Artificial intelligence has become established in many companies. The challenge is no longer simply to identify initial use cases, but to develop solutions that prove their value in day-to-day work over the long term and create genuine added value. In this interview, Civan Erbay from SOTER.studio explains what matters when developing and implementing AI and automation solutions.
Many companies are currently trialling individual AI applications. Under what conditions can these provide lasting structural relief for processes and employees?
Put simply, structural relief means that a workflow runs from start to finish without anyone having to initiate the intermediate steps. This can be seen, for example, when processes run independently of individual employees, their failure is immediately noticeable in day-to-day operations, and the end result is a finished output that requires no extensive preparatory or follow-up work.
One example: Many companies now record their video calls, but the summaries end up in a folder that is rarely used. Structural relief is achieved when these recordings are automatically used to update data records, create follow-up tasks and send the documents discussed to the relevant employees.
You develop AI and automation solutions for various areas of application. What determines whether an application is accepted and used permanently in day-to-day work?
AI automations are highly streamlined applications that emerge from employeesâ day-to-day work and follow a kind of pull principle. Instead of a complex software solution that is typically imposed from above, employees themselves identify what would help them. Nobody knows the processes and edge cases as well as the person who deals with them every day. The application is accepted because nobody rejects a tool they have helped to design.
How do you identify suitable areas of application when a company wants to use AI but cannot yet specify a concrete use case?
The right questions reliably reveal potential areas of application: Where is data copied from one system to another? Which task would you hand over immediately if you could? What does the team keep putting off every week?
Such patterns exist across companies. Some problems are so widespread that effective tools have long been available to address them. In such cases, our recommendation is to introduce these solutions rather than build something new. Where a bespoke solution is needed, however, we embark on a product discovery process together and quickly develop a clickable prototype that serves as the basis for the final workflow.
What motivates you to take on AI and digitalisation projects that also have a societal impact?
Anyone engaged in business always operates within a societal context, and with this comes a responsibility that extends beyond their own business model. In our view, entrepreneurs also have a duty to address issues such as social cohesion, climate change and how society responds to technological change. We also believe that cooperation as a principle takes us further than competition, which is all too often regarded as the only supposed source of innovation. There are now numerous exciting initiatives addressing these issues, including impact.cologne, Startups for Tomorrow and, in the field of AI, Civic Coding. We would like to see more companies commit to engagement beyond their own business models.
Data protection, GDPR compliance and the EU AI Act are integral parts of your work. How are these requirements incorporated into development from the outset?
Data protection should be a key architectural decision for any company that processes customer data or uses AI in any form. Which data flows where? Which model and hosting environment are used? In practice, this means certified infrastructure in Germany, no transfers to third countries, GDPR compliance and ISO 27001 certification.
In the field of AI, this includes open-source models operated on German or at least European infrastructure. The monopolisation of AI infrastructure outside Europe creates a dependency that can still be avoided today. A European counterweight needs customers, and anyone who chooses European providers helps to ensure that this counterweight can emerge.
