Startup Collaboration Management

A step on the way to digitization

Interview with Dr.-Ing. Thomas Wenzel

Artificial Intelligence (AI)

answers questions that no one has yet asked!

  • Human intelligence works with «If> then> because»
  • Artificial Intelligence works only with «If, then, » but with billions of nested relationships
  • AI is usually superior to human intelligence:
  • It takes many more influencing factors into account
  • It examines all connections, humans only the presumed
  • KI responds to questions that nobody has asked - and that is often the sticking point.

Startup collaboration management – can you describe that in a few sentences?

Dr. Thomas Wenzel: Startups have experts with experience in the latest technologies. Thus, they can build superior new solutions, often based on artificial intelligence (AI), but they often lack access to industries and detailed knowledge about of specific applications. On the other hand, it's hard for corporate companies to define where to make real improvements with AI – and we are not talking about 5%, we mean 90%. So, if it is obvious, that startups and corporate companies are a profitable match.

The first step of this process is a common analysis of improvement requirements, the potential availability of corresponding disruptive technologies. In the next step, I look for startups that are active in the matching area and that not just have an idea but really can master the potentially relevant technology. Step three is conducting joint development projects with company and startup experts. After this proof-of-concept is has been successfully and accepted mutually, step four is integrating the technology into the company's organization. Step five often follows, i.e. an investment in the startup to secure midterm access to the technology, to influence the startup's strategy, and to prevent competitors from using the technology.

You primarily address medium-sized companies. Why?

About half of the largest companies in Germany, such as BMW and Siemens, have established their own centers for startup competence. For other companies, however, this competence would be essentially important to not lose the connection to the current technology – as it happened to seemingly invulnerable market leaders like Kodak, Nokia, Blackberry. At first, companies have to recognize the opportunities as well as the risks of the new disruptive technologies – and this at an early stage. And, as the setup of an own center of competence for startup collaboration might not pay off for a medium sized company, we offer the complete solution «as a service».

To identify such opportunities, do you consult your customers first?

Yes, but only with regard to it clearly defined topic. We are primarily operating managers and not as much consultants. We will not turn the structures of the company upside down. Our task is the implementation of the solution.

« Our focus is completely different: We find a startup that fits your requirements perfectly - and that is the basis for your venture capital financing. »

Why is consulting prior to the project important?

Recognizing potential is a difficult thing: you need to look at many issues, from the the entire process down to for example a snap action that does not work reliability reliably. And with each issue you have to consider the wide range of possible applications of AI in order to recognize the potential for improvement. Plus, people who have just successfully reduced their costs by 6 % in the last year was will not be looking for a 90 % cut – using a technology that they scarcely know about. So, this initial consulting is about the definition of defining the potential and the appropriate disruptive technology.

Then, what is the «right solution»?

The first question is: what exactly is the problem? Only in a few cases is this easy to define. I often experience that my conversation partners did not look at a certain problem as it was deemed hopeless – due to the lack of awareness of a specific disruptive technology's potential. Then, the right solution is the right technology that improves functionality and costs by a magnitude. This may sound sophisticated, but often this is the manual part of the job.

The Internet of Things (IoT)

is networking with «everything»: objects, vehicles, people, animals ...

with «all» properties: location, temperature, pressure, speed, physical tension, vibrations, pressure, heart rate ...

This requires comprehensive sensors, for example sensors integrated into the surface of clothing or surface coating of machines, wearables for humans ...

It takes communication «to the point last corner », for example Low Power Wide Area Network (LPWAN).

IoT is the indispensable food source for Artificial Intelligence: AI lives from access to «all» data, that IoT delivers.

The «right solution» can be found at the «right startup» then?

