[Guest post] “What Getting My Time Back Taught Me About AI in Development”

By Ali Al Hamwi, Co-founder and CEO, SIFOR

I’ll be honest, I didn’t set out to become an AI evangelist. Like a lot of developers right now, I was wary of it. There’s been so much noise around generative AI in games, and a lot of it hasn’t matched what actually shows up when you sit down and try to ship something. So when we started using AI tools more seriously in our process at SIFOR, I went in with the goal of trying it out and expecting to stay skeptical. Then I realized how much of my time I was getting back.

Where things started

We build hyper-casual and roguelike titles at SIFOR, usually with a small team juggling a handful of projects at once. Abduct ‘Em, one of our recent titles, is an isometric sandbox roguelike, with dense neon and 16-bit visuals and players tearing through biome after biome, harvesting and wrecking everything in sight. A game like this needs a huge amount of props and characters, and they all need to be distinct and feel like they belong in the same universe.

Filling out this kind of a world used to involve briefing an artist for every prop, every enemy, and every little environmental detail, and then sitting through revision after revision until everything matched. Of course this process works, but it can be lengthy. And when you’re trying to move at the pace hyper-casual demands, the building time can cost you in ways you don’t see until you add them all up.

I don’t think I really clocked how much time this was eating up until I tried using a proper AI sprite and asset tool on a real production stretch.

How AI helped transform our pipeline

We used Ludo.ai’s Image Generator and 3D Asset Generator for a stretch of production work on Abduct ‘Em to see what it could do with the UFO, our main character, and what we found was a full pipeline we hadn’t anticipated. The Image Generator handled the UFO’s concept work and let me iterate on different silhouettes until things looked right. Then the 3D Asset Generator took the concept through to a game-ready mesh.

We were surprised by everything we could do with the UFO. The environmental props, the trailers, the military hardware, the shipping containers filling out each biome, all came out of the same generation ecosystem. And we were able to achieve a cohesive visual language right from the first asset, instead of building it over weeks of revision. Once we had the UFO’s look locked, every prop we used afterwards was generated against the same style reference, so nothing drifted. Setting the reference once kept everything aligned.

What stood out to me wasn’t just the speed. It was how much it shortened the gap between having an idea and actually seeing it inside the game. That changes the way you can experiment, especially as a small team.

In the end, I didn’t end up with less to do. Instead, more of my time went towards the calls that actually need a person behind them, the kind of judgment stuff an AI tool can’t do for you.

Where this leaves the industry

Whenever I hear people talking about AI in games, it’s usually about whether it will take over for games or developers. From where I sit running a small studio, I think we’re looking at more of a reshuffling. It’s great at speeding up the repetitive, technical stuff, and can give developers more room to focus on decisions that round out a good game. That said, I think studios still need to think hard about how they use these tools, and where the line sits between assistance and handing over your creative judgment entirely. Every studio has to work that out for itself.

This is especially relevant for small studios. When you have limited people and limited time, getting some of that time back doesn’t just mean producing faster. It gives you more room to think, experiment, and make better creative decisions.

My guess is that AI ends up becoming another fixture in a developer’s toolkit, the way version control or a game engine already is. I went into this planning to stay cautious, but I’ve come out the other side leaning into it. What it really gives you is more time, and when you’re running a small studio, that’s about as good as it gets.