*Thisarticle is based on a YouTube video by Sophie Hundertmark, an expert in the use of artificial intelligence with a focus on chatbots and strategic AI applications in companies and public institutions. Sophie is a researcher and lecturer at the Lucerne University of Applied Sciences and Arts and is doing her doctorate in Conversational AI at the University of Fribourg. The blog text was created using a custom GPT model that was trained on Sophie’s video content, language style and expertise. The result is well-founded, up-to-date articles based on Sophie Hundertmark‘s own expertise.
You can find the link to the video at the end of this article.
What is the SKI Pilot anyway?
The SKI Pilot is the savings banks’ internal AI solution. Some also affectionately call him Skippy or Skippid. This refers to an AI assistant that is specially designed for work in the savings bank world and is now available throughout Germany.
One thing is particularly exciting: the SKI Pilot is not just another chatbot. It is embedded in the savings bank environment, works in a secure environment and is increasingly connected to internal systems. This is precisely what makes it much more relevant for many employees than general AI tools.
While traditional AI applications often fail due to data protection limits or a lack of specialist context, the SKI Pilot starts where real support is needed in day-to-day work: with internal processes, emails, customer preparation and, in the long term, even with specific actions in the core banking system.
Why the start was initially disappointing
As with many AI projects, the SKI Pilot got off to a bumpy start. In the first phase, the solution was still able to do comparatively little. There was essentially a pure chat and text function. Savings bank-specific knowledge was largely lacking, integration into internal systems was not yet in place and the practical benefits remained manageable for many employees.
The result was foreseeable: Many people tried out the tool, were disappointed and quickly put it aside again. Those who do not recognize any concrete added value after a few tests usually do not consciously take the time to do so again in everyday life.
This is an important learning point for companies: An AI solution is not automatically accepted just because it has been introduced. If the benefits are not apparent at the beginning, skepticism quickly arises.
What has changed at SKI Pilot in the meantime
Today, the picture looks very different. The SKI Pilot has undergone significant technical and functional development. Several model changes, better performance and, above all, the connection to relevant savings bank systems have led to many employees now having a much more positive view of the tool.
Particularly important here is the proximity to the actual working day. Today, the SKI Pilot can not only answer general questions, but also access internal information. These include guidelines, instructions, information management systems and other applications specific to savings banks.
This is precisely where the difference between a general AI and a truly workable enterprise AI arises: it not only knows language, but also the technical context.
The great added value lies in system integration
The most exciting aspect of the SKI Pilot is not just the language model itself, but its integration into existing systems. Because this is where a nice technology becomes a real assistant.
During the discussion, it became clear that the SKI Pilot already provides support in several areas:
- It can access internal savings bank information.
- It helps with the preparation of customer meetings.
- It is integrated in Outlook.
- In future, it is to be linked more closely with Word, PowerPoint and other applications.
- In the future, it can initiate actions directly in systems.
This is a decisive step. Because employees don’t just want to chat with an AI. You want to complete tasks faster, more structured and more easily.
How the SKI Pilot supports e-mail work
A particularly tangible benefit can currently be seen in Outlook. The SKI Pilot is already available there as a plugin and helps with formulating e-mails or replying to incoming messages.
This sounds unspectacular at first, but is highly relevant in everyday life. Emails in particular are one of the biggest time wasters in many teams. When an AI prepares drafts, structures answers or improves formulations, this not only saves time, but often also reduces mental stress.
Of course, it remains important that employees continue to check the proposals themselves and then approve them. It is precisely this combination of AI support and human control that currently makes the most sense for many companies.
Preparing customer meetings with AI
Another interesting use case is the preparation of customer meetings. If an employee enters a customer number, the SKI Pilot can collate the relevant information and prepare it in a clear format.
This does not result in rigid interview instructions, but rather in sound preparation. For example, it can become clear which topics should be addressed in the conversation or where there is a need for action. If, for example, important information is missing or documents are no longer up to date, this can be recognized at an early stage.
This can be a considerable advantage, especially in areas that require intensive consulting. The quality of the preparation increases, information does not have to be laboriously gathered from different systems and the focus in the conversation can be more on the customer.
From information to action: where the development is heading
Things get even more exciting where AI not only provides information, but also interacts directly with systems. During the discussion, it was described that the SKI Pilot should also be able to carry out actions in the core banking system in the future, for example in the context of opening an account.
This significantly changes the role of AI. It is then no longer just a research or text tool, but an active component of operational processes.
This is precisely where the next stage of development lies for many companies: not only providing answers, but also preparing, triggering or accompanying work steps. This saves time, reduces frictional losses and can improve process quality.
