I had a pile of airline points and no time to hunt for seats. Now GATE hunts for me.

I have a loyalty account with a lot of points in it, a credit card that gives me a two-for-one voucher, and a winter trip to Bangkok for two of us. On paper that is a free trip. In practice it is a jungle.

The airline’s award search shows one route and three days at a time. Flights going west show seats everywhere. Flights going east show an empty calendar and a red banner. Switch the direction and suddenly there are seats. I would click through a week, change the airport, click through another week, and lose track of what I had already seen.

So I asked Molly, the AI assistant that runs on GATE, to take the hunt off my hands. Now GATE checks every day and every airport for me, in both directions, for two people, with the voucher applied, and hands me one page sorted by price, flying time or the best mix of both. I book the seats myself.

First, learn the rules

Before touching the search, Molly read up on how the program actually works: the award chart, the fees partner airlines add, how far ahead seats are released, when more seats tend to appear close to departure, and the small print on the two-for-one voucher. Which card earns it, which cabins it covers, how many people it applies to, and that the second traveller still pays taxes.

She also looked at every award-search tool and booking service people recommend. A few of the guides that recommend them carry referral codes, and GATE says so before recommending anything. The finding that mattered: the airline’s own search, logged in as me, is the only answer that counts. Every third-party tool is a guess about it.

That also explained my empty calendar. Airlines usually release award seats per flight and direction. Going east in winter is peak season, so those seats sell for cash. Coming home the same weeks is quiet, so the seats are released. Same plane, different answer.

Teach it once

GATE has a tool for exactly this kind of service, one with no API, where the same clicks repeat every time. It is called Autosurf. The search is written down once as a playbook: open the page, sign in, choose round trip, two travellers, the airports, the dates, search. GATE also noticed what else the page offers. My account had three vouchers on it, so the playbook ticks the one that expires first, and the same goes for any other option a site shows: filters, cabins, discount codes. My login sits in GATE’s vault, never in the playbook or the chat, and the saved session carries over from one sweep to the next.

Two tips I gave Molly turned out to matter. Searching from Copenhagen instead of Stockholm showed far more seats, so Amsterdam and Paris were worth a look too. And instead of filling in the form again for every date, the search can just click the next date tab on the results page. That made a sweep many times faster.

Then let it sweep

Now a sweep covers four departure airports, every outbound day across three weeks of November and every return day across most of April, for two people, with the first voucher ticked wherever a date has flights. Each day becomes a row of structured data: departure and arrival times, flying time, stops, airlines, cabin, points, and the points with the voucher.

Check every answer

A fast answer that is wrong is worse than none, so the sweep checks itself. Every saved result is compared with the date the page actually shows as selected. Once a click did not take and the previous day’s flights were still on screen. The check caught it, threw the row away and searched that day again.

The bigger catch was in the maths. At first I read the price on each flight as the price of the whole trip. Before showing me anything final, GATE picked one outbound and one return flight and read the airline’s own total. Each leg is priced on its own, and some partner airlines cost twice as many points per leg. Molly told me straight away that an earlier number she had given me was wrong, and the page now adds the legs the way the airline does.

What it found

The difference between airports was bigger than I expected. My home airport had nothing in either direction. The nearest hub had two outbound days and a handful of returns, all on an airline that costs double. Paris had seats on most days and Amsterdam on about half, many of them on partner airlines at the normal price.

The cheapest round trip for both of us came to half of what the same seats cost without the voucher, well inside my balance, with one stop each way. The page also shows the fastest option and what business class would cost, so the trade-off is mine to make with the numbers in front of me.

On my phone

The results live on one page made for the phone. Every bookable round trip is a card: the points for both of us with the voucher, outbound and return with times, stops and route, the number of nights, total flying time, and whether it fits my balance. I sort by cheapest, fastest, fewest stops, longest stay, or a best mix that weighs points and flying time together, and filter by airport and cabin.

What GATE does, and what it leaves to me

• It learns the rules, reads the small print and warns me about tools that are really ads.
• It runs the search on my own account, with my login kept in the vault.
• It checks every date and airport I care about, applies the voucher and verifies every answer.
• It puts it all on one page I can read on my phone.
• It does not book. I pick the trip and book it myself.

Points programs change their charts, fees and partners all the time. The playbook stays the same, and when the site changes, GATE tells me which step broke and Molly fixes it in the chat.