Nibiru Systems – Building the Future with AI

Ever since the breakthrough of ChatGPT, we cannot imagine a world without AI applications. While the most popular ones are directly interfering with us humans – a chatbot talking like customer support, Midjourney creating images like an artist – the implications are much bigger: AI can help us boost automation to a never-experienced level and use previously unimaginable amounts of data to make systems more efficient. Sometimes, these amounts of data need to be generated using novel sensor technology. 

The German startup Nibiru Systems took on the challenge of supporting companies in their journey towards a data-driven future. In today’s interview, we talk to Nibiru Systems co-founders Anton Bosse and Marvin Arnold about their contributions.

Why did You Start Your Startup?

We noticed that many small and medium-sized companies in Germany have great difficulty using data or tackling problems in a data-driven way. From various projects in the past, we noticed that we are good at using data to achieve concrete results for these challenges. In this way, we want to contribute to efficiency because Germany with its restrictive wages and social costs, has high labor costs – hence, automation projects are worthwhile to achieve higher productivity. Our motivation was to be able to solve these problems with modern AI methods.

We want to help companies, so that even more time is preserved and we do not completely fall behind other countries such as the US regarding technology. Here, we push a little harder to help customers use their data. Due to the omnipresence of AI, we believe that our technology can help every company in some way.

What is Your Product?

We currently mainly offer a service where we accompany customers from 0 to 100 in the AI sphere. If the customer has not used AI before, we start with the planning to see which data is already available and which needs to be recorded. What pain points do they have, and what opportunities can we help them with? We then create a roadmap, taking into account the assessed data and objectives. This process is the consulting part of our offering. After that, our work moves on to developing and deploying the solution  – an AI model or some software that solves the customer’s problem.

Additionally to the custom solutions, we also founded the company to look at the recurring problems that can be implemented as our product(s). We can give you a quick overview on two of them: The first is the image area, often applicable in the medical field, for example, from photos or microscopic images. In these images, objects are selected and classified, which can be used in very different tasks. We can further apply this methodology in other use cases, such as in the manufacturing sector, e.g. when checking whether an object has been sealed correctly. This can be seen with a camera as well as with the naked eye. Another example is electrical engineering, where we want to quickly detect damage, such as poor soldering or welding points, and offer an easy-to-use solution.

Our second topic is helping our customers generate documents with AI. Our solution goes beyond a simple chatbot: In a recent client project,  we have already created automated documentation. This has helped to get a plant worth over a hundred million dollars into production on time, saving our client millions of dollars in the process. We have achieved this by enabling an engineer to increase output by a factor of 5 while maintaining the same level of plant knowledge to be able to help out with occurring problems during production.

In the energy sector, we support a company that provides an app for managing the energy consumption of medium-sized companies with high power consumption (e.g. a sawmill or an agricultural company). We want the problem to be presented to the customer transparently and thus also help the customer to be better informed when purchasing energy. This is not our own product, but we advise a company on designing this app and answering data-driven questions such as: What is the forecast for next year? How can you optimize purchasing?

What are the Advantages of Your Approach?

Many companies have already collected a lot of data over many years – and may still not even know that they have collected it. Making unused data usable creates excellent potential for full or partial automation or assistance systems. To do this, you have to make a plan, approach it in a structured way, and talk to the experts on-site to see what is really possible. The broad range of projects we have worked on enables us to provide our clients with the right solutions and help them achieve rapid results through our quick-win approach.

Our own products have other specific advantages, for example, when it comes to documentation, AI can also create documents with human quality and much faster than an engineer reading a 1000 page manual. This is really revolutionary and unthinkable compared to old methods. It enables possibilities that didn’t exist before.

What is Your Biggest Challenge?

AI means that you do not have to understand much about the physics or engineering background. However, of course, you still have to be aware of the problems and product-specific aspects—the power consumer differs from the financial market, errors that occur in the medical field are different from mechanical errors, and so on. We also have to have a certain understanding of this and talk to the relevant decision-makers or experts. It is also a big challenge to acquire enough good-quality data in certain fields.

Another point is expectation management about things that AI cannot solve. Especially with the current hype around ChatGPT and other AI models, many customers expect that everything should be solvable with AI, regardless of what the data looks like. Therefore it is sometimes difficult to explain that some things do not work with the existing data. The most common problem is that there needs to be more data to make accurate predictions. We incorporate this right from the beginning in the consulting phase.

What are Your Future Plans?

Our next goal is to see how well our products are received. We want to push ahead with image recognition in the medical field and work with other institutions. We want to test it on other sectors,  e.g. manufacturing, to see how well it translates. We will also continue to develop the document generation with the large industrial customer with the above-mentioned plant and also offer it to other companies with similar problems. 

We are always looking for more use cases. It is exciting to see how we can build our solutions in such a way that customers can implement them generically. Our focus is on the data, but when you have a large company as a customer, there are often very complicated IT structures with very fixed roles and departments and a lot of data protection. It is necessary to learn what the existing processes are like and how we can integrate our products  as smoothly as possible. We have already learned a lot in the last two years, but we will definitely be able to learn a lot more in the future.

Whom to Contact?

Are you feeling inspired by this exciting idea and eager to explore more? Reach out to Anton or Marvin for a delightful discussion, or simply visit their website to learn more about their work.