The right solution is based on the right technology. This technology must be in command of the right startup – precisely in the that corresponding sector. Take AI-based 3D image recognition as an example: a startup processing 100,000 image recognitions per second will not usually be as efficient in the standard hundred-images-per-second range – and vice versa. Once the exact sector of disruptive technology has been found, it is most likely that the startups will apply that technology in a very different industrial sector than ours. In other words, the startup has probably analyzed an issue very different than our current problem. In a recent project, I needed to detect trucks, cars, and pedestrians. I found a startup that fulfilled my requirements perfectly – the reference was a medical images analysis system. By the way: a reference from another area is advantageous. The quality of an AI solution rises with the number and variety of learning cycles. So an AI system with applications in other areas is especially powerful.

That means that the technology is of prime importance and not the industry application?

Exactly. An AI-based technology is applicable in almost every industry. So you will hardly find a startup with a solution for your x-problem in your y-industry – and the search has to look for the x-problem regardless of the y- industry. The targeted technology has to be defined independent of the application and the industry. One example: a railway operator needs to know if trains are approaching railroad crossings. State-of-the-art: sensors and cables along the route in the ground are installed over many kilometers, require civil engineering - and that's expensive. I found a startup that makes this effort obsolete: It uses just one vibration sensor plus pattern recognition by AI. This can save costs in the six-digit range. This is an example for the importance of our expertise. The technology of pattern recognition technology via artificial intelligence is miles away from the ‘always used’ induction measurement - and so it was not in the scope of the company’s experts.

And how do you look for this «right» start up?

There is no supplier directory. The startup scene is too dynamic and not institutionalized. The search works only via personal network, supported by Internet research can help. There are about a million tech startups worldwide - and this scene changes rapidly. I have to know what disruptive technologies are presently available – and typical sources or location for each technology. One example: in Germany founders often come from universities and have a strength in deep-tech. For a long time, I doubted if there was not an orderly process to search for the right startup and so we sponsored scientific research. It confirmed that finding the right startup successfully depended decisively on the personal network 92 percent of the time.

You said AI also gives answers to questions no one asked?

The basis for AI is the evaluation of huge data pools. Correlations are calculated from each data group to every other data group. Billions or trillions of correlations are calculated in a very short time. Due to the sheer mass, it is possible to recognize relationships without an expert having made a previously hypothesis previously. «No hypothesis» means answering questions that have not yet been asked – and this often is the key to an intelligent solution. AI recognizes structures in the mass of relationships and thus, for example, detects failures of a single part in a large system, forecasts downtimes, calculates in detail what maintenance measures are required, whether a picture is showing a danger, etc. Autonomous machines (AM) are based on artificial intelligence. The difference: AI suggests actions, AM implements actions autonomously.

Big Data

is a collective term or also the basis for Artificial Intelligence.

It includes collecting «all» data through IoT, evaluates and uses by AI and the applications in autonomous machines. One requirement is the exploding computer performance in recent years. Until about five years ago, we have

noticed this in the higher efficiency of laptops and cell phones - meanwhile they do not have to exploit their potential anymore: for example, a computer with a i5-Core is often quite enough. Laptops are only terminals of big data centers for AI. We notice that only indirectly - for example through the good speech recognition of our mobile phones.

Artificial intelligence is only being used now. Why?

AI needs very high computing power that has only recently become available - think of the billions of correlations we've talked about. Since it usually exceeds the performance of a computer, we need and use the cloud. And another reason for the cloud: algorithms are learning even during the operational phase. At the same time, the more diverse the applications are that they calculate at the same time, the more powerful the algorithms become. And that many systems are calculated with an algorithm is another reason to operate the algorithms in the cloud. And a good range of cloud services, from different providers and worldwide is relatively new. Plus: data storage and transmission as well as sensors have also become considerably more affordable.

How do artificial intelligence, IoT and Additive Manufacturing work together?

The systems complement each other – work goes hand in hand. These are topics of digitization that complement each other. AI requires large amounts of data, permanent in real time, from everywhere and to every with all detail - without this, AI is pointless. And the Internet of Things (IoT) provides precisely these required data volumes. IoT makes data available everywhere. For example, it brings the tolerance of a plain bearing in a machine from another continent to your smart phone. And if you want to use this detailed data for flexible products, additive manufacturing (AM) makes this possible at low unit costs. And the exact specification for it probably comes from AI.

