AI in finance: between hype, hope and hands-on experience

This interview is based on an episode of the torq.partners Finance Podcast featuring guest Christophe Wagner, a principal at torq.partners. The full episode is available on Spotify and YouTube as a video podcast.

Christophe, you've set up a pilot group at torq.partners that's going through an internal AI training program. What's your approach to this?

It is important to understand that the issue of AI cannot simply be delegated to individual people. Lately, we’ve been hearing a lot about the so-called Chief AI Officer, but I think that’s problematic. AI is not a tool that can be left solely to an expert—it is a fundamental technology, much like electricity or the Internet. We must not outsource the learning curve; rather, it affects all employees.

That doesn't mean there shouldn't be enthusiasts who drive the topic forward and inspire others, but the most important thing is for everyone to try it out for themselves, experiment, and even make mistakes. AI won't take hold in the organization overnight. People need to experience it for themselves, feel the magic, and understand how AI supports them.

How do you introduce the topic of AI to the team without overwhelming your colleagues?

The easiest way to get started is to talk about how AI is actually used—that is, the specific use cases for which it’s applied. For example, we have an internal Slack channel called “Prompt and Pray.” This exchange is important because no one can know everything about AI on their own. The technology is evolving so quickly, and its applications are so diverse. That’s why it’s crucial to share experiences—such as which prompts work well.

You mentioned three levels of AI application. Can you briefly explain them?

Level one involves direct interaction with AI—that is, traditional prompting to gain knowledge or generate text. The best way to learn this is by trying it out for yourself.

Level two involves the automation of processes. For example, an AI automatically classifies incoming emails, identifies invoices, reminders, or spam, and forwards them to the appropriate system or team. Here, there’s no need for constant interaction with the AI; it handles certain tasks on its own. These are known as agents, and 2025 is expected to be the “Year of Agentic AI”—even if the necessary infrastructure isn’t quite ready yet.

Level three is programming with AI. This involves using AI to build your own tools or solutions. This is particularly exciting for areas like the finance department, where you can automate recurring processes and turn them into products, rather than having to adapt to existing software such as project management tools.

For CFOs who are just starting to explore AI: Which tools do you think are particularly important?

I recommend three types of tools:

First, a versatile AI tool like ChatGPT, Claude, or Gemini. You should dive deep into it to understand its capabilities and regularly try out new features.

Second, a tool that helps you organize your knowledge. Obsidian is a good example here, especially when used with an AI plugin. Obsidian highlights the connections between your notes and helps you reflect more effectively and recognize patterns. If you prefer a more visual approach, you can also try Notion or Tana.

Third, speech-to-text tools like Whisper or Super Whisper are very handy, especially for CFOs who travel a lot. Not only do they transcribe spoken words, but they also allow you to set up custom prompts so that the results are immediately formatted and ready to use. This way, for example, you can capture ideas for a LinkedIn post or a meeting right on the go.

That sounds like you have to actively engage with AI and not just rely on a single person in the company.

Exactly. AI is a fundamental change, not just a simple software solution that can be delegated. Everyone should get a feel for how AI supports their work and where the opportunities lie. That’s the only way to achieve a sustainable transformation.