Skip to content
RexorAI
Financial Services

An AI support agent that resolves 40% of tickets without a human

Fintra's support volume was outpacing headcount. We built an AI agent that autonomously resolves common tickets and hands off complex cases with full context.

AI AgentsAI ChatbotsWorkflow Automation

40%

Tickets auto-resolved

< 10s

First response time

4.8/5

CSAT

The Challenge

Fintra's support team faced growing ticket volume as the product scaled, with response times slipping and support costs rising faster than revenue.

The Solution

We built an AI agent connected to Fintra's knowledge base, account data, and billing system, capable of resolving account and billing questions end-to-end, with a clear handoff protocol for anything requiring human judgment.

Our process

Step 1

Ticket analysis

We analyzed 6 months of ticket history to identify automatable categories.

Step 2

Knowledge integration

We connected the agent to the knowledge base and account systems securely.

Step 3

Guardrail design

We built escalation rules so sensitive or ambiguous cases route to humans.

Step 4

Rollout

We launched in phases, starting with the highest-volume ticket categories.

Technology used

Claude 3.5Zendesk APIMakeSupabaseNode.js

Results

40%

Auto-resolution rate

Of total ticket volume resolved without human involvement.

Under 10 seconds

First response

Down from an average of 2+ hours.

4.8 / 5

Customer satisfaction

Maintained despite the shift to AI-first support.

The AI agent they built handles nearly 40% of our support volume end-to-end. Response times dropped from hours to seconds.

Marcus Webb

Head of Support, Fintra