Why I'm building Travelton

Albin Ng — Mon, 08 Jun 2026 15:09:26 +0000

Part 1 of Building Travelton. A series on building a community-first travel app with Google's Gen AI stack.

The pain is real

Ever drove three hours out to a quiet village, all hyped up, only to reach and find everything closed? Shutters down, lights off, and the one famous shop happens to rest on Wednesdays. Sian.

Or worse. You come home from the whole trip, scrolling through your photos, and realise you were fifteen minutes away from an anime event you would die-die have wanted to go to. And you completely missed it. Nobody told you. You didn't even know to look.

That's the part of planning nobody enjoys. Taking "I want to go Kyushu" and turning it into an actual day-by-day plan that doesn't have you zig-zagging across a prefecture, eating dinner two hours from your hotel, or walking right past the good stuff.

So I tried the obvious thing first: ask a general AI chatbot to plan the trip. Looked great until you read it properly. It sent me to a restaurant that closed years ago, clustered "nearby" stops that were actually 90 minutes apart, and gave me the same beige, could-be-anywhere itinerary it gives everyone. The plan read well. It just didn't survive contact with a map.

That gap, between text that sounds like a plan and a plan that actually works on the ground, is what Travelton is here to close.

What Travelton is

You give it your destinations and dates. It gives you back a day-by-day itinerary that's actually yours: mapped, planned, and worth following. Three things make it different.

It's bespoke, not one-size-fits-all. No two travellers are the same, so no two plans should be either. Travelton builds the trip around you. Your pace, your taste, the kind of holiday you actually want, instead of the generic top-10 list everyone else gets handed.

It routes you smart. Your leave is precious, so don't waste half of it stuck in transit. Travelton plans each day so your stops actually make sense together. Less time on the road, more time doing what you came for. Maximise the shiok, cut the backtracking.

You hear from people who've been there. The best tips never come from a brochure. They come from someone who's actually made the trip. Travelton brings in what real travellers have learned on the ground, so you catch the closed-on-Wednesdays, the event happening that weekend, the little things you'd only know if a friend who'd been there told you.

The hard part isn't generating words. It's making the words true. Every place real and open, every day that actually flows, every recommendation grounded in how people really travel that route.

Why this is a Gen AI problem (and why Google)

Two things have to be excellent at the same time, and they line up neatly with two parts of Google's stack.

Understanding and generating the plan. Travelton uses the Gemini API to take a loose intent ("5 days, Fukuoka and around, relaxed pace") and turn it into a structured, day-by-day itinerary. It also reads messy real-world inputs, like a PDF someone shared or a half-finished planning doc, and turns them into something the app can build on. Structured, multimodal generation is the engine room.

Grounding it in the real world. A plan is only as good as its geography. Travelton leans on Google Maps Platform (Places, Geocoding, and live map rendering) to check that every stop is a real, findable place and that a day's route actually holds together.

I'll go deeper on each of these in the posts ahead. How a paragraph becomes a structured plan, how the map keeps the AI honest, and further out, how the whole thing runs on Google Cloud.

Building this with the community

Here's the thing. Travelton was never meant to be built alone. The reason I even know about the closed-on-Wednesdays shop, or the anime event two streets over, is because someone who'd actually been there shared it. That kind of on-the-ground knowledge doesn't come from a brochure or a model on its own. It comes from people.

So Travelton is a community-first app, built for the community and with the community. The goal isn't just to hand you a nicer itinerary. It's to maximise everyone's travel experience and grow a circle of travellers who look out for each other. You share what you learned on your trip, the next person plans a better one, and it keeps compounding.

That's why I'm building this out in the open. I want you in it early. Tell me what's broken, what you wish existed, where you got burned on your last trip. Build it with me, don't just watch.

Follow along. Next time, how a paragraph becomes a plan.

DEV Community: Albin Ng

Why I'm building Travelton

The pain is real

What Travelton is

Why this is a Gen AI problem (and why Google)

Building this with the community