At the age of eight he ran his own duck farm. Make first sales. At 13 he was writing programs. At the age of 19, John Achim Holzhauer founded his first startup. In 2021 he will bring the third startup to life. A direct hit (not only) for the automotive industry. Because if you consider that the entire German manufacturing industry loses half a million euros in quality costs every minute, the AI solution comes from Semorai equal to salvation. The team is already raking in prizes and grants. The manufacturing industry can be happy.
Ariane Lindemann in conversation with John Achim Holzhauer
Why is the economy groaning under quality costs in the billions?
The goal of every company is zero-defect production. High quality, short production time and minimal costs. However, ever shorter innovation cycles, increasing international competition and increasing demands from customers pose huge challenges for production companies. In order to identify and analyze potential sources of error in production processes, optimal and early error management is necessary. In most cases, however, there is a lack of time and human resources for this.
Why is that dangerous?
The later one identifies potential errors in the product life cycle, the higher the quality costs. According to the so-called rule of ten, these grow exponentially from production level to production level by a factor of ten. It is therefore essential to identify the potential for errors during development, to take optimization measures and thus to systematically reduce the residual risk and make it manageable.
What are the options for error analysis?
In quality management, for example in the automotive industry, a so-called Failure Mode and Effects Analysis (FMEA) is usually carried out. It determines which components make up a product and what functions and characteristics these components have. So far it has taken about 12 hours per component. For very complex components up to 30 days. In the automotive industry, the FMEA is mandatory. In a Tesla, for example, there are 3,000 and in a normal combustion engine there are up to 10,000 components. This means that an FMEA costs around 10 million euros for a car. This is not only expensive, but also labour-intensive.
Surely this type of error analysis is not attractive for medium-sized companies?
On the contrary. In most cases, a lack of time and resources makes it impossible to analyze the error events in the larger overall context. The consequences are high quality and warranty costs or even image damage. German industry loses half a million euros in quality costs every minute. That is 300 billion euros every year.
You have developed a smart error management solution…
We call this “AI-based quality engineer”. The focus here is on a self-learning AI that acquires a store of knowledge from various sources, thereby understanding real connections and independently carrying out an in-depth error analysis. As a result, she not only supports employees during the highly complex design and production processes, but also supports them in error management across the company.
From duck entrepreneur to serial founder…
Entrepreneurship grabbed you early on?
I grew up in agriculture, my father had a poultry farm, so I always helped out a lot. That was a beautiful childhood. I earned my first money this way when I was six years old. I started my own duck farm when I was eight. With Indian runner ducks, they are known to eat snails. I rented them out to people with vegetable gardens and made my first sale.
At the age of 13 I started programming websites and got interested in startups. I’ve trained my mind to look for workable and practical solutions instead of seeing problems.
The first “real” startup came when I was 19: parents could buy school supplies for their kids on your portal…
Together with my best friend at the time, I founded Daba OHG in Bühl. With the brand school friend we have automated the annual school supplies shopping for parents. We talked to teachers about what the children needed, and the parents were then able to order it from our online shop with just one click. That was great. Unfortunately, the team didn’t fit anymore and I handed the company over to my co-founder.
You also penned the social media page “Bock auf Karlsruhe” …
Yes. While I was studying industrial engineering at KIT, I helped push the project together with Chris Wehle, who had the idea for “Bock auf Karlsruhe”. We originally wanted to build a recommendation platform for students for locations and events, where the higher semesters would explain to the first-year students where to go, where there are cool parties, etc. Shortly before the founding, my father died suddenly and I moved back home to support my mother. Then Corona came.
But then you ignited the third rocket…
In April 2021 I met my co-founder Joe at the SDaCathon of the research group SDaC. Together we founded Semorai. He studied electrical engineering and information technology at KIT, has been developing since his youth – mainly artificial intelligence in recent years – and has published several publications in the field of AI. During this hackathon we noticed that we complement each other very well. Me with the economic and he with the AI focus.
What exactly does Semorai’s AI do?
Using various sources, the AI gets to know the real connections around the product and independently carries out an error potential analysis. This means that potential errors and their influencing factors are already evident in the design phase.
Based on real-time data, cross-company knowledge and supplier and customer information, the identified error potentials are monitored during production and product use. This enables error situations to be predicted with an accuracy of up to 95 percent.
The AI automatically suggests suitable countermeasures to the employee or user to avoid the error situation. If this occurs nevertheless, it helps to quickly identify and eliminate the cause of the error.
Does this eliminate certain steps in the quality assurance process?
Yes, because the AI-based quality engineer for intelligent error management is something like a virtual assistant that relieves employees of work. Our mission is to give engineers more freedom to develop highly innovative solutions to the challenges of our generation.
The solution to drastically reduce quality costs…?
The potential behind this is that with our solution we reduce up to 80% of the effort for an FMEA, thereby reducing the hurdle for medium-sized companies to use this method for the first time and thus offering the opportunity to identify and resolve potential errors already in the development phase of a product . This is the greatest lever for solving the exponentially growing quality costs and thus enables companies to reduce their quality costs by up to 35%. This has an immense impact on the EBIT margin and thus strengthens the competitiveness of German industry.
The process level is your next goal then?
Exactly. We have an interdisciplinary team with age fluctuations between 18 and 62 years and can fall back on an excellent advisory board from different areas if we have any questions. But we also have extremely high demand from customers – including from the medical technology sector. We are currently working on a pilot project with a large automotive group.
You have won prizes and grants…
Joe and I came straight from university and had to figure out how we could build up a financial buffer. That’s why we applied for a few prizes and grants. We won everything we could win. We won the AI Cup, an AI competition at the University of Passau, and would have received funding of 100,000 euros from the state of Bavaria, but due to the simultaneous approval of the EXIST start-up grant, we were not allowed to accept it because of the double funding law. Then we were invited to Stage Two via Siemens. This is the largest pan-European competition for the best startups from European universities. We got an investment award from there Earlybird fetched from 100,000 euros. Then we still have each other Slush Heilbronn participated in a pitch battle, won the golden ticket and thus became the first startup in the first batch of Campus Founders’ AI Founders program. A total of around 350,000 euros was raised.
Despite your startup experience, you took the CyberLab Accelerator with you. Why?
With my first startups, it was more of a hands-on mentality, along the lines of: I’ll just do it. With Semorai, we quickly knew that it was going to be something bigger. That’s why we wanted to take a lot of knowledge from the accelerator with us on how to structure it properly. We find ourselves in a broad field of interests where certain framework conditions have to be taken into account. How do I design a product iteration? How do I validate my product? What do I need to know about taxes? On the other hand, we were able to build up a very good network here. That helped us significantly.
You are in the new Smart Production Park of the CyberLab Karlsruhe. How are you feeling?
This is a very cool and beautiful location. The open flair with large windows facing the corridor, through which you can always see who is walking by, creates a very happy and collegial atmosphere. You can exchange ideas and help shape how the community will develop right from the start.
Which customer group are you targeting?
Primarily automotive, OEMs, tier one and tier two suppliers. The difficult thing about automotive is the very long sales cycle. That is why we also orient ourselves to various other market segments. We will carry out a few other pilot projects this year, for example in the area of chemistry and pharmaceuticals, but especially in the area of medical technology, because we get a lot of inquiries here.
The interesting thing is that our AI can be generalized and scaled very well because you only have to build up domain-specific knowledge. Once you have set that up, you can use it to serve the entire market.
Semorai – the name is…
… borrowed from Japanese. In the 1990s, the topic of quality focus spilled over from Japan to us in Europe. We automate these knowledge-based methods by using artificial intelligence, the so-called semantic AI.