Walkthrough
See Skyro in action.
A walkthrough of all six intelligence modules, from Flight Intelligence to Price Prediction.
Walkthrough narrated using an AI avatar with voice cloning
At a Glance
How might we turn the overwhelming complexity of travel planning into a single, confident booking decision inside a conversation?
* Based on personal testing across multiple travel scenarios comparing Skyro against traditional platforms.
The Problem
Travel planning is broken.
Nobody's fixed the decision layer
You open Google Flights. Then Skyscanner. Then Expedia. 30 minutes later you still don't know if you should book. The data is everywhere. The problem is none of these tools actually help you decide. They just show you more stuff to look at.
Nobody needed another place to search for flights. They needed something that would just look at the options and tell them which one to pick.
Every platform out there assumes you already know where you want to go and when. They're built for searching. Nobody built anything for deciding. That gap is where Skyro lives.
How might we shift travel from search-and-compare to discover-and-decide, making the right choice feel obvious rather than overwhelming?
The Product
A travel app that lives inside a conversation
Instead of showing more options, Skyro evaluates, scores, and recommends. Each module is purpose-built to eliminate a specific friction point in the booking journey.
How I Built This
Engineering intelligence through prompt design
Skyro is entirely a Custom GPT system prompt with no code, no APIs, nothing external. That's actually what makes it interesting. The whole product lives in one text file. Which meant every design decision was really a writing decision.
Skyro doesn't just make travel planning faster. It makes smarter, more affordable travel accessible to everyone.
Reduction in planning time, from 30+ minutes of tab-switching to under 2 minutes for a confident decision.
Clear recommendation per query. No comparison paralysis. Users know exactly what to do, not just what's available.
Insider-level deal intelligence, previously reserved for dedicated communities, now available to any traveler.
Encourages cost-driven exploration. Users discover destinations they wouldn't have searched for on their own.
* Based on personal testing across multiple travel scenarios comparing Skyro against traditional platforms.
Why I Built This
It started with a really long flight search
I was booking flights from Seattle to India. What should have taken 10 minutes turned into nearly an hour: 14 tabs open, prices changing every time I refreshed, still not sure if I was looking at a good deal or not. I booked eventually. Still have no idea if I overpaid.
That stuck with me. And when I started digging into what Custom GPTs could actually do, I wanted to answer a specific question: can you build something that genuinely feels like a product using only tools that are free and available to everyone? Skyro was me trying to find out.
What I Learned
Building this changed how I think about AI
Every choice about structure, tone, and format directly shapes what the product feels like. I'd rewrite a single line and the whole output would change. It didn't feel like writing. It felt like designing.
The early iterations were slow and messy. The prompt kept breaking on edge cases. But once I figured out the right structure, the rest came together in a few hours. The hard part wasn't building, it was figuring out what to build.
My instinct was to make it flexible. But every time I loosened a constraint, the output got worse: more generic, more hedged. The more I told it exactly what to do, the more it felt like a real product instead of a chatbot.
AI as a decision layer, not just an answer machine
I didn't set out to build something revolutionary. I was just frustrated with a bad experience and wanted to fix it using what was already available. The fact that it turned into something that genuinely works, and that anyone can try right now, is what I'm most proud of.
If you can use the tools in front of you to solve a real problem, that's enough. You don't need a dev team or a budget.
Try Skyro ↗