AI is changing software development: a look behind the scenes of BBT’s AI accelerator experience in Berlin

Artificial intelligence is developing rapidly and is already changing how software is created. A company like BBT, which develops innovative software solutions for the insurance industry, this development offers enormous potential. At the same time, it raises new questions: how is AI changing the role of developers? What opportunities will this create for customers? And what responsibilities will still rest with humans?

To investigate precisely these questions, BBT took part in the Volaris Group’s AI Accelerator in Berlin. During an intensive week-long program, more than 140 participants from 19 companies in the software development, product management and leadership spaces worked together to test practical use cases for artificial intelligence, learn about new ways of working and turn ideas into concrete prototypes.

For Matthias Hirschi, a software developer at BBT, one thing in particular was paramount:

“I wanted to learn something new and deliberately think outside the box.”

Turning theory into practice

The AI Accelerator was not a traditional training program. Instead, the focus was on practical collaboration on real-world issues. It became clear right from the initial discussions that despite the different products and markets, the participants face many common challenges. Companies that develop complex software solutions in particular are confronted with similar questions: How can artificial intelligence be meaningfully integrated into existing processes? Which tasks can be automated in the future? And how can AI help develop digital solutions faster and in a way that is more closely aligned with customers’ needs?

This question is particularly relevant for BBT. The development of insurance software operates in an environment characterized by high technical requirements, complex processes and clear quality standards. AI can provide support here by speeding up development processes and creating new opportunities for innovation, without replacing the necessary expert control.

Following the official kick-off, the participants worked intensively on these issues within their respective tracks. Professional exchanges, collaborative work phases and regular discussions with experts and technology partners took place in turn throughout the week. This created an environment in which new ideas could not only be discussed, but also tested and developed straightaway.

In the Developer Track, one central idea quickly came to the fore: good software doesn’t start with writing code, but with understanding the actual problem. Before a line is programmed, requirements must be clearly formulated, relationships understood and the necessary context created. This is precisely where AI can already make a key contribution.

The moment that inspired a rethink

As the week progressed, the focus shifted increasingly to practical implementation. Initial ideas led to specifications, and the BBT team used these to develop concrete prototypes with the help of modern AI tools. During the intensive build phase, it became strikingly clear just how closely humans and AI are already able to work together.

A particularly impressive moment for Matthias was when a feature was implemented almost independently by AI.

“The basic feature was already there. We mainly had to add to the technical context, check the result and make minor corrections.”

This experience made it clear how much software development can change. In future, the focus will not be on every single line of code, but on the ability to describe problems precisely, formulate the right requirements and ensure the quality of the results.

AI as a sparring partner

One of the most important findings from Berlin is that AI is not replacing developers – it is changing their role.

Instead of taking on routine tasks, development teams will be able to focus more on architecture, technical issues and analyzing solutions. AI helps with this as an intelligent assistant that makes suggestions, creates code or prepares tests.

Matthias describes this collaboration with a fitting comparison:

“AI is like an invisible partner by your side. It challenges you, tests your ideas and can write code for you. But you still have to make the decisions and take responsibility for them.”

This is particularly important in developing software for insurers. Technical requirements, data protection, regulatory stipulations and quality standards cannot be fully automated. They require experience, critical thinking and human judgment.

Opportunities for BBT

The experience gathered from the AI Accelerator program demonstrates the potential that modern AI tools offer for BBT. New functions and product ideas can be visualized more quickly and implemented as initial prototypes. This can offer a decisive advantage, especially in the development of software for insurers and insurance companies, as requirements and new ideas can be identified at an early stage, tested and further developed together with customers.

Instead of spending a long time working towards a fully developed solution, initial prototypes enable a faster exchange of ideas: which feature actually adds value? Which processes can be simplified? And which requirements need to be taken into account to achieve an optimum solution?

This iterative approach helps BBT to drive innovation in a targeted manner and develop prototypes that are not only technically impressive, but also bring about lasting improvements to customers’ daily working lives. Models can therefore be discussed, further developed and specifically improved at an early stage in collaboration with customers.

However, the week in Berlin also demonstrated that the successful use of AI goes far beyond technology. The cultural change within a company is at least as important. New tools alone are not enough – what matters is how teams work with them, how results are scrutinized and how responsibility is taken. Critical thinking, technical expertise and a shared understanding of quality remain indispensable, even in AI-supported software development.

This approach was a recurring theme throughout the entire AI Accelerator and was repeatedly distilled into a simple guiding principle:

“We are automating work steps – not responsibility.”