Among the worries faced by game developers seeking a publisher is the gamble of a prototype – that is, “a playable build that meaningfully shows what’s good about your game – a proof of concept”, in the summary of Suspicious Developments boss Tom Francis. Specifically, Francis says a “prototypable project is one where you can build that in an amount of time you can afford to lose”.
Few independent devs have bags of time to lose, and there’s no guarantee the labour will be rewarded. And yet, many publishers today won’t even come to the table unless they can get their hands on a playable slice of a game. At Digital Dragons in Krakow this year, I spoke to people from 11 Bit, Jagex and Bloober Team, together with a couple of independent teams, about the seeming necessity of prototypes and the associated temptation to knock them together using generative AI.
Hubert Popławski is lead developer at Maverick Souls Studio, whose members include former developers of Dying Light 2 and Frostpunk 2. The studio’s first game, Before Darkness Falls, is a promising 3D survival tactics sim in which you defend and repair a collapsing industrial compound on a malevolent planet. The team had a prototype on show at Digital Dragons, hammered together in half a year after being awarded some initial funding by the Digital Dragon Accelerator programme.
A six month turnaround is pretty nippy for a team of 2-10 people. As Popławski told me after I’d sampled the build, a greener outfit would likely have fallen short. “It truly depends on the experience of the team,” he said. “I think juniors have a very hard time to deliver something that looks good enough to feel safe for investors and publishers, to look at it and consider when investing in it, right? So that’s not equal for everyone. For teams like us, for example, composed of veterans with more than five, ten years of experience, it’s much easier.”
It’s a problem both for younger devs and for smaller publishers who are looking to support riskier, hopefully breakout work, such as Frostpunk household 11 bit, whose third-party collaborations include Indika, The Thuamaturge and forthcoming squalorcore simulator Crop. “For us, I think the biggest challenge is the ever-present problem of indie developers not having budgets for prototypes, while we usually like to sign projects that have prototypes,” 11 bit’s external development director Rufus Kubica told me elsewhere at the show.
“Some developers or other publishers, or even sometimes fans, can fund prototypes, and they kind of solve the chicken-and-egg problem,” he went on. “For us, because we are so code-driven, and we need to have proof of the key features, the USPs and everything, we cannot work with a paper pitch. Unless it’s a very well-known developer approaching us, and we know the track record so we can talk about the project. But usually when you’re working with indies, it’s quite a new team, so if they don’t have a playable proof, it’s a really big problem for us.”
11 bit may not be as well-resourced as, say, Epic Games, who are prosperous enough to run prototype incubation programmes for tenderfoot teams. But Kubica added that in general, this is not a great time to be pitching videogames.
“There is this strange limbo of, if you cannot kick off things on your own, or you don’t have other funds, it may be problematic to find a publisher in the current industry,” he says. “All the publishers – and we are quite well connected with other publishers – we are very friendly because again, it’s about the spirit of the company. We want to support other indie devs and indie publishers, so we share how things work. And I see that everyone on, let’s call it our level, very generally, our level of publishing – everyone is very risk averse.”
For Bloober Team’s head of publishing Michał Gembicki, having a playable prototype is “absolutely crucial”. Speaking during a Digital Dragon panel discussion of the current publishing landscape, he insisted that securing a deal otherwise is “close to impossible, unless you are, of course, a recognised studio. [Or if] there is an IP you have, or you have a huge line-up, and you can say this is going to be like something you did in the past. Then yeah, of course, you can try, but still, [the publishers] will probably hold off until you at least deliver a video of something playable.”
“The overall point is that it’s a more risk averse backdrop,” added Anna Mostyn-Williams, senior director of business development and partnerships at RuneScape developers Jagex. “Anything you can do to help show de-risking is going to help. So of course, the prototype is going to help.”
Publishers may occasionally offer contracts funding the creation of a prototype, if they spy an idea they want to lock down before another company pinches it. They may also reach out to devs they trust and fund a prototype for a variety of game they want in their portfolio. But those opportunities aren’t thick on the ground, as I heard from other prospecting developers on the Digital Dragons show floor.
Michał Staniszewski is studio director for Plastic, a legendary Polish demoscene outfit that dates back to the 1990s. Over the past 20 years, Plastic have collaborated with Sony Santa Monica on a number of engrossing oddities – Linger in Shadows, VR game Datura and Bound – a beautiful game about a ballerina darting and spinning through a world of furious, fragmented colour.
A few years ago, a few members of Plastic split from Sony to go independent again. After posting videos of playable builds on Tiktok, they won funding from Epic for a procedurally generated sandbox game, the winningly daft Boundless Fun Simulator, in which you clip together game modes and props on the fly using stickers and sliders.
“Totally 100% agree,” Staniszewski said, when I asked if it’s harder now for an indie dev to get attention without a prototype. As for why publishers are more cautious, he points straightforwardly to rising development costs. “During our times with Sony, we were with Santa Monica external, and there was like, Journey and all the other indie studios, like Dear Esther and so on,” Staniszewski recalled. “And normally we were doing a game for four years, that was on average, and it was [doing] okay. And then all those studios, the next game was like, six to eight years. It’s a really hard, weird place that we are at, right now. It’s changed dramatically.”
