Automations
Background flows that run automatically at lifecycle events, not triggered by user input.
Automations are a special category of flows that run automatically in the background at defined lifecycle points – without generating responses visible to the user. They are designed for tasks that should happen at the end of a conversation or at other system events: generating a summary, creating a conversation title, logging data, or calling an external service.
How Automations Differ from Conversational Flows
| Conversational Flows | Automations | |
|---|---|---|
| Triggered by | User input (via Handlers) or other flows | System lifecycle events automatically |
| Visible to user? | Yes – generate responses the user sees | No – run silently in the background |
| User interaction | Can collect input, show cards, etc. | Cannot interact with the user |
| Typical use | Conversation logic | Summarization, logging, webhooks |
The Default Automation: End Conversation
Every new pipeline includes one Automation: End Conversation.
This automation fires automatically when a conversation session ends. Its typical structure:
Entrypoint (On conversation-end)
│
├── Create title ──► [LLM generates a title for the conversation]
│
└── Create summary ──► [LLM generates a summary]
│
└── End conversationThe Create title and Create summary steps use AI to label and summarize the session for the Conversations tab in the dashboard. These are useful for reviewing conversation history and understanding what users asked about.
Adding a New Automation
- Scroll to the bottom of the left sidebar
- Click + Add flow
- Enter the name of your new flow
- Select the flow type Automation
- The new automation appears in the sidebar
Automations start with an Entrypoint node that describes when the automation triggers.
Triggering an Automation Manually
You can fire an Automation from within a Conversational Flow or Handler using the Trigger Automation node. This lets you run background tasks at any point in the conversation – not just at session end.
For example, you might trigger an automation to log a specific user action (like completing a form or reaching a key step) without interrupting the flow of conversation.
Available Nodes in Automations
Automations can use most node types, in particular:
- LLM Response – For AI-generated content (titles, summaries)
- Call Webhook – For external notifications
- Store Variable / Check Variable – For reading or writing variables
- Extract variables – For extracting variables from the conversation context
- End conversation – Explicitly ends the session (use at the end of the End Conversation automation)
Automations run asynchronously. The conversation session ends from the user's perspective when the End Conversation node fires, but the Automation itself may continue running briefly in the background to complete its tasks.
Use Cases
- End-of-session summary – Generate a human-readable summary of the conversation stored in the dashboard
- Conversation titling – Automatically create a short title based on the conversation topic
- CRM logging – At session end, POST the conversation transcript or key variables to your CRM or database
- Follow-up trigger – After a user completes a key action, trigger a follow-up process in an external system
- Analytics events – Send custom analytics events to your analytics platform