Do you always offer AI solutions?

Our customers need a functional solution. The path is secondary, the result is important. AI is not a target but a tool. Nevertheless, of course, we consider all disruptive technologies equivalent, but practice has shown that most of our tasks were best solved with AI.


startups combine hardware and software optimally right up to the merger in one product. The hardware - machines, sensors or vehicles - receives a turbo-like acceleration. Currently, especially software, Internet and smartphones are being combined and modified over and over again. Deep Techs, on the other hand, generate a much higher degree of innovation and a new world of performance. Prerequisite for this is an excellent knowledge of this hardware. Therefore, Deep Techs are mainly found in Central Europe, close to leading technology companies.

Suppose my company comes to you as a customer. How do you proceed?

We go through a 5-five step process together; I do that as a project manager.

Step 1: We define potential for major improvements using disruptive technology: where can process steps be analyzed and thus be improved or simplified? What can be designed flexibly and thus tailored to fulfill the demand at lower costs? Are there any known problems without solutions? Can I edit something in total rather than in single steps? Where are market opportunities through new flexibility? Can hardware be replaced by software? Are there completely new solutions or opportunities that the competition already uses - or with which you can differentiate yourself from the competition? What are the goals of the corporate strategy - where do you deviate from it? Once the search fields have been determined, a first estimation of the possibilities by disruptive technology is made and finally a prioritization takes place for each one. It all goes through an iterative process: it incorporates technology information from me and specific engineering and market information from you. Finally, we decide on the basis of a prioritization, which points should be further elaborated with which disruptive technology.

Step 2: The search for the right startup. I search for possibilities and create a longlist / shortlist using many different sources. Then we will make the final selection together. Here you and your experience are particularly important because the selection among the last candidates can be a very subjective decision. Simply the question: Can I and do I want to have a partnership, a project with these people? Trust plays a big role.

Step 3: We are planning a pilot project that will put the new technology through its paces - under dizzying conditions, but at a reasonable cost. If it makes sense, we should integrate your customers, where necessary, certifications also belong to the proof-of-concept. This may require custom development from you or from the startup. The result of the pilot project is the proof-of-concept. Only then will we know whether the system is ready for operation and marketable, whether it meets all requirements, even under the adverse conditions of the practice.

Step 4: Now the completed solution has to be integrated into the corresponding business unit. Organizational adjustments and training of personnel may be necessary; the distribution and the project management must be integrated. Responsibilities need to be regulated and perhaps the quality process needs to be redefined. Now, the contractual arrangements with the startup must be clarified.

Step 5: The acquisition of company shares of the startup. There are good reasons for this: As a member of the Supervisory Board, you influence the startup strategy and can prevent the competition from participating in the new technology. Your and other industrial applications of the new technology increase the startup's value. You should participate in that. Finally, if there are further R & D expenditures, the costs will be borne pro rata by the other investors in other industries - a welcome leverage effect.

« Our customers need a functional solution. The path is secondary, the result is important. AI is not a target but a tool. »

Do all stages build on each other?

Each level is the base for the next stage, and this base must be safe. Therefore, after each stage, we decide again whether we are going to the next level. It may also be necessary to take a break between two stages. It happens that no perfectly fitting startup is found. Experience shows that in the rapid development of disruptive technologies it takes a few months before the right solution is available. For example, one customer was looking for a high-security radio connection because human lives depended on the security of the communication. What could be implemented so well over cable was simply not available as a wireless solution. The project rested for half a year until I found a startup that had developed the required solution for military applications.

Why do you use startups,? Don’t medium-sized companies have their own resources?

Very few medium-sized companies have an expert for AI in their ranks - and hardly for the sub-subsegment of AI, that now needs our solution. And if this expert also had years of experience in this field of expertise, then that would be like a major win in the lottery. But this experience can really be found among the 1 million tech startups. By the way: such specialized experts in high-tech niches were intellectually starving in most companies because they lacked connections and discussions.

Aren't there suitable solutions in our environment?

