I’m Paul Zammit. I’ve spent most of my career in product: building things with engineers, leading product teams, and trying to make sure the user does not get lost somewhere between a good idea and a polished roadmap.
Like most people working in technology, the last few years have changed how I think about software.
LLMs have made terminal and chat interfaces genuinely powerful. They can help people work through messy problems without forcing everything into rigid forms and menus. But the products built around them are often still a bit backwards: too much model talk, too many decisions pushed onto the user, and too much confidence from systems that have not earned it.
I kept coming back to one fairly ordinary problem: planning a holiday.
I have travelled a fair bit, and I have booked more than enough flights, hotels and rental cars to know the usual routine. Open Skyscanner. Open Booking.com. Open Airbnb. Search three versions of the same route because one website insists you pick a city when you really mean a country or region. Keep tabs open. Compare flight times, airport transfers, areas, prices, reviews, and whether saving €15 is worth arriving at 01:00 after a 14-hour connection.
It is not impossible. It is just tedious.
I thought this was exactly the sort of thing modern AI could help with. Not by planning a whole trip for me or spending my money on my behalf, but by doing some of the search work properly: understanding that “I want to go to Scotland” might mean Edinburgh, Glasgow, Inverness, or somewhere else entirely; finding the relevant options; showing the trade-offs; helping me widen or narrow the search as the trip takes shape.
Instead, I found a lot of products that either made things up, gave me stale or vague information, or pushed expensive sponsored options with a chat interface sitting on top. The specialised AI travel products were particularly strange. Ask a normal question and suddenly you are being shown €1,000-a-night hotels as though this is the natural starting point for a weekend away.
So I decided to build the thing I wanted to use.
Travelomi is not a travel agent, and it is not an itinerary generator. It will not tell you that it has planned your “perfect escape” after you type two sentences into a chat box. You are still planning the trip. You make the decisions and book with the provider you choose.
Travelomi’s job is to help you search travel properly.
Start with what you know. Maybe that is “somewhere warm next week”, “a proper wilderness trip”, or “we have two weeks and want Japan, maybe Korea, but we are flexible”. Travelomi should help you explore the real options, remember what matters as the plan changes, point out the trade-offs, and connect you to the flights, stays and sources you need when you are ready.
The goal is simple: less faff, better context, and more control.
Travelomi has just started. It will be built in the open, improve through real use, and probably get plenty wrong before it gets the important parts right. But I think travel search can be much better than it is now.
If that sounds useful, head to travelomi.com and have a look.
— Paul