SupportHub —
80% of Tickets
Resolved by AI
How we built a custom AI support assistant that slashed ticket volume, cut costs by €50K/year and made customers happier in the process.
The Problem
SupportHub had grown from 500 to 3,200 customers in 18 months. Their support team of 4 hadn't scaled with it. Average response time had climbed to 18 hours. 60% of incoming tickets were the same 12 questions — password resets, billing queries, how-to guides. Smart people doing repetitive work.
"Our best support agents were spending 6 hours a day answering 'how do I connect Slack?' We were burning out good people on questions the product docs already answered." — Priya Sharma, Head of CX
What We Built
We designed a custom AI support assistant powered by Claude and embedded it directly into SupportHub's help centre and in-app widget. The bot was trained on their entire knowledge base, 6,000 past tickets and 3 years of product changelogs — giving it deep, contextual understanding of their specific product.
Key features included smart human handoff (the AI detects frustration or complexity and passes to a human with full context), multi-channel deployment across website, in-app and email, a GDPR-compliant architecture, and a custom analytics dashboard showing resolution rates and top unresolved queries.
The Outcome
Within 30 days of go-live, ticket volume handled by human agents dropped by 68%. By month 3 it was at 80%. Response time went from 18 hours to under 2 minutes for AI-handled queries. Customer satisfaction scores improved from 4.2 to 4.8 stars. Annual savings from avoided hiring: €50,000.
"I expected the chatbot to just handle easy stuff. But it handles almost everything — and better than some junior agents would." — Priya Sharma, Head of CX