Voodoo Fishin’ is an unusual blend of cozy co-op fishing and creepy horror, and it just launched into Steam Early Access.

The game is being built by a three-person team at Lost Arcade with a strong focus on rapid iteration and systemic design. Set in a stylized bayou filled with charm, tension, and emergent chaos, the game leans heavily on player-driven moments, where relaxed social gameplay can quickly spiral into unpredictable encounters.

A key part of that development process has been Bezi, an AI-powered tool integrated directly into Unity that the team used to accelerate prototyping, build internal tools, and reduce workflow friction.

In our interview, the Lost Arcade team broke down how Bezi shaped their pipeline, how they structured scalable systems for fish and encounters, and what it means to ship a system-heavy co-op game as a small indie team.

Voodoo Fishin’ blends cozy co-op fishing with horror elements. How did that core concept come together, and what were your early prototypes like?

Lost Arcade: The first thing we knew was that we wanted to make a first-person game. Voodoo Fishin’ originally started as Voodoo Voyage, a riverboat survival/runner in which fishing was just one mechanic. Those early prototypes were fun, but we realized pretty fast that we were trying to do too much if we wanted to actually ship it in a year.

What really stood out to us was the world we had been building. We loved the voodoo setting, the bayou atmosphere, and especially the social feel of just hanging out and fishing together as a group of silly puppets. So we stripped the idea back to what was basically the “lobby” from Voyage and asked, “What if this was the game?” That became Voodoo Fishin’, and from there we rebuilt it into something more focused: creepy-cozy co-op fishing with a horror edge.

Then, we used Steam’s Fishing Fest earlier in the year to validate the product idea by shipping a somewhat polished but very early demo, and have been iterating since.

The game leans into a “creepy-cozy” tone. How do you balance tension and humor from both a design and technical standpoint?

Lost Arcade: We think the biggest factor in selling that balance is the puppet aesthetic. Players load into a warm, inviting dock that feels safe and full of charm, which gives us a strong, cozy foundation right away. We’ve also tried to lean into humor through the bait, tackle, and some of the game’s stranger, funkier details. Our goal is for the horror and tension to build naturally as you move farther away from the main dock.

From a technical standpoint, a lot of that balance comes from how we control feedback and escalation. Audio, volumetric lighting, and environmental effects do a lot of the heavy lifting. Small shifts in sound, visibility, or movement can make a situation feel tense without overwhelming the player. The game can feel calm and relaxing one moment, and then suddenly take a turn when the monsters show up as you’re trying to reel in your catch.

In the end, we want players to feel like they’re hanging out in a cozy world that just happens to be a little haunted, and sometimes a lot more haunted than they expected.

You’ve described the game as a co-op experience built around shared chaos. How did that philosophy influence your core gameplay loop and systems design?

Lost Arcade: I think it’s both shared chaos and shared chill… As we mentioned before, we wanted Voodoo Fishin’ to be a space where friends can relax, fish, and hang out, but have everything go sideways fast.

A lot of the systems support that swing: proximity chat, valuable catches, hostile threats, and recoverable lost gear. The goal was to create mechanics that naturally generate stories, where calm social moments can turn into panic, teamwork, or comedy in seconds.

Managing large numbers of fish types and encounters can quickly become complex. How are you structuring your data and systems to keep things scalable and maintainable?

Lost Arcade: The team has a lot of previous xp building digital card games, so a data-driven approach was a natural fit for us. A lot of the way the fish work in Voodoo Fishin’ is structured much like a card game. Each fish is basically a card that gets dealt to specific Fishing Area decks at the start of a session.

When players fish, the system checks their Bait against the available fish in that area, filters the pool based on valid matches, and then rolls within that filtered set to determine which bites and which get hooked. That gives us a nice balance of “predictable randomness” we’re after. It still feels dynamic, but players can learn the system and make smart choices when they’re targeting specific fish. 

To keep all of that scalable, we built internal editor tools that let us quickly browse, filter, and tune fish data in one place. That tooling has been incredibly important for keeping the game maintainable as the number of fish and tackle combinations grows.

The game uses proximity voice chat as a core feature. How does that system influence gameplay design, and what technical considerations came with implementing it?

Lost Arcade: Proximity voice chat was one of those features we saw as essential from day one. In co-op horror, especially, it adds so much to the experience that it quickly becomes part of the game’s baseline. It changes how players move, communicate, and react to danger, and it creates a lot of both tension and comedy.

Because we implemented it early, it influenced gameplay design right away. Players naturally make different choices when communication depends on distance. Technically, the main challenge was making it feel reliable and seamless, so it supports the experience without drawing attention to itself.

The swamp itself feels like an active threat. How are AI behaviors and environmental systems structured to create that sense of unpredictability?

Lost Arcade: We try to make the swamp feel like it has agency, not just a backdrop. A lot of that comes from layering AI behaviors, environmental hazards, and atmosphere so the whole space feels hostile and alive.

With AI, we want threats to be readable enough that players can learn them, but varied enough that encounters do not become routine.

A lot of the unpredictability comes from systems colliding, where danger builds naturally from the world before a monster ever appears. 

As a small indie team, how did your production pipeline evolve over time, especially after transitioning from traditional studio environments?

Lost Arcade: After years of working in bigger studio environments, this was the first game we’d built fully for ourselves at this scale. That meant adapting our pipeline to work without the usual external structure, while staying disciplined about scope and honest about what we could realistically ship as a trio.

