AI and R&D: Reimagining how HR solutions are designed

For more than two years, the Research & Development teams at Sopra HR Software have been exploring the practical potential of artificial intelligence. From the outset, our goal was not to follow a passing trend, but to understand what AI could genuinely contribute to our work as an HR software provider operating across complex, highly regulated, multi-country environments with different technologies.  

This initiative began with targeted experiments involving volunteer employees across different product lines, technologies, and geographies. The objective was to observe, test, measure, and above all identify use cases capable of delivering tangible value for both our teams and our customers. 

Today, we are entering a new phase. AI is no longer simply an area of exploration; it is gradually becoming a driver of transformation in the way we develop, design, and enhance our HR solutions. 

From experimentation to deployment 

The first use cases focused on two areas: automated regression testing and support for software development. Test generation, code analysis, code completion and technical documentation all helped identify the first opportunities to improve efficiency, quality, and reliability. 

However, integrating AI into an R&Dorganization like ours cannot be reduced to simply adopting standard tools. Our products use a wide range of technologies, operate in different regulatory environments and cover demanding areas such as payroll, time and attendance and HR administration. 

We therefore chose to proceed gradually. As agentic AI technologiesmatured, some of our most promising experiments began to deliver consistently strong results. What worked only partially in the past has now become reliable enough to be deployed within specific areas. 

The benefits are already visible in some projects: shorter production cycles, higher quality and a greater ability to shorten the time between an idea and its delivery to the market. However, this scaling-up process must remain carefully managed. In a complex organization, gains must be measured, consolidated, and adapted to the realities of each product line. 

The developer as orchestrator 

This transformation is not limited to tools. It is fundamentally reshaping shaping roles within R&D. 

The role of the developer is evolving. It is no longer simply about writing code line by line. Value is shifting toward the ability to clearly define objectives, guide AI systems, assess their output, make corrections, exercise judgment, and ensure that the expected standards are met. 

In this new model, developers are gradually becoming the conductors of teams of specialized agents. They draw on complementary areas of expertise, coordinate them around a common objective, and ensure that the resulting output meets our standards for quality, architecture, performance, security, and accessibility. 

This shift does not diminish the importance of human expertise. On the contrary, it makes it more strategic. As AI takes over certain execution-oriented tasks, people need to define their intent more precisely, exercise closer oversight and take greater responsibility for design decisions. 

An agent factory tailored to HR business needs 

One of the key lessons from our experiments is that AI must be contextualized to be truly useful. 

At Sopra HR Software, we do not operate in a generic environment. We develop HR solutions that must meet stringent requirements relating to payroll, HR administration, regulatory compliance, multi-country deployment, data security, accessibility, performance and continuity of service. 

For AI to be relevant, it must reflect this business reality. With this in mind, we have begun developing a platform of specialized agents. Some provide functional expertise in areas such as payroll, HR administration and time and attendance. Others focus on more technical domains, including specific programming languages, software quality, and code analysis. 

The strength of this approach lies in the ability of these agents to work together. The goal is not to assemble a collection of isolated assistants, but to have them contribute collectively toward a shared objective within a consistent and controlled framework. 

To maintain consistency, we use a common set of rules in the instructions given to AI. These instructions help guide output in line with our architectural, development, performance, security, and accessibility requirements. They are continuously refined, adjusted, corrected, and improved based on real-world usage, following a process of continuous learning. 

Managing usage, costs and environmental impact 

The deployment of AI also raises new operational challenges. Adoption across teams has been strong, which is encouraging. However, it also inevitably increases the use of tokens, with economic and environmental implications that must be taken into account. 

This is now one of our key priorities: how can we optimize that usage while maintaining a high level of output quality? How can we continue to expand valuable use cases while keeping costs and environmental impact under control? 

Here again, the challenge is to strike the right balance. AI should be used where it creates genuine value, within a framework that is responsible, measurable and sustainable. 

People are the real challenge 

Since the beginning of this transformation, one thing has become clear: the greatest complexity lies not in the technology itself, but in the human dimension. 

We need to explain what AI is, what it can and cannot do, where it creates value and how it should be used. We must also support our teams, because people do not all approach these new tools in the same way. 

Some employees embraced AI immediately. Others expressed greater caution or concern. These differences are perfectly natural. They are part of any profound transformation of the way we work. 

Our responsibility is therefore to support the adoption process by providing clear guidance, sharing good practices, creating the conditions for responsible use, and gradually developing new ways of working. There is not yet universal playbook for working with AI. We are creating it collectively as we go. 

AI is evolving at an unprecedented pace. In this context, staying informed is not a secondary concern; it is a prerequisite for success. Technology watch and knowledge sharing must become a collective habit, with teams dedicating time to monitoring developments, comparing experiences, testing new approaches, and rapidly sharing valuable insights. Our ability to learn together will need to evolve at the same pace as AI itself. 

AI will transform our processes, but it will only deliver its full potential if the transformation works both ways: integrating AI into the way we work, while adapting our organization, roles and processes to make the most of its capabilities. This is not about simply adding another layer of technology to existing practices. It requires rethinking certain workflows, responsibilities, and modes of collaboration so that AI can deliver its full value. 

This human dimension is essential because AI is not only changing how software is developed. it is also transforming how we think about work, collaboration, and the creation of value. 

Towards a new user experience 

AI is not only changing R&D internally It is also opening up new possibilities in the very design of HR solutions. 

For many years, software has been built around structured user journeys: menus, screens, buttons and a sequence of steps. Users had to navigate through the application to achieve their objectives. 

With AI, a different approach is becoming possible. Tomorrow's HR solutions will be driven more by the user's intent. Users will express a request, ask a question, or describe a need, and the solution will suggest the most appropriate way forward. 

This marks a major shift in how people interact with technology. For HR professionals and employees alike, it paves the way for experiences that are more intuitive, more contextual, and more focused on objectives than on predefined workflows. 

It is also changing the way we design products. R&D teams must now think in terms of solutions that can understand intent, guide users, and dynamically generate relevant responses. 

Shaping the future of work together 

AI is driving a profound transformation in the way we work. It is challenging long-established habits, reshaping roles, and prompting everyone to rethink where they create value. 

This transformation is demanding. It calls for new skills, methods and working habits, together with constant vigilance over the quality of the results produced. 

But it is also exciting. We are actively shaping the way we will work tomorrow: how we organize teams, design products, support users, and enhance our solutions. 

In this context, the role of the software engineer becomes clear. Software engineering is not just about coding or applying established methods. It is about finding solutions, understanding complex requirements, orchestrating expertise, and ensuring that technology serves real-world needs.  

This is precisely where AI brings value in R&D: not by replacing human expertise, but by allowing experts to focus on what matters most: analysis, decision-making, quality, and innovation.