"The performance of AI is already very strong today and improving at a rapid pace."

It's easy to lose track of what's going on in a store - and not just in the run-up to Christmas. For the companies behind the stores in particular, it is of great importance not only to see potential shoplifting via cameras, but also to recognize, among other things, when goods are sold out and need to be restocked. With the support of Artificial Intelligence, cameras can do even more here - this is what the Berlin-based company Signatrix thought and has been developing an AI camera solution for retail since 2017. We talk to CEO and founder Philipp Müller about the origin of the idea, the benefits of artificial intelligence and how to keep a cool head while shopping in the run-up to Christmas.

Hello Mr. Müller, thank you for taking the time for the interview! Since you're marketing a solution for retailers: What is your personal strategy for the shopping rush in the run-up to Christmas?

You're probably most relaxed if you've already taken care of most of it in the fall, but then you never manage to do that. If you know which stores are well organized and customer-friendly, it makes shopping in the run-up to Christmas much more convenient, I think. And a bit of shopping hustle and bustle can be fun, too.

Signatrix is developing an artificial intelligence solution for the retail industry. What exactly does this solution look like and how did you come up with the idea for this product?

When we founded Signatrix in 2017, we noticed that existing sensor technology such as camera infrastructure in supermarkets and other retail stores produces large amounts of relevant data that goes largely unused. However, this data contains information that can be enormously useful to businesses by providing a comprehensive picture of what is happening in a store. Our AI software is able to analyze the camera data and provide retail companies with such information. This includes, for example, statistical analysis of aggregated data - for example, to learn what customers are interested in during their shopping trip - or real-time notifications when goods are sold out or unauthorized behavior, such as shoplifting, is detected.  Because each store has different requirements, we provide Signatrix AI Suite, a platform through which an individual selection of computer vision apps can be put together for each store. These can be very different solutions, from loss prevention to customer analytics - anything that makes the store more secure, efficient or customer-focused. After the one-time installation of the hardware, you can then use an intuitive dashboard to test out further apps at the touch of a button and implement them in no time at all. This means you can save yourself the time-consuming evaluation and implementation processes that would otherwise be involved in introducing every new AI solution, and make the store smarter and smarter at your own pace without any complications.

What advantage do cameras with your artificial intelligence have over conventional surveillance cameras?

A conventional camera doesn't notice any difference between what it's recording and what it's not. Whether it's a theft, for example, or simply a customer shopping, that always has to be determined by a human watching the video stream. This is something that store detectives like to do these days, but they also have a hard time effectively reviewing dozens of camera streams in real time for days on end. In addition, camera technology can be useful for many other applications that have nothing to do with shoplifting. Application in these contexts is still in its infancy in retail and requires additional AI technology.

You pay particular attention to data protection when using your solution in cameras. Was this part of your concept from the start or did you respond to the needs of your customers here? Was the adaptation to the DSGVO 2018 a major effort?

It was clear from the beginning that camera-based solutions cannot function or scale without thinking about data protection aspects in the software and solution design. In particular, it was important to us to use the existing infrastructure of sensor technology from the outset, as appropriate checks had already been carried out here. Due to their business model and the associated margin structure, retailers have a particularly high risk as a result of the GDPR due to the structuring of possible penalties under the GDPR. For this reason, we have gone through extensive and holistic auditing procedures, particularly with our larger customers. Independently of this, however, we already consulted experts from renowned law firms during our own examination of the topic as part of product development in order to optimally implement the corresponding requirements from the very beginning. Accordingly, the GDPR did not mean a major change for us.

In times of COVID-19, many measures are implemented to keep interpersonal contact to a minimum. Can cameras with your AI software replace on-site security personnel? Or will they be used more as a supplement to existing personnel?

In our opinion, the software can rarely replace staff entirely. It automatically detects incidents and notifies the staff in real time - but in most cases, humans still have to intervene and take action. Artificial intelligence, however, makes their work much easier: A large part of the monitoring work is eliminated and they only have to react to the automatic signals of the app, which are sent to a terminal device of their choice. The use of the software thus leads to significantly better results in the prevention of theft and a reduction in the workload of security personnel, as well as in the replenishment of shelves, the placement of products or the optimal advice to customers.

