The agent calls Customer.io's Trigger Broadcast API endpoint with the broadcast ID, recipient overrides, and Liquid template personalization data. You can pre-configure which broadcasts the agent is allowed to trigger, or let it dynamically select based on conversation context. The broadcast fires across all channels configured in Customer.io, whether email, SMS, or push.

Yes. The agent uses Customer.io's Get Messages API, filtering by campaign ID, message type (email, SMS, push), and time window. It retrieves metrics like attempted, delivered, opened, clicked, and converted counts. You can ask for specific campaigns or date ranges, and the agent returns structured performance data within the conversation.

Tars needs an App API Key from your Customer.io workspace. This key grants access to the App API endpoints for reading segments, messages, newsletters, and triggering broadcasts. It does not require Track API credentials. You can generate a dedicated key in Customer.io Settings under API Credentials and revoke it anytime.

No. Tars queries Customer.io in real time during conversations. Segment membership, message metrics, and subscriber identifiers are fetched live and used only to respond to the current request. We do not maintain a copy of your Customer.io data. All data in transit is encrypted.

Yes. The agent calls Customer.io's Suppress Person endpoint using the customer's email, ID, or cio_id. Suppression permanently deletes the profile and prevents re-creation. This handles 'right to be forgotten' requests during the conversation itself, so your compliance team does not need to manually process each request in the dashboard.

Customer.io's native triggers are event-based or scheduled. Tars adds a conversational layer where anyone on your team can trigger broadcasts, check metrics, or manage subscribers by simply asking the AI agent. No dashboard navigation, no API knowledge needed. It is like giving your whole team a Customer.io power user on demand.

Customer.io handles empty segments gracefully by simply not sending any messages. The agent reports back that the broadcast was triggered successfully but notes the segment had no qualifying members. No errors, no wasted sends. You can then ask the agent to check segment membership to diagnose why the audience was empty.

Yes. The agent calls the List Transactional Messages endpoint to retrieve all available transactional templates in your workspace. This lets your team verify which transactional messages exist, review their configurations, and ensure the right templates are set up before triggering sends from your application backend.