Experts: Kevin Kuhn (Gopf)
Gopf, a Lucerne-based start-up, has developed software to source and analyse market data. The software uses technologies such as automated web content collection and the identification of patterns and market movements using large language models. This allows Gopf to create an AI bot that monitors competitors, suppliers and customers, identifies trends and summarises information.
Picture: Chor Tsang (Unsplash)
It seems paradoxical. We often hear about how overwhelming, uncertain, unknowable or even confusing things are these days – despite information and knowledge being more easily accessible than ever before. But that’s precisely where the problem lies: too much noise, too little meaning. The static makes it difficult to identify signals, to distinguish between what is important and what is not. We’re flooded with trivial stories that crowd out the big picture.
To make sound decisions, we need factual foundations. Everywhere we turn, people lack the attention, time and money to extract relevant information from the noise and mass of data and to turn it into manageable packages. That’s where Gopf comes in. An AI bot searches for relevant data, processes it, automatically summarises it and generates reports. Why Gopf? How did Kevin Kuhn and his colleagues come to develop a piece of software like this?
Kevin Kuhn (Founder and CEO of Gopf) recounts how, in 2020, he addressed the question of clustering information and displaying it in 3D space. However, he quickly realised that such visualisations are trickier to read than he thought.
This led him to grapple with the question of how to find meaning in vast volumes of data. His amazement at the possibilities offered by analysing large volumes of text still resonates today in the name of the company: ‘Gopf!’ – a Swiss German word that is typically used to express astonishment.
Closely linked to the company’s history is the question of how the continuing education landscape is structured across universities. Back in 2021, Mr Kuhn’s goal was not only to create an overview of the content taught by the countless Certificate of Advanced Studies (CAS) and Master of Advanced Studies (MAS) programmes, but also to identify the various teaching formats.
He talks enthusiastically about how he and his colleagues built a program back then that collects this data on the continuing education landscape from the internet, downloads it and then analyses it. Their analysis used common machine language processing methods (known as semantic similarity maps) to map the similarities between course descriptions. Once the various course contents and formats have been sorted according to similarity, it is possible to identify clusters and form relevant groups.
In 2023, Gopf obtained the commercial register data of 653,000 Swiss companies. However, the purpose of the company recorded in the commercial register gives little indication of the actual activities of the organisation in question. An automated internet search helps. Every single one of these 653,000 companies is searched for online, and the most relevant results are downloaded and evaluated. This process involves similar methods to the ones that Kevin Kuhn and his colleagues used at that time to understand the continuing education landscape – simply with the aim of collecting, evaluating and clustering data from the business world.
The software is able to provide information about industries, their products and the challenges they face by trawling through the websites of relevant companies and corresponding trade journals. The tool thus provides insights into entire sectors, identifies market trends and monitors suppliers and customers. The tool is an AI system that, similar to Gemini or ChatGPT, analyses text data using large language models. Users can specify search terms, questions, sources and the type of output.
One segment where Gopf sees potential is the mechanical engineering industry. This sector is plagued by shrinking margins and increasingly strong competition from Asia. Incremental innovations – i.e. the gradual improvement of products – are on the rise, while the proportion of disruptive innovations has been declining for years. And this at a time when it is clear how important market innovations are and that they not only offer tremendous entrepreneurial potential, but also have enormous economic significance (see SATW study on Innovative strength).
Kevin Kuhn emphasises how important it is to well and truly know the market and your own data, and to understand what your competitors are doing. Developing an understanding of customers and suppliers can help with making the right decisions at the right moment. Kuhn thinks big. He envisions being able to interact with the AI bot directly in Microsoft Teams or Zoom.
‘Such programs are technically demanding and require considerable expertise,’ commented Kevin Kuhn, ‘but that’s not where the major difficulties lie, as the individual components have essentially already been developed. What’s new is the how they’re linked and the services derived from this approach.’
While the difficulties encountered in building Gopf are more subtle, they are also weightier. Initially, developing systems like Gopf takes time and money – the costs are in the mid six-figure range. The result is a piece of software that can be used for many different things. And that is precisely where the major challenge lies: explaining to interested individuals what the system is capable of, what it can be used for and how it can be deployed in the most profitable way – without overwhelming them with the sheer abundance of possibilities.
Telling stories remains a challenge, even in the age of machine language processing. What will change is how these stories come about, and what data they are based on. And that’s where it’s important to use all the tools available.
