Artificial intelligence: strategies and projects at CANCOM

CANCOM Slovakia: Katja, you are the Internal AI Chapter Manager at CANCOM and thus deal with cutting-edge topics. How did your journey into the world of artificial intelligence begin?

Katja: I originally studied communication design and joined a SaaS company where my position served as an interface between marketing and sales. Both during my studies and in my first job, I came into contact with AI-based technologies and CRM systems, which sparked my interest immediately. I later developed an AI-based sales training program at an IT consultancy that was rolled out across the group. I was also able to take over the training for it, which taught me a lot, especially about user adoption and employee enablement. I joined CANCOM at the beginning of 2024, where I am now responsible for internal AI projects.

CANCOM Slovakia: What strategy is CANCOM pursuing when it comes to using AI-supported tools within the company?

Katja: At its core, the aim is to increase employee productivity and efficiency and to replace repetitive, time-consuming tasks with AI. Often, these are administrative tasks that require much less or no manual effort with the use of AI, thus making the actual working day much more efficient. The time saved can then be used for meaningful and important tasks. This is a great advantage, because it leaves more room for strategic, creative and value-adding work.

CANCOM Slovakia: What measures are you implementing to introduce employees to the new technologies?

Katja: We have developed a multi-layered approach. On the one hand, we are taking a top-down approach by training management so that they and multipliers can carry these technologies further into the business units and their teams and also act as points of contact. At the same time, we are pursuing a bottom-up strategy with webcasts, training and newsletters to address all employees directly. For example, there are job talks on strategy and kick-start training to introduce new users to Copilot in Microsoft 365 and thus make it easier for them to get started working with AI. Here we explain to our colleagues in detail how the AI works, what they need to bear in mind when using it, where they can find the tool and provide specific use cases.

CANCOM Slovakia: What are your specific tasks in implementing the strategy?

Katja: To get a better understanding of this, I’ll briefly explain what the Internal AI Chapter actually does. The scope of activities covers four pillars:

• We have already addressed the first pillar in detail – the topic of user adoption: How can we create transparency and sensitize and train our employees in the best possible way in how to use AI?
• The second pillar covers the strategy and thus also our internal roadmap. The aim here is to align the requirements of our employees with the regulations set by the EU, and thus contribute to the CANCOM corporate strategy. The other two pillars ultimately arise from this.
• The third pillar is the use of AIs that are relevant for the entire CANCOM Group. These include, for example, the CANCOM Assistant – our internal chatbot, the use of Copilot M365, other bot solutions, but also the use of tools for more productivity and efficiency in deadline and task management, as well as voice-to-text solutions, which will play an increasingly important role in relation to AI in the future.
• As a fourth and final pillar, the Internal AI Chapter acts as an advisor and initiator for the development of AIs for certain units of CANCOM. Here we look at whether we have development resources in-house or whether we may need to work with external providers. In these projects, I take on the project management both internally and externally.

So to answer your question about my specific tasks: I am in informal contact with our employees and suppliers on a wide range of topics, develop ideas and strategies for implementation from this, also take responsibility for and manage the resulting projects, and ultimately train colleagues in how to use the tools.

CANCOM Slovakia: You mentioned the CANCOM Assistant. What exactly is it and how does it make your day-to-day work easier?

Katja: The CANCOM Assistant is our in-house AI assistant designed to support employees in their daily tasks and in their search for information. Often, quick access to the right documents or tools is a challenge, and this is precisely where we save time with the CANCOM Assistant. If it doesn’t have a specific solution at the ready, especially for technical questions, it can even create tickets that are then forwarded to the responsible employees for processing. In the future, it is to be the first point of contact for all administrative tasks, thus minimizing the manual effort involved in booking vacations, reporting sick, traveling and ordering hardware for new employees, for example.

CANCOM Slovakia: What are the differences between having your own chatbot like the CANCOM Assistant and solutions like ChatGPT, LLama, Perplexity and Co.?

Katja: The freely accessible models such as ChatGPT and Co. are open source by default. This means that any information we enter into these tools is processed and stored in data centers outside our company and used for training purposes of these AI models. However, especially in the work environment, we usually deal with highly sensitive data that should not fall into public hands and should never be used as a database for training public AI models. To prevent this from happening while still allowing employees access to AI, an in-house bot such as the CANCOM Assistant is a secure solution that we not only use internally, but have also already successfully rolled out to customers.

CANCOM Slovakia: What does your team look like and is there such a thing as a “typical” working day in your job?

Katja: The Internal AI Chapter is an association of different key players from various departments. Marcel Reifenberger, our Chief Information Security Officer and Vice President Internal AI, heads it. Depending on the AI and deployment scenario, we work closely with GIT, Platform Enablement, the Competence Centers, Legal and other specialist departments, which makes the projects varied and dynamic. I don’t have a typical working day. Although there are some fixed meetings, apart from that my day is often flexible in terms of topics and arrangements. As fast as AI is developing, so fast-paced is my working day. One constant that arises from this is the need to keep up to date with information.

CANCOM Slovakia: Which tasks do you particularly enjoy?

Katja: I particularly appreciate the variety of my tasks and the close exchange with colleagues. It is fascinating for me to see how we can make everyday work easier through AI processes, and I enjoy dealing with new technologies. A highlight for me is working on business-unit-specific topics because there is a lot of innovation potential there. I also find the area of user adoption interesting – in other words, how well the tools are received by employees. I get valuable feedback in this regard, for example after training sessions or webcasts. Recently, I was able to take over a slot at CANCOM Microsoft Connect and spoke there about the use of AI in sales, for example how to use AI to prepare for meetings in order to get the relevant information in a more targeted and, above all, faster way. Internal and external events give me the opportunity to exchange ideas with many different contacts and thus gain a new perspective on both the potential and the solutions offered by AI.

CANCOM Slovakia: What challenges does the introduction of these tools entail and how do you deal with them?

Katja: A big issue is actually getting colleagues to use the solutions. It’s not enough just to provide new technologies – employees want to understand how they can use the tools in specific cases. In the day-to-day work of many, there is no time to try out new tools independently – for a higher adoption rate, in addition to transparency, spaces for exchange and contact persons for questions and for the transfer of information must be created.

Another issue that definitely needs to be addressed is raising awareness of data verification – is the answer I get from the AI correct? But data protection and security also need to be given special consideration – in other words, which AI model do I enter my data into?

That’s why every AI application we use undergoes strict testing before it is deployed. If someone wants to use an AI tool, it has to be submitted for approval, and we then check whether it meets our data protection requirements. Data protection and security are our top priorities.

CANCOM Slovakia: Thank you for the interview. In the next part, Katja will tell you what trends, challenges and risks she sees in the use of AI.