Smart replies
Smart Replies categorizes prospect replies, drafts responses, escalates what needs a human, and automates follow-up. Test in simulation before enabling auto-reply.
Smart Replies handles what happens after a prospect responds. It categorizes the reply, drafts a response based on your guidelines, escalates anything needing a human, and automates follow-up so conversations keep moving. You set handling per category, and a testing mode lets you simulate conversations and review drafts before enabling auto-reply.
Reply triage is where outbound actually stalls at volume. Plenty of tools help you receive replies; most still require a human to read every one. This is a throughput capability, not a convenience one.
Reply categories
Categorization is based on patterns from thousands of campaigns.
| Category | Typical content | Sensible default handling |
|---|---|---|
| Information requests | "Send me pricing", "What does it integrate with?" | Auto-reply from the knowledge base |
| Product inquiries | Specific capability questions | Auto-reply if covered; escalate if not |
| Meeting requests | "Happy to talk next week" | Auto-reply to book, always notify the rep |
| Objections | "We already use X", "No budget", "Not now" | Escalate early on; automate once drafts prove reliable |
Handling options
For each category, choose one of three:
Respond automatically
Alice answers directly. Use where your knowledge base reliably covers the question.
Skip
No automated action. Use for categories you want handled entirely by humans.
Escalate
Route to a person, with the conversation context attached.
Escalation is only as good as its destination. Decide who receives escalations and confirm they see them — an escalation into an unwatched channel is worse than no escalation, because it looks handled. Wire it to Slack.
Testing before you automate
Run simulation mode
Simulate conversations and review the drafted responses without anything sending.
Test the objections you actually hear
Not generic ones. Take your five most common real objections and check the drafts.
Test a question you know isn't covered
Confirm it escalates rather than being answered confidently and wrongly. This is the most important test on the page.
Enable per category, not all at once
Start with information requests and meeting requests. Add objections last.
Keep sampling after go-live
Read a sample of auto-sent replies weekly for the first month.
The failure mode to design against is a confident answer to a question your knowledge base doesn't cover. A reply that says "let me bring in a colleague" costs you nothing; a wrong answer about pricing or compliance costs you the deal.
Guidelines and boundaries
Smart Replies drafts from the guidelines you provide, and from your knowledge base. The boundaries you set there apply directly here — claims Alice must never make, and topics that always route to a human.
Topics worth making unconditional escalations:
- Pricing beyond your published posture
- Contract, legal, or procurement terms
- Security and compliance specifics
- Roadmap or delivery commitments
- Anything from a named strategic account
- Anything from an existing customer
What reps should do
Reps receive escalations and booked meetings. Their job here is feedback:
- Flag a bad auto-reply. That's a knowledge base or category-handling problem, and it's fixable.
- Flag a missed escalation — something answered automatically that should have come to them.
See Getting started as a User.