Minesweeper Gets a Roguelite Remake — Interview with Duo Forge Games

Made with Bezi, Coinsweeper is a roguelite that heightens Minesweeper’s 50/50 adrenaline into a high-stakes gamble. To celebrate its smashing breakout release in April 2026, Bezi spoke with Duo Forge Games on Coinsweeper’s dev process.

Bezi: A roguelite that spins Minesweeper into an addictive gambling experience — it’s such a good idea that it almost seems obvious in hindsight. How did you two arrive at the idea?

Duo Forge: We’ve always loved the core tension in Minesweeper, that moment where you're not sure if a tile is safe or not, and wanted to make that feeling the entire game. Casino push-your-luck mechanics were a natural fit: the exponential cashout curve means every single reveal is a genuine decision. Games like Balatro showed us that combining a classic format with roguelite progression and gambling aesthetics could resonate with a wide audience, and we wanted to bring that energy to Minesweeper.

"Games like Balatro showed us that combining a classic format with roguelite progression and gambling aesthetics could resonate with a wide audience, and we wanted to bring that energy to Minesweeper."

B: Speaking of Balatro, it seems to be a trend — indie devs taking a classic game (poker, in Balatro’s case, Minesweeper in yours) and wrapping roguelite and deckbuilder mechanics around it. Why do you think this resonates so well with audiences? 

DF: Familiarity dramatically lowers the barrier to entry to your game. Everyone knows Minesweeper and poker. You don't have to explain the core mechanic to players, their brain is already locked in. That saved energy can then be invested into understanding the roguelite layer without overwhelming them.  Also, both Minesweeper and poker have a built-in risk structure. They are naturally games about uncertainty and decision making under pressure. That makes them ideal foundations for roguelites, where exactly that tension is the emotional core.

B: One reviewer pointed out that Coinsweeper finally gives players "a meaningful way of dealing with 50/50s" — the classic Minesweeper frustration where the game is unwinnable by logic alone, and you’re forced to guess at a coin flip. How did you think about that design when building Coinsweeper?

DF: We think the traditional Minesweeper community has mostly treated it as a bug to be fixed, generate boards that are always logically solvable, eliminate the guess entirely. And that's a valid approach. But it always felt to us like it was solving the wrong problem.

Because here's the thing: the 50/50 moment isn't just a frustration, it's an adrenaline-packed feeling. That moment of standing at a coin flip, knowing one tile could end everything, is genuinely tense and dramatic... the GAMBLE. The problem was never the guess itself. The problem was that the game gave you nothing to work with when it happened. No agency, no tools, no way to tip the odds. You just clicked and hoped.

"What Coinsweeper does differently is treat the 50/50 as a design opportunity rather than a design flaw."

What Coinsweeper does differently is treat the 50/50 as a design opportunity rather than a design flaw. The roguelite and deckbuilding layers exist precisely to give you meaningful tools for those moments. Power-ups that reveal tiles, mechanics that let you hedge your bets, risk/reward decisions that make even a bad situation feel like your decision. The guess is still there, but now you have context and resources around it.

In a way, the whole game is built around that one emotional moment. Instead of removing the cliff, we gave the player a parachute. We also give players the option to cash out, so they always have the choice to stop gambling and continue safely on a new board. 

Now, balancing that risk/reward curve was a tough design challenge, and in some ways, the most difficult part of making the game. We wanted every run to feel fair but never predictable. Too punishing and players feel cheated, too generous and the tension disappears. Getting the 70 power-ups to interact in interesting ways without breaking the economy took a lot of iteration. We had spent a long time working on the core balance, but Bezi helped us take it from "good enough" to truly polished.

B: 70 power-ups! How did you go about that? Did you draw from existing Minesweeper iterations?

DF: We didn't lean much on existing Minesweeper iterations; most of them stay too close to the original. 

"Our starting point was always the player's emotional state. What does it feel like to be cornered? To be on a lucky streak? We worked backwards from feelings into mechanics."

Our starting point was always the player's emotional state. What does it feel like to be cornered? To be on a lucky streak? We worked backwards from feelings into mechanics.

Bezi was a huge part of the prototyping process here. Because it actually understands your project, your scripts, your architecture, your existing systems. We could spin up a new power-up idea, test how it interacted with everything else, and see if it felt right, all really fast. Without that we'd have spent way more time on implementation and way less time on the actual design. It genuinely let us try more ideas.

For keeping 70+ power-ups from conflicting, we categorized everything early, risk modifiers, reveal mechanics, economy boosters and set rules for how categories interact. That gave us a framework instead of just patching problems one by one.

B: Any interesting insights you gained about balancing, throughout that risk/reward curve optimization process?

DF: The biggest lesson: players aren't rational calculators. We spent a lot of time tuning numbers based on expected value, but how something feels is completely disconnected from what it actually does mathematically. A small but visible advantage will feel better than one that quietly doubles your odds in the background. Perception of agency matters as much as actual agency.

The other thing that surprised us: the middle of a run is the most dangerous place. Early game is forgiving, late game has momentum. But the middle is where players quietly disengage if the pacing drifts. We spent a disproportionate amount of time just on that section.

And one thing we didn't expect, some player-discovered "exploits" turned out to be the best parts of the game. If something feels powerful but earned, that's a feature, not a bug. Playtesting with real players surfaced interactions we never would have found internally.

B: How did you know the game was ready to release?

DF: Honestly, we knew it was polished when we just started playing it instead of testing it. That mental shift, forgetting you made it, that's the moment.

Two things got us there. Bezi was honestly a game changer for balancing. It helped us spot broken power-up interactions and edge cases way faster than we ever could manually. For a 2-person team, having that kind of support was huge. They say "AI for building real games, not generating them" and that's exactly how it felt. We genuinely couldn't have iterated as fast without it.

And of course, nothing beats real humans. We had about 10 friends from the Fragout Community playtesting basically the whole development. Shoutout to them, seriously. They were there through the ugly early builds and kept giving us honest feedback all the way to launch. That's what really pushed it over the line.

B: Coinsweeper went out with "Positive" reviews from day one and people logging serious hours (one reviewer's at 172 hours already). That’s great positive signal for your first-ever release. How are you thinking about game marketing, and what are you learning?

DF: Honestly? It was mostly "release and pray" sending emails everywhere and hoping something sticks. Getting attention as a first-time indie is just hard, full stop.

Reddit ended up being our most effective channel by far, which was a bit by necessity. We couldn't crack short-form video. Coinsweeper is a slow, thoughtful game and it just doesn't translate well to a 15-second TikTok clip. So Reddit, where people actually read and engage, felt like a much more natural fit.

Our budget was basically zero, so paid promotion wasn't really an option either.

We also talked to a lot of publishers, including a AAA publisher, which was a wild experience. They wanted us to scale the game in a direction that wasn't our vision. Maybe that would've meant more reach. That said, we're genuinely grateful for those conversations. They were great people and we learned a lot. Who knows, maybe we work together down the road on something that's a better fit.

So we kept it independent. No regrets, but yeah, marketing as a first-time dev is humbling.

B: What’s next for Duo Forge Games?

DF: We're two brothers from Germany in our mid-to-late twenties, both working professionally in the IT industry. We've been developing small games as a hobby for about 5 years, but all previous projects were either small demos or never meant for the public. Coinsweeper is our first commercial release and our first game on Steam. Now that we've cleared that first big hurdle, we already have 2 new games in the pipeline that we'll be announcing soon.

Check out Coinsweeper on Steam now.