Trusted senders

Overview

Make sure your AI Agent always receives and triages email from specific senders by adding them to the Trusted senders section at the bottom of the Config > CHANNELS > Email page.

By default, the Agent’s Automatic reply handling uses AI-powered detection to identify automated messages — such as out-of-office replies, ticket confirmations, and system notifications — and doesn’t reply to them, so it can’t get caught in a loop with another automated system. That detection can also stop messages you want handled, like system mail your workflows are meant to triage to a human. Trusted senders skip the AI-powered detection layer: their messages run your normal Agent flow, so your playbooks and handoffs can process them.

The trusted senders list is global across all email channels on the AI Agent — a trusted sender is trusted on every configured email address.

Use cases

Trust senders whose automated mail your Agent should process instead of ignore.

• Institutional system mail: Route IT helpdesk auto-acknowledgments or ticketing-system notifications into your triage playbook and on to a human team.
• Security and quarantine digests: Make sure spam-quarantine reports and security notifications reach a human instead of being treated as noise.
• Automated partner updates: Process order, shipping, or status notifications from partner systems that your workflows depend on.

Capabilities & configuration

Trusting an email address changes how the AI Agent handles messages from that sender.

Trusted behavior

Emails from trusted senders:

• Are never dismissed by AI-powered auto-reply detection, even when the content looks machine-generated
• Run your normal Agent flow — knowledge, playbooks, and handoffs — so they can be triaged to a human where configured
• Appear in your conversation view and reporting like any other conversation

What trusting does not change

• Header-based auto-reply detection still applies. Messages that declare themselves automatic through standard email headers (Auto-Submitted, X-Autoreply, Precedence) are still dropped and don’t create conversations, exactly as described in Automatic reply handling. Trusting a sender bypasses only the AI-powered detection layer.
• Content moderation still applies. Trusting a sender is not a safety exception — sender addresses can be forged, so messages that trip content moderation are still declined regardless of the sender’s trusted status.
• Loop protection still applies. If a trusted sender’s system gets stuck sending the Agent the same message over and over, or sending messages at machine speed, the Agent stops replying until the burst ends. The messages themselves still appear in the conversation, each with a note explaining that the AI Agent’s response was suppressed — nothing is lost. Normal back-and-forth conversations are never limited.
• An address can’t be both trusted and blocked. Adding an address to one list is rejected while it’s on the other; remove it from the other list first.

Permissions

Any dashboard user can view the trusted senders list. Only Admins and Owners can add or remove trusted senders.

Quick start

Trust an email address in a few steps.

To remove a sender, click the … next to the address in the Trusted senders table and select Remove from trusted, or select multiple addresses and remove them in bulk.

FAQs

Related features

Explore additional Email configuration and management options.

• Block email addresses: Prevent specific senders from receiving Agent replies.
• Email configuration: Activate and configure your Email channel.
• Launch controls: Control how emails are routed during rollout.
• Conversations: View and manage AI Agent conversations.