Use ChatGPT Agent to test points-based travel plans
Jitendra Jain tested ChatGPT Agent with a family holiday to Iceland and a restaurant reservation in Dubai. The travel brief asked for five days for four people in mid-October, using Marriott Bonvoy points and Emirates Skywards miles, with a PowerPoint as the deliverable.
Original prompt
Shared by the source author. See source notes for context.
Plan a 5-day trip for four, mid-October. Use 200k Marriott Bonvoy points + 200k Emirates Skywards miles.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Jitendra Jain (JJ)
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Ten-slide Iceland planning deck with missing availability and points-value checks; Dubai dinner search stalled at WhatsApp and required manual completion.
- Human involvement
- Jain checked the output, found an unavailable hotel and finished the WhatsApp booking step manually.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Specify dates, party size, loyalty balances and required deliverable.
- Inspect whether proposed flights and hotels have live availability.
- Take over blocked reservation steps and review missing travel constraints.
Jitendra Jain tested ChatGPT Agent with a family holiday to Iceland and a restaurant reservation in Dubai. The travel brief asked for five days for four people in mid-October, using Marriott Bonvoy points and Emirates Skywards miles, with a PowerPoint as the deliverable.
The agent produced a ten-slide presentation containing a daily schedule and activities such as the Golden Circle, Blue Lagoon and a glacier hike. It considered the loyalty balances conceptually, but Jain found important gaps. It had not checked live flight or hotel availability, recommended a Marriott that was already full, and supplied no valuation of the points. Visa timing was also absent. Roughly half the reported twenty-six-minute run went into making the slides.
His dinner test exposed a different execution boundary. The agent searched restaurant sources, identified romantic features and attempted form filling for an anniversary meal. When the reservation required WhatsApp, it stalled and Jain finished manually.
Together, the runs produced planning material and candidates while leaving decisive booking checks unresolved. The useful lesson comes from those observed omissions: a presentation and a list of restaurants do not establish availability, suitability of a points redemption or completion of a reservation.
Prompt
Original prompt
Source: hotelemarketer.com · Jitendra Jain (JJ)
Frequently asked questions
What was missing from the points trip?
Live flight and hotel availability, points-value analysis and visa timing were not handled in Jain's test.
Did a polished deck imply usable reservations?
No. One recommended Marriott was full, despite the deck containing a structured daily schedule.
Where did restaurant execution stop?
The booking required WhatsApp; the agent stalled and Jain completed it manually.
More details
About this account
The agent researched itineraries and restaurant options, made slides and attempted form filling.
Firsthand source opened and read. Prompt is a verbatim excerpt of the published original instruction.
Credit: Jitendra Jain (JJ)
Source published
Added to Agents In Use
Source checked
Tools and topics
ChatGPT agent · PowerPoint · Travel websites · Restaurant websites
Loyalty points · Availability checks · Booking limits
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.