Irrespective of this issue, however, the trend is indeed in the direction of customers expecting more and more to be able to carry out the payment process in particular without assistance from staff. Self-checkout terminals, for example, are being installed in more and more stores. Computer vision software is very helpful here, as it recognizes whether the process is being carried out correctly. This prevents losses and ensures a better customer experience.

By the way, it does not seem to be the case that the use of AI in the retail sector will result in the loss of jobs. People usually have so much to do that any relief is gladly accepted. In the future, employees may be increasingly used to advise customers. This will not only make the shopping experience more enjoyable for customers, but also make the employees' work more interesting and varied.

You assume that the error rate of machines is now lower than that of humans. What is your forecast for the future? In which direction will this trend develop?

The performance and reliability of AI is already very strong today and improving at a rapid pace. For many tasks, machines are already en par with their human counterparts today and for all those where this is not yet the case today, it can be assumed that it will be in the near future.

I think we will be able to automate more and more in many areas what are still cumbersome or annoying tasks today. In brick-and-mortar retail, the use of AI still has to establish itself in many places, but it is becoming increasingly clear that the industry has understood that it needs to change and can improve - not least due to competition from online retail. Today, ultimately, almost everyone still shops at brick-and-mortar retailers, and their stores are important to the attractiveness of communities. But margins are relatively low, especially in grocery retail. AI can be used to optimize operational results, make processes better, make the shopping experience more attractive to customers* and be more lucrative to the bottom line. So the supermarket of tomorrow will have almost no losses due to theft, day-to-day tasks will be excellently organized, employees will be relieved, and the shopping experience will be enhanced by continuous automated customer analysis and personal customer advice.

You just won second place with Signatrix in the AI Beyond Borders Award 2022, congratulations! What does the win mean for Signatrix? What are your plans with the prize money?

The award confirms once again that both the AI industry and politicians attach a lot of importance to the topic of retail tech and that Signatrix's approach is appealing. Events like this are important to generate attention for innovative technologies and to signal to German companies that they are working in a location where their work is valued and supported. This is of course very motivating, it's a great recognition of our work.

The prize money is not allocated to any specific project, it mainly goes towards the development of new apps for our AI Suite.

Part of your prize was also a membership in KI Park. How important are such ecosystems for AI companies and startups?

They are enormously helpful. An AI ecosystem like the KI Park enables exchange with other entrepreneurs in the field of AI, but also with journalists*, scientists*, politicians*, investors* and so on. The platform that KI Park offers is very useful for making contacts or promoting one's own company. Moreover, you get good input at the events, which is not to be neglected. An exciting talk can give you some good ideas.  In addition, there is something "atmospheric". The fact that there are such institutions in Berlin, which have many members, organize events, provide coworking spaces, etc. - that simply gives you the feeling that you are in the right place with your project.

You founded Signatrix in Berlin in 2017. What advantages does the location hold for you as an AI startup? How do you feel about the climate regarding innovations in the field of artificial intelligence?

On the one hand, it's great to be based in a city that is so open and attractive to many people from all over the world, because this makes it enormously easier to assemble a high-caliber and international team. Of course, this is also related to Berlin's universities. There are already many people in Berlin, especially in the IT sector, who are very talented and want to work in a young company. And those who don't yet live in Berlin are happy to come here.

Second, the city offers a very good institutional network in the area of startups and AI. Berlin has produced many great software and online companies and also just such institutions as the AI Park. So the city offers an exciting environment. Many people here are fascinated by following or working on the development of innovative technologies. In addition, many funding programs at the state level help to drive the further development of technology companies in Berlin.

What is your general experience of Berlin as an AI hotspot? How will the city still develop in terms of its bandwidth as a location for artificial intelligence?

The development of AI technology usually works particularly well in close association with the companies that end up using the technology, as they often have and generate the data that is relevant for researching, developing and delivering successful AI solutions. Therefore, I consider it important to continue to provide appropriate incentives so that Berlin becomes even more attractive as a location for national and international companies of all stripes, so that appropriate partners can be directly integrated into the ecosystem.

In addition, we need to make it even easier and less frictionless to recruit top international talent - and in this context, issues such as effective employee participation should finally be resolved.