Example: a support inbox
A customer writes to an address assigned to your support agent. The incoming message reaches SenderPermit, where the agent can draft a reply using its instructions. Your policy evaluates the proposed send. A matching rule may allow it, block it, or require a person to approve it. If no rule matches, approval is required.
- Give the agent an identity. Assign an email address and instructions in your workspace.
- Receive and draft. Configure inbound email, then enable a separate AI drafting allowance for the workspace.
- Apply a policy. Choose which agent replies may be sent, held for approval, or blocked.
- Review the outcome. Inspect the proposed reply before approving or rejecting it, and follow the decision in the audit trail.
What the controls do
Policies govern whether an agent reply can be sent. Human review lets someone inspect a held reply before release. The audit trail records the resulting activity and decisions. These controls do not verify that an AI-generated reply is factually correct, so use approval rules for sensitive or high-impact messages.
How to start
First, verify a domain and configure receiving. Then follow the agent approval guide to create an identity and policy. You can try the AI workflow before connecting a real inbox.
New workspaces include zero AI draft attempts. AI drafting requires an allowance enabled separately from the email plan. See pricing and allowances, or contact support to request access.