Peter Jeitschko, Solutions Sales Specialist for Artificial Intelligence at Microsoft Austria, talks in an interview about trends, collaboration between data teams and business departments – and what all this has to do with smoking at petrol stations.
He has been working in various IT consulting and sales positions for over 20 years, for leading brands including Volvo Cars, IBM, Oracle, and the data specialist Datameer. He is currently a Solutions Sales Specialist and AI expert at Microsoft Austria.
Mr. Jeitschko, big data has been a trend for years. Is it now more than just a buzzword?
It is certainly more than just a buzzword. However, “big” implies that we are talking about large volumes of data. But most companies don’t have a volume problem, they have a data complexity problem.
For this reason, the term “Big Data” is misleading in the vast majority of cases. What does it mean in concrete terms?
Companies usually do not have a central data repository, but instead many different data sources in various formats, from which it is necessary to generate added value for organizations. We are not necessarily talking about huge amounts of data. Data quality, scalable data platforms, and user-friendly analysis tools are essential to tap into this treasure trove of data.
Is that unlikely to change in the future?
That’s right. For one thing, new data formats are being added all the time. In many cases, it’s audio or video files that need to be combined with structured information from databases. Cloud applications are then added to the in-house on-premise databases. This, in turn, increases the complexity of data integration and data analysis.
Could new technologies, like machine learning and artificial intelligence, help?
We clearly see that the trend is moving in this direction. After all, the concepts behind machine learning and AI are not new but are based on thinking that dates back, in part, to the 1970s and 1980s. What has changed is the computing power and the price of processing data. Today, this enables new opportunities: for example, AI and the right analytical tools can be used to analyze machine noise and detect early on when a machine is about to fail. These use cases fall into the category of “predictive maintenance”.
Other use cases in image and video recognition go in the direction of security. For example, AI can be used to automatically detect whether someone is smoking at a petrol station, or whether someone is wearing appropriate safety clothing at a construction site to trigger an alarm if necessary. In the future, we will see more and more use cases like this.
Which organizations find it particularly difficult to deal with complex data?
I’ve had the privilege of working on projects at listed corporations, as well as smaller mid-sized companies and start-ups. In data projects, the size of an organization plays a subordinate role in dealing with data complexity. However, there are cultural differences when dealing with data. In Anglo-American areas, data is handled differently than in Austria, Germany or Switzerland, for example.
The justified desire for data protection is often used as an excuse for not tackling data projects in the first place. Particularly those organizations that in some cases have invested decades in building up their IT infrastructure and are now suffering from “digital legacy”. In the banking and insurance sector, for example, it is not uncommon to find systems that were introduced 40 years ago. However, it must be clear: The problems of 2021 cannot be solved with the methods and infrastructures of 1970. Other industries such as retail, telecoms and logistics have already recognized this and are leading the way.
How can problems with such legacy issues be solved?
It is important that organizations have a digitalization strategy and, derived from this, also a data strategy. This strategy must be supported and driven by top management and consistently implemented by a “Centre of Excellence,” or similar. This also means that you have to throw the traditional frameworks overboard and possibly set up completely new IT structures. From a risk perspective, however, this is not possible in every case, so it is important that new technologies in the cloud area integrate as well as possible with the existing IT infrastructure – complementing each other. These hybrid cloud application scenarios will be with us for a long time to come.
What is your advice to organizations when tackling data projects?
It’s no use building a costly data infrastructure if it doesn’t ultimately solve the problem in question. So first you need a concretely defined problem or question, then a solution idea, and only then the individual technical components to be able to solve the respective use case in a meaningful way.
Why do most data projects fail today?
Data projects fail for two main reasons: The first is the appropriate skill set of the company’s own employees. Management must understand the opportunities offered by digital business models and provide employees with the opportunity for targeted training. Microsoft supports organizations in this area, for example, with the “Enterprise Skills Initiative” – an education and training program tailored to the organization’s individual needs.
And the second reason?
The second reason is a lack of a digitalization strategy or one that exists only on paper and is not lived out in the company. It is essential that the employees are closely involved in the implementation of the strategy. It is important to remember that digitalization and data projects are not purely IT projects. The respective experts from the business departments must work closely with their IT colleagues and specialists, such as data scientists. Otherwise, experience shows that data projects are doomed to failure.
Here you can download the magazine including the expert interview “Digital transformation” for free:
Fill out the contact form to receive the magazine issue. You will then receive the entire issue by mail.
Preview of the Digital Magazine Issue: