Prepare a flight-delay compensation claim safely

An agent checked a flight’s timing, estimated the payout, and drafted the airline claim without submitting it.

Done with Custom agent· Intermediate

What they did
A custom agent started with a flight number and date, decoded the airline code, then compared gate timestamps from FlightAware with AirHelp. It also pulled the relevant legal text from legislation.gov.uk, used a rules table to work out the amount owed, and prefilled Air India’s online claim page in a cloud browser with a built-in block against sending the form.
How it went
On Air India AI162, it calculated a 207-minute delay and an entitlement between £260 and £520. A complete run took two to five minutes. A smaller model sometimes entered the wrong values, and 8 of 10 submit-trap pages still got through the first text-based check.
Tips
Use a stronger model; gpt-5-mini worked, while gpt-4.1-nano made field mistakes. Put the anti-submit guard inside the page flow itself instead of depending on button-label matching.

Try it yourself with Claude in Chrome

Check my flight [airline/code] [flight number] on [date] and work out whether I qualify for delay compensation under [rule/jurisdiction]. Use public flight-tracking sources to verify the departure and arrival timing, consult the official legal text for the rule, and calculate the amount I can request. Then open the airline’s claim page, enter my details: [full name], [booking reference], [email], [phone], [address], and any flight information needed. Stop after the form is ready and wait for my approval before any irreversible action, including submitting or sending anything.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on dev.to ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.