For example, in Germany we have world market leaders with excellent experts. Certainly, we can find good solutions in medium-sized companies. But everyone should be realistic about their own strengths: even if the idea comes from your people - are your resources fast enough? Startups are incredibly creative. They like to work a lot and the latest technology stimulates them. Because of their structure, they can tackle tasks openly and playfully, and with completely new ideas. And in our case: the new solutions are developed jointly by the startup and your people. Both sides bring their strengths and challenge the other side. An example: solutions often have to be certified. Neither the startup nor I know your industry-specific customer requirements or the standards for certification in your environment. In other words, it is a true collaborative effort and you have an influence on the accuracy of fit and solidity of the solution down to the last detail.

How do you choose the startups?

In the first step, we jointly defined the technology. I then look for companies with real expertise and references exactly in the defined technology segment. The search is predominantly conducted via my international personal network with people and institutions, incubators or business angels, sometimes matchmakers. Some countries have amazingly dedicated consulates, which are very helpful. And there are exclusion criteria: A startup must be well beyond the idea phase - there must be a working prototype that I can test - even if we will not use it later. Finally, it's a cultural issue, a question of mentalities: Do I believe in their skills and reliability, will they work well with my staff as a team?

Where do the startups come from and how many do you address?

The origins are really international and very different. In Germany and Northern Europe, we have strengths in deep-tech, that means a software that works as a «turbo for hardware». That does not only work in itself or with other software. Startups from the US have a lot of software that supports e-commerce business models. Startups from Europe often improve an existing business with completely new technical solutions. Countries like Israel, Finland or Lithuania are interesting. They have a young, well-educated, motivated and technically-minded population - and startups. 0n the other hand, there are few big companies that could be customers - so the startups are very open for international cooperation. Many startups come from the field of technical universities - and as in the special case of Israel, they come from the military's elite technical units.

How risky is it for medium-sized companies to work together with startups?

The risk can be controlled. The tasks are clearly described and the results are tested in the proof-of-concept under your guidance according to your criteria - including the necessary certifications. The greater risk is to miss the new technologies because they are not recognized or applied. See Kodak, Nokia or BlackBerry.

« From 100 potential Startups remain in the end about three, who will master this technology. »

Why should you acquire shares in a startup?

After the proof-of-concept has been successful, investments in company shares offers advantages. With a seat on the company's board, you can influence the strategy in your sense. As a co-owner, you decide who participates –you can exclude competitors, for example. And as the startup develops a true industrial application through you, its value increases. You should participate in that. And finally, if further developments need to be funded in the future, your co-investors will also participate - a welcome leverage effect.

Do you also offer such a participation in the startup?

Venture Capital (VC) is part of our offer. As a technical VC, it differs fundamentally from the classic, purely financial-oriented VCA. The classic venture capital (VC) investment is geared to financial variables: expected increase in value of the industry, valuation of the other investors involved. Hardly any consideration is given to the technological performance of individual companies. The goal: to get out of as many investments as possible after a few years with a high profit. Venture means adventure! Our focus is completely different: We find a startup that fits your requirements perfectly. We validate it together in the «proof-of-concept»: are your requirements met under difficult conditions, is it the solution that will help you to advance the market? - and that is the basis for your venture capital financing. Of course, it's a goal to participate in the value creation that comes from your industry application. But the primary target is to lead an innovative.

You can also download the interview as a brochure.

Do you have an example of the different applicability of a startup technology?

A startup from a completed project has an AI application that can very accurately detect, analyze, and evaluate three-dimensional objects from a coarse point cloud. In a previous project with another company, the system examined radiological images, and identified and distinguished benign and malignant tumors. With this performance as a reference, we used the startup for a completely different purpose; the difference may surprise you: on a parking area on the highway, trucks, cars and pedestrians were recognized, and their position and size were determined. Safety warnings could even be generated from the movement patterns. The effect that one application benefited from the fact that other applications have also trained the algorithms becomes obvious here. This again is an argument or the question from earlier, namely that other applications and investors are very welcome for a startup.