀

  • AI evaluates user input and conversation context
  • Routes to different paths based on conditions
  • Each condition needs a label and description
  • Configuration:
    • AI model: Select which LLM to use
    • Temperature: 0.0 (deterministic) to 2.0 (creative)
    • Max tokens: Token limit for decision (default: 10000)

​
Webhook 🌐

  • HTTP requests to external APIs
  • Methods: GET, POST, PUT, PATCH, DELETE

​
Apps ⚡

  • Connect to 2,700+ services
  • No code required
  • Authenticated connections

​
Creating your first flow

  1. Access flows: Navigate to Flows section
  2. Create new: Click “New flow” button
  3. Name your flow: Be descriptive
  4. Add nodes: Double-click canvas or drag from toolbar
  5. Connect nodes: Drag from output to input handles
  6. Configure each node: Double-click to edit
  7. Test: Use test mode before publishing
  8. Publish: Make it live

​
Quick example: Lead capture

​
AI-powered features

Flows include two powerful AI features:

​
1. Decision node: AI-powered routing

The Decision node uses AI to analyze the conversation and route to the appropriate path: How it works:
  1. Define conditions with labels and descriptions
  2. AI evaluates the full conversation context
  3. Selects the best matching condition
  4. Flow continues along that path
Example conditions:
  • Label: “interested” | Description: “User shows interest in the product”
  • Label: “not_interested” | Description: “User declines or shows no interest”
  • Label: “needs_info” | Description: “User asks questions or needs more details”
Configuration tips:
  • Use clear, distinct condition descriptions
  • Lower temperature (0.0-0.5) for consistent routing
  • Higher temperature (1.0-2.0) for nuanced interpretation

​
2. AI field generation: Dynamic content

Look for the ✨ icon to switch any supported field to AI mode. How to use:
  1. Click the sparkle icon on a field
  2. Enter a prompt describing what to generate
  3. AI generates content during flow execution
Example prompts:
  • “Summarize the user’s main concern”
  • “Extract the email address from the conversation”
  • “Create a friendly response acknowledging their interest in the product”
  • “Generate the webhook URL based on the user’s location”
Available in:
  • Send text: Message content
  • Send template: Parameter values
  • Send interactive: Body, header, footer text
  • Webhook: URLs, headers, request body
  • Apps: Configuration values
Context available to AI:
  • Full conversation history
  • Previous node outputs
  • System information

​
App connections

Connect your flows to external services without coding. Search for available actions across thousands of apps.

​
Setting up connections

  1. Add Apps node to your flow
  2. Search for desired action (e.g., “create spreadsheet row”)
  3. Connect your account when prompted
  4. Configure action parameters
  • CRM: HubSpot, Salesforce, Pipedrive
  • Sheets: Google Sheets, Airtable
  • Communication: Slack, Email, SMS
  • Tasks: Trello, Asana, Notion
  • Payments: Stripe, PayPal

​
Example: Save lead to Google Sheets

  1. Apps node → Search “Google Sheets”
  2. Select “Add Row to Sheet”
  3. Connect Google account
  4. Configure fields with customer data

​
Testing and debugging

​
Test mode features

  • Initial message: Set custom trigger
  • Mock messages: Preview WhatsApp conversation
  • Flow events: Track execution steps

​
Debug process

  1. Enter test mode
  2. Send initial message
  3. Follow conversation flow
  4. Check events tab for errors
  5. Inspect variables at each step

​
Coming soon

  • Connect with agents