Support
Support resolves customer problems and is the company's early-warning system for product issues.
Key processes
- Ticket intake and triage
- Resolution
- Escalation
- Knowledge base management
- Feedback loop to Product
KPIs that matter
- CSAT
- Customer satisfaction after an interaction.
- First contact resolution
- Tickets solved without escalation.
What AI agents can do here
- Chatbots for common questions
- Ticket classification and routing
- Suggested replies
- Agents resolving most tier-1 tickets end to end
- Voice agents
- Agents handling complex cases with access to systems, humans for empathy-heavy cases
What good looks like
- Build and maintain a knowledge base; answer from it, not from memory.
- Measure first-contact resolution and CSAT, not just ticket counts.
- Send the top recurring issues to Product every month with numbers.
Common pitfalls
- Closing tickets without solving them
- Agents inventing policies or refunds
- Treating Support as a cost center instead of an early-warning system
Agent tasks and the human control they need
| Task | Maturity | Human control |
|---|---|---|
| Answer common questions and suggest replies from approved content | Today | Answers grounded in the knowledge base; samples reviewed weekly. |
| Resolve most tier-1 tickets end to end, including simple account changes | Emerging | Clear escalation rules and audit logs of every action. |
| Handle complex cases with access to internal systems | Future (3–5 years) | Humans for empathy-heavy, high-value or legal cases. |
Today: in common production use. Emerging: piloted by early adopters. Future: plausible within 3–5 years, not yet dependable.
End-to-end processes this department is part of
Related processes
Incident management
Detect → Triage severity → Contain → Resolve → Communicate → Blameless post-mortem
Mature teams measure mean time to recovery (MTTR), not just how rarely things fail.
Issue-to-resolution
Ticket intake → Classify → Resolve or escalate → Confirm with customer → Feed back to Product
Grounding support agents in an approved knowledge base is today's main defence against hallucinated policies.
Customer onboarding
Kick-off → Configure → Migrate data → Train users → First value milestone → Handover to success
Time to first value is the strongest early predictor of churn.