Side, a seasoned services provider, has entered into a six‑month memorandum of understanding with Danish artificial‑intelligence specialist Modl.ai to co‑develop an AI‑powered quality‑assurance (QA) platform. The partnership blends Side’s extensive experience in the services sector—formerly operating under the name PTW—with Modl.ai’s relatively fresh perspective on gaming technology.
Modl.ai, founded in 2018, secured $1.7 million in seed funding two years later and has since been exploring ways to embed AI into the testing workflow. Martin McBride, Side’s chief commercial officer, describes the agreement as a convergence of the two companies’ strongest assets into a singular, differentiated solution. “One of the holy‑grail aspirations for the games industry today is to harness AI to augment QA services more effectively.
No one has cracked that problem yet, and we believe we’re on the brink of doing so,” he says. Earlier AI‑enabled QA tools—some of which were also created by Modl.ai—required developers to hand over entire builds or even source code to the testing firm. Those approaches relied on deep integration via plugins and SDKs, which introduced considerable friction and, as co‑founder Christoffer Holmgard notes, became cumbersome over time.
The new system eliminates the need for any internal access. Instead, developers deliver the product exactly as they would to a QA tester or, ultimately, to a consumer. The game runs on a device that mirrors the target platform, while the AI watches the screen, simulates player inputs, records every observable event, and then passes that data to a second AI layer that flags potential bugs.
Human reviewers then decide whether each issue warrants action. The first phase of the workflow is what Modl.ai calls building “game knowledge.” Developers upload assets, design documents, screenshots, or any other relevant material and converse with the AI in plain language. “We simply chat with the system about our game, supplementing the conversation with images or design docs,” Holmgard explains.
This natural‑language interaction helps the AI construct a contextual model of the title. Next, developers specify the testing objectives. These can range from traditional test cases to novel scenarios crafted to exploit the AI’s strengths.
Holmgard emphasizes that human expertise is still crucial at this stage: “You need skilled people to shape the prompts and guide the AI so it performs the right actions.” Once the parameters are set, the system executes a “run,” generating bug reports that include textual descriptions, short video clips as evidence, and comprehensive log data. These artifacts are handed to seasoned QA specialists, who review them and determine the appropriate next steps for the production pipeline. Distinguishing a genuine bug from an intentional feature remains a nuanced challenge for AI. Holmgard says that feedback loops from human testers help the system learn these subtleties.
“Some learning comes from the prompts, some from the pre‑built game knowledge, and a significant portion comes from ongoing tester feedback that labels each report as a bug, not a bug, or a feature,” he notes. Over time, this iterative training refines the AI’s judgment.
Side envisions a future where QA personnel transition into AI‑engineer roles, thereby elevating the skill set and compensation of its workforce. McBride highlights the strategic benefit: “Instead of a linear career path from junior tester to senior tester, our staff can become AI engineers, opening new career trajectories and allowing us to repatriate jobs to regions where they have historically disappeared due to low‑cost labor pressures.” By automating repetitive, low‑value tasks—such as smoke tests for each new build—testers can focus on strategic analysis, creative testing, and deeper gameplay evaluation, leading to more fulfilling careers.
Benedikte Mikkelsen, Modl.ai’s chief product officer, echoes this sentiment: “Many current tests are monotonous and can be automated. This frees humans to concentrate on higher‑order activities like assessing gameplay feel or conducting exploratory testing, which is where real value lies.” Both leaders stress that AI will not replace QA workers but will shift them up the value chain. McBride points out that in regions where testing costs are extremely low, a full AI takeover is unrealistic.
Providing advanced tooling, however, enables staff to upskill and remain relevant longer, benefitting both the individual and the company. Side’s philosophy treats AI as a “sliding scale,” with a human always present in the loop.
“That’s our ethos,” McBride affirms. “The tool lets us capture the best of AI—speed, consistency, error reduction—while still relying on human judgment for the final call.” When asked about the types of games the solution currently supports, Chief Operating Officer Oliver Deslandes admits the focus is on simpler titles—puzzle games, card games, and other experiences with repeatable mechanics—rather than fast‑paced 3D open‑world or fighting games.
“We’re targeting projects where the mechanics are predictable enough for the AI to learn effectively. More complex, high‑speed genres will come later as the technology matures,” he says. Deslandes adds that the initiative is a collaborative journey. Early adopters willing to pilot the system will help shape its evolution, providing real‑world data that informs future enhancements.
“If you join us now and run pilots, you’ll see firsthand where this technology can go,” he concludes. In summary, the Side‑Modl.ai partnership aims to create a hybrid QA ecosystem where AI handles the heavy lifting of repetitive testing, while human experts provide strategic oversight, nuanced decision‑making, and continuous feedback. This collaborative model aspires to raise the overall quality of game releases, improve employee career prospects, and potentially revitalize QA employment in regions that have seen job erosion due to cost‑driven outsourcing.