Building a Startup Solo With AI: How Itamar Built Kivoon With Claude

It started with a very ordinary kind of frustration. Day one on the Isle of Man. Hungry. And absolutely no idea how to get to the nearest McDonald's.

If you've ever landed somewhere new and stared blankly at a bus stop, you know the feeling. Most of us shrug, give up, and order a taxi. But that small moment of friction is exactly where Itamar's story takes a turn, because instead of letting it go, he turned it into Kivoon, a public transit app for the entire Island, complete with timetables, live tracking, and a marketing engine that practically runs itself.

The best part? He built it solo, over a few weekends, with Claude doing a huge amount of the work right alongside him. In his recent Launchpad Talk, Itamar walked the room through the whole journey, and it's a genuine blueprint for anyone wondering whether building a startup solo with AI is actually realistic. Spoiler: it is.

From a missing McDonald's to a real product

The lesson buried in Kivoon's origin story is one every founder eventually learns: the best ideas rarely arrive as grand visions. They show up as annoyances. A problem you personally felt, in a moment you actually lived.

Itamar didn't set out to "disrupt public transit." He just wanted to get somewhere and eat. But because he paid attention to that friction, and because he had the tools to do something about it, a personal inconvenience became a product that helps everyone on the Island move around more easily.

That reframe matters. You don't need a billion-dollar idea to start. You need a real one.

Wrestling the data into shape

Here's where most people would have stalled. Public transit data is messy. The raw information sitting behind timetables isn't exactly handed to you in a neat, ready-to-use package.

Itamar's breakthrough came when he found GTFS, the General Transit Feed Specification, the industry-standard format for transit data. Once the raw information was parsed and mapped into GTFS, everything else suddenly became possible. Timetables, routing, live tracking: they all had a solid foundation to build on.

This is one of those quietly inspiring details. The hard part of a project often isn't the flashy feature at the end, it's the unglamorous groundwork of getting your data into a shape you can actually work with. And this is exactly the kind of task where working with AI shines. Parsing, reformatting, debugging edge cases, figuring out an unfamiliar standard: Claude could move through that grind quickly, letting Itamar stay focused on decisions rather than drudgery.

Claude as a one-person team

The phrase Itamar keeps coming back to is the idea of AI as your team. Not a magic button that builds the app for you, but a genuine collaborator you can hand real work to.

Throughout the talk, he gets specific about the actual prompts and decisions that moved Kivoon forward. It wasn't one giant request that spat out a finished app. It was a series of practical conversations: solving a parsing problem here, structuring a feature there, working through a tricky bit of logic, then moving on to the next thing.

That's the realistic picture of solo founder life in 2026. You're still the one making the calls, choosing what matters, deciding what "good enough" looks like, holding the vision. But the heavy lifting that used to require a small engineering team can now sit with an AI partner who never gets tired and never runs out of patience.

The self-improvement loop and the real-time vehicle matcher

One of the most impressive parts of Kivoon is its self-improvement loop, and at the heart of it sits the real-time vehicle matcher, the piece that figures out which bus is actually which, live, as vehicles move around the Island.

This is where the project graduates from "neat side project" to "real software." Matching live vehicles to scheduled routes in real time is a genuinely hard engineering problem. The fact that one person, working weekends, could ship something like this with Claude alongside them is exactly the kind of thing that would have sounded far-fetched just a couple of years ago.

It's a reminder that the ceiling for solo builders has moved. Dramatically.

A marketing engine that runs on its own

Building the product is one battle. Getting anyone to use it is another, and it's the one that quietly kills most indie projects. Founders ship something great, then run out of energy for the relentless grind of content and promotion.

Itamar sidestepped that trap by building an automated content engine into Kivoon. The marketing, in his words, ended up running on its own. Instead of manually posting updates and scrambling for attention, he set up a system that keeps the app visible and growing without demanding his constant attention.

For a solo founder, that's gold. It's the difference between a project that fizzles after launch and one that keeps building momentum while you sleep.

So… could you ship the thing in your head?

That's really the question at the centre of this whole talk. If you've ever had an idea rattling around your brain, an app, a tool, a fix for something that bugs you every single day, and assumed you'd need a co-founder, a budget, and a team of engineers to make it real, Kivoon is a quiet, convincing argument that the rules have changed.

You need a real problem. A willingness to wrestle with the boring parts. And an AI partner you're prepared to actually collaborate with, prompt by prompt, decision by decision.

Itamar started with a missing McDonald's. Look what it became.

▶️ Watch the full video: https://www.youtube.com/watch?v=4yluSY-9d9s

🚀 Register at the Launchpad community — join the Isle of Man's startup hub of founders, freelancers, mentors, anyone curious and investors building on the Island. Come for the talks, stay for the community. discoverlaunchpad.com