Refund Eligibility Check Agent · Use case
How can an airline or a hotel group use an AI agent to check whether a booking is eligible for a refund on its own infrastructure?
A refund eligibility check agent reads one refund request and its booking, sets them against the fare or rate rules the booking was sold under and the refund rules the company has indexed, says which condition the request meets or fails, with the rule, and prepares the case. The refunds team decides and pays. The agent issues no refund, changes and cancels no booking, reads no card number and sends nothing to the customer.
Every fare and rate has its own rules.
Each request is read against them by hand.
- ✗ 01A refund depends on the fare or rate the booking was sold under, what has already been used, when the request was made and why, and the rules differ from one booking to the next.
- ✗ 02Requests queue after every schedule change and every cancellation, and a request wrongly refused comes back as a complaint.
Where it stops for a person
Always. The refunds team decides and pays.Always. It checks a request against the rules, it does not decide it: eligibility, the amount, any goodwill and the reply are the refunds team's decisions, and the refund is made by a person in your own system. A request the rules do not cover, or one where the booking and the rules disagree, is marked for a supervisor.
It prepares the case.
The refunds team decides and pays.
Step · 01
A refunds agent opens a request, or names a booking.
Step · 02
The agent reads the request and the booking through your own APIs: the fare or rate, what was used, the dates and the reason given.
Step · 03
It finds the rules that booking was sold under and the refund rules that apply to it.
Step · 04
It says which condition the request meets or fails, citing the rule, and lists what is missing or unclear.
Step · 05
The refunds team decides, and makes any refund in your own system.
Reads from · connectors
Your own APIs
The refund request and the booking, read through your own APIs with the operations that change something left out, and the fare and rate rules and the refund rules that apply, as your legal team has set them out, indexed as a knowledge source. No card numbers: payment data stays in the payment system.
Acts through · MCP servers
Nothing. It issues no refund, creates no credit or voucher, changes or cancels no booking and sends nothing to the customer.
What you set up
- Index the fare and rate rules, and the refund rules that apply to you as your legal team has set them out, as a knowledge source.
- Import the description of the API that holds bookings and refund requests, and keep only the operations that read.
- Build the agent in Katonic Studio and test it on requests your refunds team has already decided.
- Publish it to the refunds team, one fare family or one property first.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service. Bookings and refund requests are personal data and stay inside the company.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Does it issue the refund?
No. It has no tool that pays or credits anything. The refunds team decides and makes the refund in your own system.
Does it read card details?
No. It reads the booking and the request. Card numbers stay in the payment system and are not among the operations you import.
Does it tell the customer the outcome?
No. It has no tool that sends anything. The refunds team writes to the customer.