Cubeia, a Swedish software company, is progressing through its journey towards AI-assisted development. Having transitioned from an informal adoption of AI by developers to a more organized AI-driven development process, the company is now confronted with a critical question: how to effectively operate when coding is no longer the primary limitation.
Stefan Grenstad, Cubeia’s COO, affirms, “We’re doing this 100%,” regarding their commitment to AI-integrated development.
The company has refocused on leveraging the productivity boost that AI provides. While just six months ago the initiative revolved around substituting human-generated code with AI-generated code, it has since broadened to encompass a revamped development pipeline, a redefined role for developers and quality assurance (QA), restructured teams, and an evolving engagement with customers.
In Cubeia’s initial phase, developers enjoyed an open approach to AI utilization. The second phase introduced a structured framework where all engineers operated under the same AI-driven pipeline. This evolution required Cubeia to tackle challenges concerning quality and reliability while ensuring employees adapted to the new workflows. Grenstad indicates that this adjustment phase is largely complete.
In the early part of the year, during a hybrid operational phase, Cubeia resolved 259 issues. Following the implementation of the AI-driven process, that number surged to 421, marking a 62% increase. The volume of larger projects escalated from 17 to 58. Grenstad notes, "So the answer is yes: moving to AI-driven development has increased our output tremendously."
With coding taking less time, the focus has shifted toward understanding and creating business value. Grenstad mentions, “Because we’re not spending as much time coding, we’re spending more time on the business: talking about value and understanding the domain.” Developers are now expected to consider the rationale behind their builds and the value they generate.
Cubeia's head of marketing, Paul Crisp, emphasizes the importance of measuring success, stating, “If the objective is 10% more traffic, for example, how do we measure it?” He illustrates the process: establishing a baseline, implementing changes, then measuring the effect to assess whether the goal was met.
As customer involvement grows, Cubeia is actively engaging clients to determine their needs regarding AI capabilities. Grenstad elaborates, “We [recently] started asking, 'What does this mean for the customer? What can the customer actually do with this?'” He shares examples of clients wanting to harness Cubeia’s APIs to enhance their own gaming platforms.
In a notable case, a customer independently created a casino landing page and sought to integrate it with Cubeia’s system. Another client aimed to utilize Cubeia’s player account management system for bespoke features like personalized bonuses. This shift signifies that Cubeia is less about merely providing AI tools and more about offering a platform that integrates with customer-driven AI infrastructure.
The company is currently piloting a project centered on the landing-page integration and anticipates being able to accommodate customers with products developed through their own AI tools. Grenstad reflects on the organization’s evolving role, stating, “We’ve realized we’re going to be part of it.”
In light of the increased productivity, Cubeia is restructuring its development teams. They are experimenting with different configurations, including a rapid-response team for smaller client requests alongside another team focusing on larger, strategic projects. Current teams consist of five to eight members, but Grenstad envisions smaller teams—perhaps two or three individuals—could be more effective.
Despite the enhanced capacity, bottlenecks persist. With multiple AI agents generating parallel work streams, managing the review process remains a challenge. Grenstad notes, “The bottleneck becomes the person reviewing everything.” While critical parts of the system still undergo thorough code reviews, not all code is scrutinized to the same extent as before.
Grenstad's perspective on recruitment is also shifting. In the past, technical skills such as Java proficiency were prioritized, but with the changing landscape, he recognizes that domain knowledge is increasingly valuable. He states, “I’d rather take someone who’s good in the domain but doesn’t know any Java.” This raises questions about the pathways for new developers entering the industry, challenging the traditional progression from junior roles to senior positions.
The company is redefining the development process, shifting from a sequence where developers code and then pass their work to QA. In the updated model, QAs are involved throughout the process; product engineers now consider automation and manual testing from the project’s planning stages. The responsibility stretches from problem identification to execution and outcome evaluation.
Despite these advances, some team members long for the coding aspects they once enjoyed. Grenstad acknowledges their struggle, noting, “We still have people who loved the coding part – solving problems with code and writing beautiful code. They’re struggling with the change.”
Strategically, Cubeia is also navigating reliance on AI providers, particularly the implications of price increases or shifts in service quality. Grenstad expresses confidence in their ability to adapt to these uncertainties, stating, “We’re confident that the question marks will be solved as we work through them.”
Cubeia initially aimed to eliminate coding as a hurdle for software development. This transformative journey revealed that coding was just one element in a larger system, and as new bottlenecks form, the focus has shifted toward domain expertise, product development, quality assurance, prioritization, and client needs. As the spotlight moves away from technology itself, the emphasis is now on how Cubeia will utilize this technology moving forward.
