In April, Cubeia, an iGaming platform developer, decided to fully embrace AI-assisted coding across its systems, with a specific focus on implementing Claude's large language model (LLM) to enhance the speed of development and updating processes. By mid-July, the company’s COO, Stefan Grenstad, reported that the front end of Cubeia was 100% driven by AI, with back-end systems closely following at 90% to 95%. This strategic shift has met Grenstad's expectations, resulting in faster release rates and an increase in the number of updates compared to the previous reliance on human-written code.
Despite these advancements, the transition posed unexpected challenges for the development team. Grenstad noted that team members who struggle with context switching found the new demands taxing. "What’s been very obvious here is that those of my developers that have a hard time with context switching are struggling because to increase productivity, you need to be a multi-tasker," he explained. He elaborated that developers can run multiple tasks concurrently, and while this capability theoretically allows for many releases in a single day, effective multi-tasking remains essential. It is critical to determine the optimal number of parallel tasks a developer can manage, which Grenstad admitted was a surprising side effect of the transition requiring new training and organizational adjustments.
As part of addressing these challenges, Grenstad reported ongoing internal discussions about establishing limits on the number of daily tasks. There has been varied feedback from the team regarding the shift to AI-driven coding. Some tech leads have expressed skepticism, while others have adapted quite positively. "One of our senior guys that was one of the big skeptics has actually been using Claude as his sounding board," Grenstad noted. He observed that for some developers, querying AI has enhanced their solutions, demonstrating a significant change in mindset.
Nonetheless, some team members have raised concerns over losing control as release speeds increase. Before the transition, Cubeia typically executed four to six releases per month with around 50 support tickets raised. In May, following the initial changes, the company produced only five releases but faced 86 tickets. June saw a jump to 10 releases accompanied by a rise to 96 tickets.
In light of this rapid evolution, Grenstad is considering how to integrate the quality assurance (QA) function more effectively into the development workflow. He explained that while QA is technically part of the development team, the traditional handover process has not adapted well to the new, quicker AI-driven developments. He advocated for a structure where developers carry responsibilities into the production phase, emphasizing that QA should be included in the planning process.
Overall, Grenstad believes it is crucial to support teams through this transition, recognizing that many developers are committed to the goal of adopting AI-assisted coding. "My tech leads are doing a tremendous job with those that are struggling a bit, and you know, supporting them, helping them," he stated. He concluded, "I think everybody agrees now that this is the way we should be working. This is the future. I don't think they are thinking this is the wrong way to go," even as team members, including himself, maintain a degree of skepticism about ensuring consistent quality amid the new pace.
