Trust narrative

Trust is earned when every action has a gate, a witness, and a fallback.

This is the governance page. It should feel like a review pack for a buyer who wants to inspect the operating model before they let AI touch revenue workflow.

Core questions

What is allowed to run?
Which policy approved it?
Who can inspect the state after send?
What happens when the model is unsure?
Trust-only section

Review the control ledger the same way a security-conscious buyer would.

ActionGateVisibilityFallback

Action

Draft a first-touch follow-up

Gate

Approved proof blocks only for enterprise accounts

Visibility

Operator sees prompt context, draft, and reviewer notes

Fallback

Escalate to AE owner if confidence drops or proof is missing

Action

Switch channel after no reply

Gate

Only if the account allows WhatsApp fallback

Visibility

Timeline records why email paused and why fallback was selected

Fallback

Return to waiting state if fallback policy is blocked

Action

Route a positive reply

Gate

Conversation intent and owner map must be present

Visibility

Inbox state changes to needs reply with named owner and SLA

Fallback

Route to manual triage if intent is ambiguous

Review posture

Buyers should leave the review knowing what runs, what is blocked, and how the team intervenes.

We walk through gates, visibility, and fallback behavior against your actual workflow before asking anyone to trust the automation.