How to turn on AI Auto-Responders in Agent Hub, choose draft-for-review or send-automatically, decide what to respond to by message type, ground replies in your knowledge bases and customer records, give the AI tools to act on demand, and see what gets routed to a human instead.
AI Auto-Responders read incoming emails and texts, sort each one into a message type, and draft a reply for you, grounded in your business's own knowledge bases. You decide, per phone number and per email address, whether replies wait in your approval queue for a human to check first or send straight away, and which message types the AI should handle at all.
Find every Auto-Responder at https://app.trustpager.com/ai-agents/agent-hub.
The Auto-Responders section lists every phone number and every workspace email address you have, whether an auto-responder is set up on it yet or not. Each row shows the number or address, and its current status. Click any row to open that Auto-Responder's detail page and configure it.
Each Auto-Responder detail page has an on/off switch at the top right. Nothing drafts or sends until you switch it on.
Choose how confident you want to be before a reply goes out:
Review, edit, and approve queued drafts at https://app.trustpager.com/settings/api?tab=approvals.
Every incoming email or text is automatically sorted into one of ten types. For each type, you choose what the AI does with it:
The ten types, with their defaults out of the box:
These defaults mean your Auto-Responder won't draft replies to cold sales pitches or automated no-reply mail, and complaints or billing questions always get a person's eyes before anything goes out. Change any of the ten to whatever suits your business on the Auto-Responder's detail page, at https://app.trustpager.com/ai-agents/agent-hub.
Tick which knowledge bases the AI can draw on when it writes a reply. The more relevant material you attach, the more accurate and specific its answers will be. Create and manage knowledge bases at https://app.trustpager.com/ai-agents/knowledgebase, then come back and tick the ones that apply to this number or address.
Tick which of the customer's own records the AI can use as background when it drafts a reply: Opportunity, Contact, Customer account, Products, Work orders, Bookings, Invoices. The AI works out who it's talking to from their email address or phone number, then only pulls in the ticked records if the customer's message is actually about them. It never volunteers this information unprompted.
Tick which actions the AI can take mid-reply, on top of just drafting text, when a customer's message calls for it.
Read-only, no approval needed: Look up account, Work order status, Check pricing, Check availability.
Write actions, each with its own "Requires approval" toggle: Book appointment, Cancel booking, Reschedule booking, Request callback, Take a message, Unsubscribe, Create payment link. Switch it on and the action queues at https://app.trustpager.com/settings/api?tab=approvals for a person to approve before it happens, the same place drafted replies wait. Switch it off and it happens immediately. Customer-visible actions (bookings, unsubscribe, payment links) default to requiring approval; internal ones (take a message, request a callback) default to happening immediately.
Use Tone for a short style note, for example "warm and concise" or "formal". Use Instructions for anything business-specific the AI should always follow when replying: products and pricing to mention, policies to stick to, and anything it should never promise. Both are optional, but the more specific your instructions, the closer the drafts will match how you'd actually reply.
Turn this on and the Auto-Responder only drafts or sends outside your set business hours. Outside those hours, the message simply sits in your inbox, waiting for a person to reply as normal, the same as if the Auto-Responder was off.
The "Bound to" card shows the incoming number or address this Auto-Responder is listening on. That's fixed, since it's what defines the Auto-Responder. For email, you also get a Reply from selector: choose which of your workspace's sending addresses the reply actually goes out from. The AI signs off as whichever address you pick, so a shared inbox like info@ can still have replies go out under the right name.
Each Auto-Responder's detail page shows four running totals over the last 30 days: replies drafted, sent to human, total messages handled, and failed. If any drafts are currently waiting in your approvals queue, a banner on the page tells you how many, with a direct link through to review them.
Talk to Evie right now, or book a real human for a deeper walkthrough. Whichever feels right.
Hear her in 10 seconds.

60 minutes with a real human · Google Meet · Free.