IO Interactive aims to accelerate the game development process | GI Sprint

When Ulas Karademir last worked at IO Interactive, the organization was under the ownership of Square Enix and primarily known for developing the Hitman series.

After leaving the company in 2014, Karademir took on significant roles at Unity and later collaborated with Crytek co-founder Cevat Yerli on creating RealityOS, an engine for distributed development.

Returning to IO Interactive as the chief technology officer, Karademir now finds himself at an independent company with multiple studios in cities like Malmo, Barcelona, Istanbul, Brighton, and Copenhagen. The studio is engaged in various projects, including Hitman, a fantasy game, and a James Bond title.

Having accumulated a decade of experience since his last stint at IO Interactive, Karademir is now focused on enhancing the company’s technological capabilities to streamline production.

"My primary interest lies in the production-related aspects," Karademir remarks. "I'm concentrating on dev ops, modifying our infrastructure, and modernizing our software development processes. Additionally, I am exploring tools that can reduce costs, expedite production, and enhance quality."

The concept of 'cheaper, faster, better' aligns with the new GI Sprint editorial series, underscoring the need for rapid iteration and testing in game development.

"In any game development pipeline, we've adhered to traditional methods for the past two decades," he notes. "We use digital content creation tools that are exported to the engine, edited, and then built for a platform, which is a time-consuming process. Improving iteration speed is crucial for feedback and faster production."

"In the realm of games, you're crafting enjoyment, which is inherently subjective. It's like delivering a joke without knowing if it will land. The key is to iterate swiftly, test, and bring ideas to life to determine their fun factor."

"Fun is kind of unmeasurable. It's telling a joke and you don't know if people will laugh. The best way to do it is to iterate faster and test it"

Developing tools to accelerate and automate some processes is fundamental to achieving this goal.

"Imagine upgrading from a simple screwdriver to a power drill," he analogizes. "For a professional carpenter, investing in high-quality, reliable tools is essential for efficient work. The same principle applies in game development; repetitive tasks necessitate advanced tools to ensure higher quality and faster production."

"IO Interactive has a legacy of 25 years, enriched with learning from past experiences. Effective production stems from these learnings," Karademir continues. "Boyd's principle highlights that iteration speed triumphs over iteration quality."

Karademir dismisses the notion of fully automating game development, emphasizing the need for human interaction to identify and address specific requirements.

"Automation has two avenues: replacing humans or enhancing their efficiency," he details. "Pure automation without human feedback is challenging and often falls short. Constant dialogue with game designers and artists allows for the creation of valuable tools, fostering a product-minded approach within the company."

One significant advantage for IO Interactive is its proprietary Glacier engine, a project Karademir contributed to during his earlier tenure. This customized engine allows the company to focus specifically on its game development needs.

"Using proprietary technology offers numerous benefits," he explains. "It permits swift decision-making and technological adjustments. The feedback loop between developers and users is remarkably tight."

"Commercial engines like Unity and Unreal constantly add features to cater to diverse user bases, resulting in a saturated engine. Users might only utilize a fraction of these features but still require stability and reliability," he adds. "In contrast, our engine is game-centric, enabling us to develop targeted tools that significantly impact production and can be rapidly evaluated for efficiency."

Despite the advantages, developing proprietary tools and technology comes with time and financial investment. IO Interactive's decades-long experience makes this feasible, but Karademir cautions newer studios against adopting this approach hastily.

"I wouldn't recommend this pathway for new studios," he advises. "There is little reason to build proprietary technology for games replicating existing trends and mechanics. Developing unique tech is justified only when pursuing innovative and unprecedented gameplay."

"IO Interactive has accomplished numerous impressive feats, such as our crowd systems… no other engines are equipped to handle these capabilities, but ours can. Hitman is a highly intricate game. Typical games might feature an NPC with varying behaviors based on the weapon it has, but Hitman differs significantly. Its AI integrates disguises, level setups, and lengthy gameplay segments that can extend up to 90 minutes. Generally, games function with simple cause and effect when a button is pressed. However, Hitman involves infiltration, costume changes, preparation, eavesdropping, and learning the AI's actions… it’s complex. Such unique games demand tailored technology."

For certain companies, opting to utilize pre-existing technologies like Unity and Unreal Engine and then customize them may seem logical. Nonetheless, this strategy has its downsides.

"Typically, developers adopt a Unity or Unreal engine, add custom plug-ins, and significantly modify the editor or engine. Consequently, they end up tethered to outdated versions during production. That issue doesn’t apply to us. We only have one engine, which we continuously update according to our requirements."

"I strive to minimize rework. Creating new features is exciting and enjoyable. But dealing with repetitive tasks or bug fixes is tedious."

Returning to the discussion on enhancing the iteration process, it’s not merely about cutting development costs but also addressing the surge in live service games. Developing a live game necessitates a user-centric, reactive approach, which transforms the engineering team’s mindset.

"We must adopt more of a platform-engineering and dev ops perspective," he elaborates. "This necessitates the integration of new tools, including advanced AI tools, to automate repetitive tasks. I aim to incorporate AI in writing tests; test creation for game development is cumbersome due to constant changes. It's indispensable for a software-as-a-service (SaaS) company. As we shift to a live service model, we must develop a SaaS-oriented engineering mindset."

Karademir concludes by emphasizing the potential of AI in eliminating unnecessary, repetitive tasks in development.

"AI can revolutionize areas such as study analysis, code analysis, and automated testing," he notes. "There are numerous start-ups focusing on this field. It’s a fascinating area."

"In general, I aim to reduce rework. Starting new tasks is invigorating and fun. Yet, dealing with redundant tasks or rectifying bugs is monotonous. These issues arise from misunderstandings, poor management, or lack of expertise. Once a process is understood, it can be automated, and I believe AI can significantly contribute in this domain."