I Build AI Tools for a Living. This Is the First One I Built for Myself.

I build with AI every day at work. Online, I’m constantly seeing people ship creative personal projects with these tools. But everything I’d built up until this point was for someone else. I love what I do, but I was tired of only building for other people’s problems.

I wanted to build something I was actually passionate about. Not for a portfolio, not to prove a concept to a stakeholder, but something I would actually use. I wanted to be my own target audience for once, and build a tool that would genuinely light me up.

Coincidentally, around the same time, I’d been getting really into this video game Helldivers 2. I was playing almost every night with my friends, and sometimes even by myself if they were busy. Somewhere between the drop pods and the bile titans, an idea clicked.

The itch

As I moved up the difficulties and started running Super Helldives (the hardest missions in the game), I noticed myself defaulting to the same loadouts every time. I’d try to mix it up, pick a new weapon, swap in a stratagem I hadn’t used. But constantly building around unfamiliar items got exhausting. It felt forced.

I kept thinking about how fun it would be to just get a random loadout and figure it out from there. And then I realized: I have the skills, I have the tools. Why am I not building this? That’s where Built on Break was born.

What was already out there

A few loadout randomizers already existed online, but they were pretty bare bones. They’d randomly pick your equipment with no real thought behind it, like rolling dice for every slot independently. You’d end up with no way to handle the harder enemies, or a build full of items you hadn’t even unlocked yet.

That second point was a big one for me. The game releases new equipment through seasonal content drops, and I didn’t have access to all of them yet. I wanted the ability to filter out items I didn’t own so I wasn’t constantly rerolling, trying to land on something I could actually take into a mission.

That alone was enough reason to build my own.

The main interface. Three modes, one button. Roll for democracy.

From filters to something smarter

The more I worked on it, the more I realized I could go beyond just filtering. I could add constraints to make the randomizer generate loadouts that were actually viable. Things like limiting how many of each equipment type you could get, or making sure you always had at least one support weapon. The randomizer would still surprise you, but it wouldn’t set you up to fail.

The friend who pushed it further

I started testing early builds with one of my close friends, and honestly, he’s the one who inspired me to take it to the next level. We were using the Balanced mode, which has basic structural constraints. The loadouts were always playable, but they didn’t go further than that. No guarantees you’d have the right tools for the mission, just a reasonably put-together build.

And we were already having a blast. Rolling loadouts and dropping into missions with whatever we got forced us to adapt, improvise, and use equipment we’d never have picked on our own.

But we kept running into the same issue. Sometimes a loadout would have no way to deal with the game’s toughest enemies, the heavily armored ones that require specific tools to take down. You’d load into a 40-minute mission and realize five minutes in that you just couldn’t win. That’s not a fun challenge. That’s a waste of an evening.

Mission Ready mode

That’s what led to Mission Ready mode, the feature that turned this from a novelty into something actually useful.

I built a scoring system that rates every stratagem for anti-tank and crowd control capability, then ensures your loadout covers both. You might get weird picks, combinations you’d never choose on your own, but you won’t be helpless against a Bile Titan.

A Mission Ready roll. Weird picks, but you can handle anything the mission throws at you.

It was the hardest part to build, but it’s what makes the whole thing work. My friend and I found ourselves running more creative loadouts, learning more about the game, and pushing our boundaries more than ever before.

Everything that came after

Once the core was solid, the features started stacking:

• Squad mode generates four complementary loadouts with zero stratagem overlap across the team
• Daily challenge gives everyone the same loadout worldwide, one per day
• Shareable image cards let you download your loadout as a branded PNG to send to friends

Squad mode. Four loadouts, zero overlap. Everyone brings something different.

None of that was in the first version. The first version was ugly and minimal. But I shipped it, watched how people used it, and added the things that would make them come back. Ship it ugly, polish it live.

What’s next

This was my first personal project built primarily with AI tools, and it went from idea to live site in about two weeks. The biggest thing I took away: build for yourself first. My friend and I were the first users, and that kept every feature decision honest. If we wouldn’t use it, it didn’t ship.

That’s all I wanted out of this. Build something real, ship it, and learn something along the way. Everything else is a bonus.

More to come.