As in other walks of life, one way in which indie developers have sought to trim their expenses is by using generative AI tools. There’s a rash of chatbot software that is aimed specifically at prototyping fast, regardless of expertise. In November 2025, Amazon Web Services introduced “Agentic Arcade”, a tools package mocked up to resemble an old coin-op dive, which boasts of being able to deliver “playable game prototypes in minutes”. Unity have their Unity Muse, aimed at “creators of all skill levels”. Within large, established teams such as Crystal Dynamics, meanwhile, generative AI prototyping has become routine at many stages of development, despite vociferous reactions to Steam disclosures.
Boundless Fun Simulator doesn’t, as far as I know, feature any generated materials, but Staniszewski is receptive to generative AI usage, specifically for smaller teams of seasoned operators. “I think that like, generating whole games is like, [an idea for the] distant future,” he said. “But automatic code generation for simple stuff like UI – if you are a good programmer with 10 years of experience, you will get a 10 times boost of your productivity. So I believe that the indies who consist of one person or two people will get really much more productive in a year, maybe, because it’s going so fast.”
For all the promised efficiencies, however, publisher executives I spoke to at Digital Dragons were highly ambivalent about pitches that include generated or partly-generated prototypes. “I’m not against AI myself, in general,” 11 bit’s Kubica told me. “It’s a useful tool, but a useful tool for particular things, and I like when things are human-made in the end.” The basic problem for a publisher, he went on, is that a prototype that makes extensive use of generative AI offers little proof of a team’s capabilities. “How can I be sure that this developer can make the quality they’re promising?” Kubica added. “If everything is done with AI, how do you know they have the artists to make it happen?” He adds that thanks to generative AI, “it’s kind of easier to overpromise quality early”.
Bloober Team’s Gembicki feels “there’s nothing worse than an AI generated pitch”. Speaking during the aforesaid panel discussion, he commented that “I would [prefer] to see a sketch or something work in progress, but done by hand, by that team, showcasing this art, they want to achieve – usually it’s about art, right? Rather than this AI content.” He does feel generative AI usage is “fine” for the market research aspects of a pitch, such as collecting data and producing charts. “But definitely not for writing text, and not for doing the artwork.”
Jagex’s Mostyn-Williams agreed. “Yes, use it for market info, for that side of things, but people can tell,” she warned. “It’s so obvious. So be really, really mindful of it. And if you are going to use it, make sure that you go through it afterwards and ‘de-AI’ it. I know, that’s an obvious point, but a lot of people don’t.”
For Maverick’s Hubert Popławski, a further consideration is the possibility of a PR disaster, even before you’ve signed a publisher. Generative AI pervades today’s games industry but it’s a much-disliked technology – culturally cannibalistic, energy-intensive, error-prone, and a pretext for workforce reduction. “The backlash from using AI is so scary,” he says. “People still don’t like the idea that you’re using AI. If you do, it will be ‘Oh, that’s an AI. No, thank you.'”
I myself despise generative AI, partly because I see it as part of a wider anti-social oligarchal project, but as with any tech, I think there are less-bad applications given some degree of responsibility and transparency – for example, proprietary, bespoke and circumscribed models that pay or otherwise obtain consent for the material they use for “training”. I mostly avoid games of any scale that make use of AI generation, but I can understand the appeal for a smaller team trying to catch the eye of a publisher, during a time of cutbacks.
One thing I forgot to ask Popławski at Digital Dragons is simply whether creating a prototype is fun. It feels like it should be the aspect of development that people relish the most. It’s the part where ideas become a tentative reality, in which the project has some solidity but is still very malleable. I find it sad, in general, that generative AI advocates focus so heavily on its usage for speedy “ideation”, brainstorming and testing of concepts. Surely, this is the part of the project where you feel least inclined to hand off work to some form of automation.
Speaking to me later over email, Popławski predictably took a less idealistic view. For him, making a prototype can be enjoyable but is more often a grinding gamble. “Creating a prototype is first and foremost stressful,” he said. “In some ways, it’s a kind of ‘First Judgement Day’ for the project. It can be fun when you have the funds and the team’s ambitions are aligned with the project’s scope, but if you’re starting from scratch and still need to prove yourself, there’s a lot of pressure, and that can take the fun out of it.”
The fixation with having a prototype can make discussions with potential backers feel utterly arbitrary, Popławski went on. “We had a situation in a B2B meeting where someone didn’t take us seriously regarding scope and funding until we showed them our gameplay reveal trailer,” he recalled. “Once they saw what we had actually built, they relaxed and nodded along, as if the claims that had sounded ambitious a moment earlier suddenly became completely reasonable.”
This feature is based on a trip to Digital Dragons in May 2026, with the event organisers paying for travel and accommodation.