Revising texts, gathering ideas, answering questions, and occasionally seeking medical advice or having a best friend stand in: ChatGPT was already thoroughly familiar with these tasks. The author was also aware that artificial intelligence could do quite a bit more. But she had found it difficult to imagine how entire workflows could be automated with it.

That changed at the beginning of November during the three-day course "Certified Applied AI Expert" (CAAIE) at HWZ. Under the leadership of Kimia Kiser, participants explored the possibilities and pitfalls of artificial intelligence in 16 modules. Daily rotating tutors contributed different perspectives and practical experience. Their own laptops were constant companions throughout. The continuing education program costs 2,200 francs and concludes with an exam. Those who pass receive a certificate.

Using the code FINEWS, all readers receive a CHF 200 discount.

Kimia Kiser, head of the CAAIE course and co-founder of Kimiana. (Image: provided)

ChatGPT, time to get to work!

A prompt is quick to write. But what happens when AI is supposed not just to carry out individual instructions, but to take over recurring tasks? In the course, the tool used was not ChatGPT, but its competitor Claude. A fictitious company was used to test this. Participants fed Claude the fictional business documents, had it draft emails to employees, and taught the language model to maintain a consistent tone in customer communications.

Not everything that sounds convincing is true

For all the enthusiasm for experimentation, the downside of the technology was also addressed. AI can not only take work off people’s hands, but also produce errors, sometimes with remarkable confidence.

A traffic-light model illustrated where its use makes sense and when caution is still required. The green zone applies to those who condense texts, gather ideas, or have complex issues explained to them. Yellow applies to research, calculations, and data analysis: here, the necessary expertise is required in order to verify results. Red applies, for example, to confidential data without appropriate safeguards or sensitive personal communication, such as terminations. In such cases, responsibility should remain with humans.

For the financial industry, this distinction could be particularly relevant. Anyone working with confidential client data must not only know what AI can do, but also which information may be entered into such a tool in the first place.

Learning by doing: In the AI course at HWZ, participants were able to try out what they had learned directly on their own laptops. (Image: provided)

A digital sleuth for finance stories

How the new knowledge could be used in her own day-to-day work became clear during the course itself. Using ChatGPT and Claude, the author developed an automated topic radar for finews. Since then, the tool has searched the internet every morning at a set time for developments that could be of interest to the finance portal’s readership.

Just in time for the start of work, the automation delivers up to five topic suggestions, providing possible starting points for deeper research. Whether a story actually emerges from them is still decided by the editorial team.

The digital sleuth is already in use, but it also has its limits. Paywalled content, such as that of the Neue Zürcher Zeitung or Bloomberg, remains out of reach without the appropriate access. The topic radar is therefore to be further refined and expanded with additional functions.

Nikolai Madlener, tutor at CAAIE and co-founder & CTO of miora. (Image: provided)

"Finanzfabio" is not alone

It is not only media professionals who are considering how AI can make working life easier. Since the first CAAIE session in June 2026, around 20 people from the financial industry have taken part, according to the organizers. Among them were a chief investment officer, private bankers, relationship managers, as well as executives from asset management firms and family offices.

Swiss financial influencer Fabio Marchesin, better known as "Finanzfabio", also attended the November session.

Fabian Marchesin, "Finanzfabio," was also along for the ride. (Image: provided)

The applications of AI in the financial sector are diverse: preparing client meetings, reducing extensive research reports to the essentials, drafting presentations, or analyzing Excel data. Especially for repetitive tasks, the technology promises relief.

However, banks often have strict rules governing the use of AI. The methods taught in the course can, in principle, also be transferred to internally approved tools. Whether and how they may be used depends on the respective policies.

New tasks for old acquaintances

During the three course days, the author realized that with the right tools and the necessary know-how, a simple chatbot can become a rather self-sufficient assistant that takes many tedious tasks off her hands. But without human oversight, it still cannot do everything. And despite all its newly gained independence, AI still has its limits:

You still have to get the coffee yourself.