Images, PowerPoint and language are also becoming more important
Multimodal skills are another area of development. The SKI Pilot can now handle images and PowerPoint files better. OCR, i.e. text recognition in images and documents, makes accessing content much more practical.
Speech recognition and voice output are also planned. This opens up new possibilities, for example for taking minutes of meetings or for more natural interactions in everyday working life.
Here, too, it is clear that the added value of AI does not come from a good language model alone. It is created by functions that are linked to real work processes.
Why good technology alone is not enough
Despite all the progress made, one point became very clear during the discussion: even the best AI is of little use if employees don’t know how to work with it.
This is precisely why the introduction of corporate AI is never just a technology project. It is always also a change project. People need to understand what the tool can do, where it helps and how they can use it sensibly.
Many employees expect a perfect result immediately on the first test. If this does not happen, the AI is quickly labeled as useless. However, the problem is not always due to the technology itself, but often also to a lack of support, unclear expectations or a lack of practice.
What companies can learn from the launch
What is particularly valuable about this example is that the introduction of the SKI Pilot did not simply end with an activation. Instead, continuous support, explanations and readjustments were provided.
Some of the most important success factors were
- Training in small groups
- Regular exchange meetings
- Testimonials from colleagues
- Coaching for managers
- Concrete work on real use cases
- Visible communication of new functions
It is precisely these points that often make the difference in practice. Because acceptance rarely comes from announcements. It arises from experienced benefits.
Managers play a central role in AI
One particularly clever approach was the targeted work with managers. Instead of just offering general training, we looked together at where the biggest time wasters are in teams and which prompt templates or AI applications could specifically help there.
This is so important because managers have a dual role. They must not only recognize a benefit themselves, but also provide orientation for their teams. If they understand how AI can be used sensibly, it is more likely that their employees will also be more open to working with it.
It also turns an abstract AI topic into concrete work support. Instead of talking about technology, people are suddenly talking about e-mails, customer preparation, processes and saving time.
Prompting is not a side issue, but the key
A key learning from the discussion was the importance of good prompts. Because even a powerful system only delivers helpful results if the task is clear.
This sounds banal, but is often underestimated in everyday life. Many people test an AI with vague inputs, get mediocre results and conclude that the tool is no good. The bottleneck often lies in the wording.
Perhaps the most important piece of advice from the interview is therefore: take your time to develop good prompts.
A prompt does not have to be perfect at the first attempt. On the contrary. Good prompts are often created iteratively. You test, refine, add context and improve the instructions step by step. This effort is particularly worthwhile if a task recurs regularly.
Why three hours of prompt work can still be worthwhile
A very pertinent thought from the interview was: Even if you spend three hours on a prompt, it can still pay off.
As soon as a properly developed prompt saves one or two hours every month, there is clear added value over the year. This is exactly how companies should look at AI: not as a magical instant solution, but as an investment in better ways of working.
This is an enormously important change of perspective. After all, anyone who expects AI to deliver perfect results immediately without any training is almost bound to be disappointed. However, if you understand that good results come from clear instructions and smart use, you will recognize the actual benefits much more quickly.
Using AI like an intern
A very vivid image from the interview was the comparison with an intern. This is often the best way to explain how to deal with AI.
If you give an intern an unclear task, you will rarely get the desired result. However, if you explain clearly what you need, what the context is and what the result should look like, the quality increases significantly.
The same applies to AI. Good results need good instructions. Anyone who understands this logic will use tools like the SKI Pilot much more realistically and successfully.
The real success factor is human hand-holding
In addition to all the technology, one point remains particularly important: the introduction of AI requires support. People need to be picked up, empowered and motivated.
Employees who are not already deeply involved in the world of AI need examples, security and space to try things out. They need to exchange ideas about what works and what doesn’t. And they often need someone to work with them to make the first real added values visible.
It was precisely this human hand-holding that became very clear in the conversation. And this is probably one of the biggest success factors of all.
Conclusion: The SKI Pilot shows how corporate AI can be really useful
The SKI Pilot is a good example of how an internal AI can develop from an initially disappointing solution into an increasingly valuable work tool.
The key to this is not just better models, but three things above all: secure framework conditions, close integration into existing systems and active support for employees.
Companies can take a lot from this. Introducing AI does not mean simply providing a tool. It is about identifying use cases, empowering people and continuously making added value visible.
This is exactly when AI becomes not just a trend topic, but a real productivity lever in everyday working life.
Any further questions?
Do you have any questions? I am happy to support you, act as a sparring partner and answer your questions. I am always happy to receive your messages, preferably by WhatsApp message or e-mail.
