Automated Customer Support Pipeline
Built an agentic workflow that triages, categorizes, and resolves L1 support tickets autonomously using RAG over the client's knowledge base. Human agents now handle only escalations.
Challenge
A B2B SaaS company was drowning in L1 tickets. Average first response time hit 14 hours, CSAT was slipping, and hiring more agents was not scaling with ARR growth.
Approach
We mapped the top ticket categories, built a RAG knowledge base from help docs and resolved tickets, then deployed an n8n + Claude agent stack that classifies, answers, or escalates with full audit trails.
Outcomes
- 62% reduction in tickets reaching human agents
- First response time cut from 14 hours to under 4 minutes for automated resolutions
- CSAT recovered 11 points within one quarter
- Support headcount stayed flat while ARR grew 40%
Stack
RAGn8nClaudeZendesk API
Want results like these?
We can map a similar system to your stack and timeline.
View AI AutomationStart a Project