At a big studio, you usually have outside constraints like clients, stakeholders, or existing IP helping define the rules. Here, the buck stops with us. That’s exciting, but also a little scary. Over time, our process has become more focused on practicality, cutting what doesn’t serve the game, building tools where needed, and making sure we stay laser-focused on what matters most.

What tools and workflows have been most critical for maintaining iteration speed across art, design, and engineering?

Lost Arcade: The biggest lever for us has been Bezi, an AI tool that plugs directly into Unity. It has dramatically increased our iteration speed and allowed us to take on ideas we probably would have cut as a three-person team.

For example, our fish tool might have taken a week or two in a more traditional workflow, but we were able to build it in under a day. It also helps that each of us effectively has a personal assistant with context on the whole project, so we spend less time distracting one another for answers and can keep multiple areas moving in parallel.

You’ve mentioned moving away from more traditional production models. What changes had the biggest impact on your team’s efficiency?

Lost Arcade: One of the biggest changes has been the speed at which we can validate ideas. We do a lot less traditional upfront planning now, because with tools like Bezi we can prototype features very quickly. In a more traditional setup, we might have spent much longer debating idea A versus idea B before committing. Now we can stand something up in a few hours and see if it actually works.

That does not make polish or final production any easier, but it massively speeds up the validation phase and helps us make decisions with less hesitation. It gives us more shots on goal in less time.

What tools or processes do you rely on for debugging and profiling in a co-op, system-heavy game?

Lost Arcade: Aside from the standard Unity tools, one of the best debugging resources we have is our Discord community. They are incredibly good at finding the edge-case bugs and strange situations we miss.

They also bring a wide range of hardware, network conditions, and playstyles that we simply cannot afford to fully replicate as a small team. For a co-op, system-heavy game, that kind of real-world testing has been invaluable.

You’ve publicly discussed using Bezi during development. At what point did you introduce it into your workflow, and what problem were you trying to solve?

Lost Arcade: We’ve been using Bezi on this project for about 9 of the 14 months it has been in development. Early on, we experimented with other AI tools, but they often felt clunky, disconnected, and inconsistent. What made Bezi different for us was that it plugged directly into Unity and could maintain a continuous context for the project, which was a huge usability shift.

We initially brought it in to help with tooling and internal workflows, especially anything that could remove friction for a small team. Over time, as the tool improved, it became useful across more and more of development. At this point, it is helping generate a large share of our code, which has had a major impact on how quickly we can iterate and validate ideas.

For those unfamiliar, how would you describe Bezi’s role in your pipeline compared to traditional tools or coding workflows?

Lost Arcade: It feels like each of us has a personal Unity expert at our fingertips. We have all used Unity for over a decade, but traditionally there is still a lot of time lost jumping to forums, Stack Overflow, or documentation when you hit a gap.

Bezi changes that by keeping that support inside the workflow. It reduces context switching and helps us move from question to implementation much faster. Because it understands the project, it is also far more useful than a traditional search-based workflow.

One of Bezi’s strengths is its ability to understand the full Unity project context and assist with scripting, debugging, and tooling. How has that changed the way your team approaches development tasks?

Lost Arcade: Like we’ve mentioned, Bezi having full project context has been a game-changer for us. For each of us, it saves a huge amount of time by cutting down on context switching, interruptions, and time spent searching for answers. We can stay focused and move faster without constantly pulling each other away from our work.

For a three-person team, that’s been massive. It has also made us much more willing to prototype features, build internal tools, and test ideas quickly, because the cost of validation is so much lower than it would be in a more traditional workflow.

The use of AI in game development is a sensitive topic in the industry. What are your thoughts on how tools like Bezi fit into that larger conversation right now?

Lost Arcade: We understand why it is a sensitive topic. A lot of people in games are worried about creativity being flattened, jobs being displaced, or studios using AI as a shortcut instead of investing in people. We think those concerns are valid, especially because we have seen that side of the industry ourselves. We have each been laid off recently and understand how larger studios can operate when new efficiencies emerge.

We have also been around long enough to see major technology shifts change what team sizes look like. Tools like Unity dramatically lowered the number of people needed to build and ship games. At one point, it was striking to watch teams go from hundreds of developers to something much smaller and more agile. This AI wave feels like the next version of that in some ways, with smaller teams gaining the ability to do more.

Our thesis at Lost Arcade is that we make hand-crafted, quirky games with no generative art. For us, tools like Bezi are about speeding up iteration, prototyping, and tool creation, not replacing the human voice behind the work. If anything, we think that in a world with more AI-generated content, distinctly human-made art and design stand out.

Finally, what general (or specific) advice do you have for other small teams looking to get their games over the finish line?

Lost Arcade: There has never been a better time to make games as a small team. The tools are better, faster, and more accessible than ever, which means the barrier to actually building and shipping something is lower than it has ever been.

Our biggest advice is to find one or two people you genuinely work well with and just start making things. That trust and momentum matter a lot. Small teams can move incredibly fast when everyone is aligned and willing to stay practical about scope.

The other big thing is to finish. Keep the idea focused, cut more than you want to, and get something real in front of players. You learn so much more by shipping than by endlessly planning. A finished small game teaches you more than a dream project that never gets out the door.

Lost Arcade, Game Development Studio

Interview conducted by David Jagneaux

Don't forget to subscribe to our Newsletter, join our 80 Level Talent platform, follow us on Twitter, LinkedIn, Telegram, and Instagram, where we share breakdowns, the latest news, awesome artworks, and more.

Are you a fan of what we do here at 80 Level? Then make sure to set us as a Preferred Source on Google to see more of our content in your feed.