← All work
Axioprax Lab
Lab build, fictional business, synthetic data
AI Support Chatbot
A chatbot trained on product documentation. It answers FAQs, hands complex questions to a person and logs every conversation to the CRM with a topic tag.
Problem (test scenario)
The same product questions hit support every day. Complex ones needed a human, but FAQs did not.
What we built
A RAG chatbot trained on docs, with human handoff and topic-tagged conversation logs.
How it works
- 01Customer message received
- 02Intent classified
- 03Knowledge base retrieval
- 04AI reply generated
- 05Reply sent or agent alerted
Tech
Claude APIRAGIntercomNode.js
Result
Pattern for scoping chatbot builds. Labelled as Lab, not a client case study.