When I first spoke to Juergen Mueller, SAP's chief technology officer, earlier this year, generative AI was the hot new thing.

Nine months later, "a lot has evolved. Everyone learned a lot and natural language models are amazing and they can do a lot of things," Mueller told Business Insider in November.

One issue with language models and generative AI is that it often has a cutoff point and lacks up-to-date knowledge. The German company — which focuses on cloud-based subscription services for its financial-reporting, inventory-tracking, and human-resources applications — looks at the processes of companies, including financials, procurements, HR, customer relations, or the supply chain.

"We develop a method that you can combine the strength of a large language model with all the real-time, company-specific information that you need," he said.

Mueller explained it to me this way: Imagine if I gave a colleague all the articles I ever wrote, and asked her to write one article summarizing everything, and "write it in a style like I always write." The SAP cloud is similar. Instead of articles, it has access to all employee records, or invoices, or emails — other important documentation necessary for a company's daily functions —and can use that data to complete tasks faster.

SAP is giving the large language models "means to access the know-how," Mueller said. SAP has trained more than 50,000 of the company's 105,000 workers on what it sees as the opportunities that Gen AI brings.

SAP has worked within the machine-learning AI space for almost nine years. "We have a lot of components and process in place," he said. The company has been trying to answer questions around how to work with data; how to protect data; how to use data ethically; what tools data scientists need; "and how can we rethink and upgrade to generative AI."

Mueller opened his iPad and showed me one of the tools at work — taking financial slides and summarizing them in an email for an executive meeting. It also gave data points and a widget. It basically did hours of work in about 5 minutes.

Of course, security is central to what SAP does, as is checking for bias in data.

Mueller views it all as an opportunity. "Technology will not stop and what we have to do — everyone in every role — is we have to continue learning," he said. He compared AI to past conversations about truck drivers becoming obsolete, which didn't happen. He referenced the movie "Hidden Figures" and how there were human, and not mechanical or digital, calculators. Then there were maps: paper vs. Mapquest vs. Google maps.

"There is risk, always. So we have to acknowledge that, work with that. But also then see what is the opportunity," he said.

Mueller's insights are part of Business Insider's year-end leadership package, "Looking Ahead 2024," which digs into vision, strategy, and challenges across corporate America.

The following interview has been edited for length and clarity.

What are you most excited about for 2024?

I look forward, in 2024, to seeing all these things in real life, in action. So, as I said, many cases that customers built — and also that the industry built — overall are in preview mode. And now actually it's "the rubber hits the road" in 2024. And I'm very much looking forward to that.

What are you most concerned about for 2024?

Regarding the adoption of AI, including programs like ChatGPT, language-learning models:

What I'm most afraid of is that globally, we are not continuing to learn fast enough. And I think the risk is not that there's new technology. The risk is that we don't lean in and don't invest the extra time, extra hours. Because it needs a little bit of time; it is a little complex. ChatGPT luckily makes it easy to access that so that is good. But in a business function, it is still relatively complex to deploy this technology. Therefore everyone needs to lean in, learn. That, I think, is the largest risk. Then you have divergence in society.

The worst thing is if only a few would know how and what to do with it and the majority wouldn't. That's, I think, one of the large worries. So you are very important because you're an amplifier and people read a lot about it and people should be optimistic. But also, it should be clear that you need to invest a little bit of time, like with every skill, to be good at it.

What's one thing you got right in 2023?

In 2023, the R&D we've invested in business AI since 2015 paid off, with over 25,000 SAP Cloud customers using at least one of our 130 AI scenarios, the launch of a generative AI hub, and the addition of vector database engine capabilities into our solution offerings. This strong foundation of AI expertise has been enabling us to infuse generative AI capabilities across our solutions and deliver immediate value to customers.

What's one thing you got wrong in 2023?

I underestimated the need to meet our customers where they are. This year, it became clear that in-person interactions — such as our annual developer and customer conferences at locations around the world — are as important as virtual event options. For us, striking the right balance between virtual and in-person meetings is key to fostering a dynamic and effective collaboration and work environment.