Use casesOutbound email guardrails
02

Outbound email guardrails

Rewrite risky emails before they send.

Scope a content policy to email so safe drafts pass unchanged and risky promises are replaced with policy-approved language before delivery.

Trigger

An agent proposes a customer-facing email through the SDK, gateway, support workflow, CRM, or another application integration.

What breaks

A draft guarantees a refund, exposes sensitive information, violates policy, or uses wording the business cannot stand behind.

Control point

Evaluate the proposed message against the email-scoped content policy, then return permit or a safe rewrite before the customer mailer sends anything.
01Proposed email
02Email policy
03Permit or rewrite
04Customer mailer

Live flow

Follow the action through the decision.

The walkthrough below is native page UI: proposal, policy facts, decision, and the point where the customer-owned executor continues or stops.

01

Proposed action

The agent proposes a customer email

This is a guaranteed refund.
Operation
send_email
Channel
email
02

Policy check

The email policy checks the draft

Match
guaranteed refund
Action
transform
Scope
support-agent
Policy on
03

Decision

Return an explicit effect

Safe draftPermitSend unchanged
Risky promiseTransformUse safe replacement
04

Execution

The customer mailer applies the result

The application sends the original permitted draft or the policy-approved replacement.

Execution boundaryFeatherlane AI evaluates the proposed message and never sends the email.

How it works

  1. 01
    Choose

    Scope a policy to outbound email

    Set the email channel, the risky wording or semantic match, the desired action, and a policy-approved replacement when the action is transform.

  2. 02
    Publish

    Manage one policy through the registry

    Validate and publish the YAML, then confirm it is enabled for the intended environment and agent before customer-facing traffic reaches it.

  3. 03
    Propose

    Check the message before delivery

    Submit the proposed output with context.channel set to email. A safe draft returns permit; risky wording returns transform with the configured rewrite.

  4. 04
    Apply

    Send only the policy-safe result

    The customer application keeps ownership of delivery and applies the returned decision before calling Gmail, Outlook, a support platform, or another provider.