If there's one thing we'd all like to see automated, it's tedious work processes that take a lot of time and effort, especially when it comes to finances. Fortunately, there are now AI solutions for such processes that can make the effort much easier for us. One of them is being developed by the Berlin-based company Deep Neuron Lab, which was founded out of TU Berlin in 2019. We talk to CEO Andreas Schindler about Deep Neuron Lab's "Natural Language Processing", what fascinates him so much about artificial intelligence and where to find the most refreshing ice cream in Berlin.
Hello Mr. Schindler, thank you for the interview! Where do you prefer to cool off on hot summer days in beautiful Berlin? Do you have any insider tips?
We currently office space at AI Campus Berlin – fortunately it's air-conditioned. But if it does get too warm, I stop by JONES ice cream in Schöneberg.
You originally come from the insurance industry. What made you want to move into the field of artificial intelligence?
I was a student in conjunction with my employment and investigated the financial markets with AI in the course of my master's thesis. After that, the subject just wouldn't let me go. I have always been interested in many different topics – thus the insurance industry as well. Here I was able to deal with different topics without having to change fields. From drought risks in Germany, to the insurability of offshore wind farms, to global cyber risks and their consequences.
Two things in particular fascinate me about AI:
1) I am convinced that artificial intelligence is the most important technology that humans can work on. We are the first generation in the history of mankind that has the chance to develop an AI that resembles humans and even surpasses them in certain abilities.
2) The field is evolving so fast that I can hardly keep up. And that fascinates me. In five years, we will see leaps in development for which we used to have to wait 30 or 50 years – and development is still accelerating.
Even though I don't work directly in AI research, in my job I have the opportunity to be exposed to the latest research findings on a daily basis.
Your AI startup was founded out of TU Berlin in 2019. How did you come up with the idea for Deep Neuron Lab (DNL) and how did your early days proceed before and after its founding?
There was not the one moment when the idea for DNL came to us. The basis for the foundation was the motivation to work with technology and to solve problems with it. In the beginning, we did a few individual projects with clients to find out what kind of problems they have and how we can solve them. Our customers at the beginning were mainly banks, insurance companies, auditors, etc. – we observed that they all had challenges, especially in dealing with financial reports. This is how we came to focus on automating the analysis of financial reports and offering product solutions for this purpose.
The founding was (as with all founders I think) a highly dynamic and exciting time Yet we founded DNL at the start-up centre of the Technische Universität Berlin – the Centre for Entrepreneurship. They helped us a great deal in overcoming the initial hurdles.
Deep Neuron Lab (DNL) offers technology that uses AI to speed up and automate work processes. How exactly does your artificial intelligence work? What can be understood by "natural language processing" that you use in your solutions?
We extract and structure data from financial reports. These are not always depicted in the same form. That's why we need solutions that can handle different forms of presentation. For this purpose, we use AI methods from the field of "image processing" (IP) and "natural language processing" (NLP). Let's take a table, for example. In order to understand the content of this, we need to understand the texts in them on the one hand – for this, we use NLP. On the other hand, we also need to understand how they relate to each other – this is where IP comes in. Together, this allows us to understand the contents of a table and convert it into a machine-processable format.
You have arrived at your motto – focussing on the humans, and not the process. Where does AI find its place in this – between process and human, or as part of either?
Often, the expectation is that AI will fully automate processes and thus replace humans. We deal with processes that are too complex for full automation and therefore require collaboration between humans and AI. AI supports humans through automation – humans monitor AI. This puts humans at the centre of attention for us: Our applications are designed to support them, not replace them. AI is supposed to make work easier and, above all, assist with the monotonously time-consuming extraction of data from documents, so that humans can concentrate on the essentials.
Your focus is on the financial sector. Why is your solution particularly suitable in this area? Does AI have more uses in the FinTech world?
I have always been fascinated by the financial industry, which is why I studied and worked in the field. Additionally, there is a great need for document processing in the financial industry. A tremendous amount of essential data and information is published in confusing documents, which significantly limits transparency with regard to them. Cases like Wirecard show that just because information is known, it is not necessarily processed and understood by everyone. We see a reason for this in the publication in document form. In order to obtain information from documents, it used to be necessary to employ people who laboriously searched for, understood, structured and extracted the information. We facilitate access to such information, in order to make it easier or possible to understand.
Why should someone establish a company in Berlin? In your eyes, what makes the city particularly attractive for young AI companies?
For us, the number one reason was, of course, that I am a Berliner, but Berlin is also a perfect location from an objective point of view. A company is the sum of its employees. Berlin is highly attractive for talents from all over the world, making it easier to attract excellent employees. In Berlin itself, young talents are continuously trained in the award-winning universities. In addition, the start-up scene in Berlin is very well networked and thus enables an exchange among founders. Especially at the beginning, such exchange is highly valuable and has helped us.
What are the next steps for Deep Neuron Lab (DNL)? What are your plans for the near future?
We are in an exciting phase right now. We closed our seed funding earlier this year and have since been able to build a great team. We are currently entering the test phase of our products among our customers, which we will launch by the end of the year. At the same time, we are being approached by various parties about our solutions, and we have just been able to acquire a large auditing firm as a customer.