In many companies, artificial intelligence raises as many expectations as it does questions. For a small-to-medium-sized enterprise (SME) in French-speaking Switzerland like SwissCaution, however, the challenge is not simply to follow a technological trend, but to identify concrete applications capable of improving internal processes, employees\u2019 day-to-day work, and the quality of service provided to customers. The company has therefore moved forward in stages, starting with the most repetitive tasks and needs, before gradually integrating certain tools into its operations. This approach\u2014based on experimentation, close collaboration between business and development teams, and a strong capacity for adaptation\u2014illustrates how an agile organization can embrace AI without compromising its coherence. To learn more, Romain Paci, Product Owner, and Nayel Maharmeh, developer, detail the gradual implementation of AI among SwissCaution\u2019s employees and customers.<\/p>\n\n\n\n
How did SwissCaution begin integrating artificial intelligence into its operations? Nayel Maharmeh:<\/strong> This initial phase also helped us establish a methodology. At SwissCaution, we moved forward step by step, without trying to roll out large-scale systems right away. This incremental approach allowed us to test prototypes, put them to the test in real-world conditions, and then scale them up once they delivered real value to managers. Ultimately, agility and AI work in synergy: with agility, we learn faster, and with AI, we produce better and more quickly.<\/p>\n\n\n\n What were the first use cases implemented? R. P.:<\/strong> We\u2019ve also worked on accounting, using solutions capable of sorting and processing invoices, particularly through data extraction. Not long ago, this type of automation seemed almost out of reach or reserved for very large organizations. Today, these technical projects have become a realistic possibility. This opens up new opportunities, including for smaller companies that want to improve efficiency without overburdening their operations.<\/p>\n\n\n\n What principles guide your approach to managing these projects? And regarding the customer experience, what developments have been undertaken? R. P.:<\/strong> In practical terms, this means that certain basic checks\u2014such as verifying identification documents or ensuring consistency between different elements provided in the application\u2014can be performed automatically. This speeds up processing while making the onboarding experience smoother for the customer. Behind this apparent simplicity lies a whole process of automation that both secures and streamlines the process.<\/p>\n\n\n\n What effects are you seeing internally today? N. M.:<\/strong> In reality, the main challenge is also to free up managers\u2019 time for tasks with higher added value. When we reduce the proportion of repetitive tasks, we improve not only productivity but also the work environment. The workload becomes less burdensome; employees are more relaxed, more available, and more focused on tasks that truly require analysis, judgment, or human interaction. That\u2019s where a virtuous cycle begins.<\/p>\n\n\n\n Does AI replace certain human skills, or does it rather complement them? R. P.:<\/strong> AI enhances capabilities; it does not eliminate the need for critical thinking. In our context, its primary purpose is to allow teams to focus on what requires experience, nuance, and responsibility. That\u2019s also why we\u2019re moving forward cautiously and gradually: we need to learn quickly, but without losing sight of our priorities.<\/p>\n\n\n\n Is the agility of an SME an advantage in this transformation? N. M.:<\/strong> This ability to stay close to the ground is invaluable. We can identify a need, devise a solution, experiment, and then make adjustments. Agility doesn\u2019t mean improvising\u2014it means learning faster. It\u2019s a real advantage in a field like AI, where applications evolve very quickly and where you have to be able to adapt continuously.<\/p>\n\n\n\n How do you see the next steps? N. M.:<\/strong> The next step could be a more executive form of AI, capable not only of assisting but also of triggering certain actions or initiating processes based on clearly defined rules. For now, we\u2019re still operating within an assistive framework. In the future, certain components could become more autonomous. But this evolution must remain under control, consistent with our needs, and aligned with our quality standards.<\/p>\n\n\n\n As issues of technological sovereignty and environmental impact gain importance, what choices do you intend to prioritize? R. P.:<\/strong> This aspect is becoming increasingly important in how we approach innovation. For a company like ours, it\u2019s not just about adopting high-performance tools, but also about choosing solutions that align with our values and market expectations. Here again, agility means staying attuned to developments without locking in technological choices too early.<\/p>\n\n\n\n
Romain Paci:<\/strong> AI was able to be integrated largely as a result of preliminary work aimed at establishing a modern IT architecture, which facilitated its modular implementation and gave us a competitive advantage. So AI didn\u2019t come to us as some big theoretical project decided from above. It naturally became part of our agile way of working. We first wanted to test, understand, and identify the most relevant use cases. About two years ago, we launched small internal projects to challenge the teams and identify the most tedious tasks\u2014those that added no real value but took up a lot of time. We quickly realized that simply using generic conversational tools wasn\u2019t enough: the key was to integrate AI into business processes, where it could have a tangible impact.<\/p>\n\n\n\n
N. M.:<\/strong> We started by automating the sorting of incoming emails by language and email category. For a highly customer-focused company like ours, this was a top priority, as we receive a large volume of messages that must be processed quickly and accurately. Automation has allowed us to better manage these workflows and reduce the burden of certain repetitive tasks.<\/p>\n\n\n\n
N. M.:<\/strong> The close collaboration between business teams and developers. Our technical teams work largely in the same open-plan office as our other employees. This accessibility makes a big difference: communication is more direct, feedback is faster, and adjustments are smoother. In terms of team cohesion and collaboration, this is a major asset. It allows us to iterate quickly and ground our solutions in operational reality, while also giving us the ability to develop our own tools\u2014or even our own AI models\u2014in-house.<\/p>\n\n\n\n
N. M.:<\/strong> The goal was to make thecustomer<\/strong>journey as simple and seamless as possible. For a customer, registration should be quick and involve as few unnecessary hurdles as possible. So we\u2019ve automated a number of simple checks related to the data and documents submitted. For example, AI can analyze the information received, compare it with what was entered, and verify that everything is consistent.<\/p>\n\n\n\n
R. P.:<\/strong> The first effect is, of course, significant time savings in the long term. For certain processes\u2014particularly in handling acquisitions or requests\u2014we\u2019ve been able to save a great deal of time. This allows us to process more cases and, potentially, serve a larger client base. But the benefit isn\u2019t measured solely in terms of volume.<\/p>\n\n\n\n
N. M.:<\/strong> Today, we clearly view AI as an assistant. It allows us to work faster and automate certain checks or operations, but it does not replace human judgment. It remains essential to know what is relevant, what needs to be validated, and what deserves special attention. The real strength lies in the synergy between humans and technology.<\/p>\n\n\n\n
R. P.:<\/strong> Yes, very clearly. An SME can often make decisions faster, test more quickly, and make corrections without excessive red tape. At SwissCaution, this agility allows us to develop a prototype, present it to the committee, and then move forward if the results are convincing. We don\u2019t need to go through endless cycles before taking action.<\/p>\n\n\n\n
R. P.:<\/strong> We\u2019re currently in a stabilization and fine-tuning phase. We\u2019ve already implemented tools that are proving their worth, but we still want to gauge their potential, understand their limitations, and see how far they can take us. The challenge is also to stay on the lookout, as models and solutions are evolving at an impressive pace.<\/p>\n\n\n\n
N. M.:<\/strong> We are closely examining the most sovereign and responsible solutions possible. The idea is not to be dependent by default on the very large international technology companies if credible alternatives exist. As a company regulated by FINMA, we must also comply with its regulations and requirements. The market is evolving rapidly, and we\u2019re seeing the emergence of more local, more ethical offerings that have already caught our full attention.<\/p>\n\n\n\n