01 — The challenge
Client challenge
BookieLink's support volume exceeded what their small team could manage. Repetitive questions were consuming hours daily, delaying responses to complex cases that genuinely required human attention.
Case study · AI agents
Customer Support RAG Bot
A RAG-powered support bot trained on thousands of BookieLink Q&A records. It handles repetitive customer questions automatically, reduces support tickets, and frees the team for higher-value cases.
Technology stack
- n8n
- OpenAI API
- Vector Database
- Webhooks
02 — What I built
What I built
I built a RAG-powered support chatbot trained on 5,000+ Q&A pairs covering BookieLink's full product surface — deployed as a front-line agent capable of handling the majority of incoming queries autonomously.
- 5,000+ Q&A dataset for RAG training
- Vector-based semantic similarity search
- Automatic escalation for unrecognized queries
- Webhook-based deployment and session management
- OpenAI-powered natural language response layer
Inside the build
01 / 04
03 — The outcome
Measured outcomes
The bot achieves 85% query accuracy and reduced support tickets by 70% in the first month. The team now handles only complex escalations — recovering hours of daily capacity that were previously consumed by repetitive queries.
Next project
Invoice parsing & KPI dashboard
Custom software · Next.js · Supabase · Airtable
Read the case study