Approval-Safe Customer Replies
Customer communication is a high-trust surface. Rocky can help most by drafting accurate, polite, context-aware replies that a human approves.
When this matters
This page is for operators who want Rocky to produce usable work without turning the system into a mystery box. Use it when you need a practical path, a clear verification step, and a boundary between suggestion and action.
The operating pattern
- Read the source. Use the actual email, order, ticket, or transcript.
- Collect relevant facts. Status, deadlines, prior replies, policies, and customer tone matter.
- Draft plainly. Use short paragraphs and avoid internal vendor or partner details when inappropriate.
- Label uncertainty. If a fact is unknown, ask or escalate instead of inventing.
- Require send approval. Sending remains a human decision until the workflow is proven safe.
Pre-flight checklist
- Draft cites source facts
- No internal/private partner names leak
- Tone matches the situation
- Refunds/payments are not promised without approval
- Human reviews before send
Common failure modes
- Plausible invention: A confident wrong answer is worse than a short escalation.
- Internal leakage: Operational details can undermine trust.
- Auto-send too soon: Start with drafts until accuracy is measured.
Verification
A page is not done because it was drafted. Verify the source, run the workflow, inspect the output, and record what changed. If a step touches money, customers, accounts, permissions, or private data, keep it behind an explicit human approval gate.