Article

Enhancing Customer Support through AI-powered ticket issue categorization

Razorpay Engineering (Swati Agarwal) · engineering.razorpay.com

An Indian fintech at scale showing how LLM ticket categorization went from 60% to 85% accuracy and feeds product fixes.

Open on engineering.razorpay.com ↗

The question it answers

  1. How do founders use support conversations as product research with AI?

Read with it

  1. Four failure modes that make LLMs hallucinate 'insights' from your ticket data, and how to prevent each one. lennysnewsletter.com Article How to do AI analysis you can actually trust Caitlin Sullivan (Lenny's Newsletter)
  2. 50 real workflows, several of them exactly this: dump support transcripts in, get validated product hypotheses out. lennysnewsletter.com Article Everyone should be using Claude Code more Lenny Rachitsky
  3. A hands-on course from a user-research veteran on turning raw feedback corpora into decisions with Claude Code. maven.com Course Claude Code for Customer Insights Caitlin Sullivan
  4. Notion cut its monthly user-insights report from 2 weeks to 3 days by unifying tickets with every other signal. enterpret.com Article How Notion supercharges its product feedback loop using Enterpret Enterpret
  5. The founder story behind feedback intelligence and why Canva, Figma and Notion all bought it. notoriousplg.ai Article Notorious: How Enterpret Disrupted the Customer Intelligence Space Zachary DeWitt (NotoriousPLG)
  6. The reference tool for this workflow: feedback from every source unified and auto-tagged into roadmap-ready themes. enterpret.com Tool Enterpret Enterpret
eChai Partner Brands The partnership →