AI-driven customer support triage
Client disclosure: EU SaaS company, 200-500 FTE (k>=5 cohort)
- first-response time
- 8h12min
- tickets auto-resolved
- 0%38%
Use the live console to inspect this record, compare it with your workflow, or start a fit check.
A mid-sized European SaaS company asked us to design an LLM-powered support triage system that could handle their growing volume of tier-1 inquiries.
Situation
Their support team was overwhelmed: 8h average first-response, no auto-resolution, attrition driven by ticket-volume burnout.
Approach
We deployed a Claude Haiku-based triage agent with retrieval over their internal knowledge base, integrated with their existing Zendesk instance.
Outcome
- First-response time: 8h -> 12min
- Tickets auto-resolved: 0% -> 38%
- Support team NPS: +24 points