AI customer-support assistant for a commerce marketplace
We built a RAG assistant over Quantia’s knowledge base, deflecting 60% of incoming tickets and cutting average response time from 8 hours to 2 minutes, with automatic human handoff for complex cases.
Scope
- Quality evals & guardrails
- RAG pipeline
- Helpdesk integration
- Production monitoring
Outcomes
Cut human-handled tickets by 60%
Avg response time: 8h → 2m
94% CSAT
To production in 4 weeks
Gallery
Tech stack
PythonLangChainPineconeGPT-4oFastAPIRedis
Our role
AI Engineer, ML Ops, Product Designer
Read the full case study
AI assistant deflects 60% of support tickets for Quantia
