Leadership Insights: AI in the Enterprise

Transcript

Amanda Razani: Hello, I’m Amanda Razani with Digital CxO. I’m excited to be here today with Yaad Oren. He’s the managing director of SAP Labs US. How are you doing today?

Yaad Oren: I’m doing great, Amanda. Thank you for having me.

Amanda Razani: Glad to have you on our show. Can you share a little bit about SAP Labs and what services do you provide?

Yaad Oren: Yeah, absolutely. Maybe first, a really, really short introduction to what SAP is doing. Basically, we are doing things. We are selling and leading the enterprise application market. When we are selling application around financial solution, manufacturing, and so forth. We’re also building platform product, database and a leading integration, AI, which I’m sure we’re going to talk about. We are leading those markets for the recent 52 years.
The Labs is the nickname for our development organization. We are a European company. We are based in Germany. I have the honor to lead the Labs in the US, where we are talking about 6000 women and men that build product around two dimension mostly, around the cloud solution, where we are the leading application as a SaaS model. Also, a lot about driving innovation, embedding AI and very advanced technologies in our portfolio. This is what we are doing in the Labs, and specifically in Labs US.

Amanda Razani: Thank you for sharing. I believe you recently came off hosting the TechEd 2023 Conference, I believe?

Yaad Oren: Right. We had amazing TechEd Conference. This was the biggest conference to date that SAP did, when we came up with a lot of interesting announcement and engage with our ten-thousands of customers and partners.

Amanda Razani: From your experience, and in speaking to some of the business leaders there and over the last year, what are some of the key issues and concerns in regard to implementing new technologies and digital transformation initiatives?

Yaad Oren: Yeah. This is a great question, Amanda. I think that, first of all, from my observation, there is a huge hunger to experiment new technologies, especially around generative AI. This was the main theme and the main interest from huge ecosystems.
When it’s come to the main challenges, I can divide it into two. First one is around the reliability and responsibility of the technology. It’s no secret that generative AI is going like wildfire. Also, many other advanced technologies can really give great result, great output. But in the end of the day, they’re not always fully reliable when you run them in production. What does it mean? It means the result, the output can be really misleading. In AI, all of us are familiar with this term hallucination, when sometimes the system just invent facts. Customers are really bothered by that.
Also, by the fact that it might be some data privacy violation or some security breaches. You start moving data around, you’re more exposed. While the north stars look very promising about AI result, the [inaudible 00:03:34] there is still need to be built, especially about responsibility and reliability.
The second thing is very simple, like any customer, is around cost. Advanced technologies, especially around gen AI, are super cool but also come with a price. Then, you need to make sure you are really targeting the AI or the technology to solve a dedicated problem. You don’t shoot the fly with the cannon, not take too much to solve a problem that can be solved with many maybe a little bit more down-to-Earth technology and with a lower cost.

Amanda Razani: Okay. Let’s take the first key concern, which is that reliability and the security. What are some of your key tips or advice, when it comes to making sure that it is reliable and it is secure?

Yaad Oren: Fs, it might surprise you, Amanda, but many enterprises doesn’t have what we call data strategy. They are collecting data, it’s being stored in silos. It’s being stored in different environment, which sometimes you don’t have a single pane of glass to view it. If I need to give an analogy, sometimes the data flow into the organization like a Brownian motion of molecule bumping into each other.
In order really to reduce and optimize the reliability and security, you need to have a strong data strategy, which means very simply, data governance structure that you can observe all your data. You need to have some sort of policies regarding the quality of data. A lot of the data in the system, that goes back to your first question, result in hallucination and lack of reliability is because the data is not a great quality.
I can give this a silly analogy that you’re only dependent … Like any gourmet dish in a restaurant, it depends in the ingredients you have. Also, in AI. If you don’t have the right data ingredients in the right quality, the result you cook out of this probably not going to be the best satisfying.
You need to have a strong data strategy about the quality of the data, how you govern it. You need to actively look in errors in data which is not normalized or standardized. You must have a single view on your data so you know, in any given moment, what’s happening in your system. This is something many customers missing or not having to the right extent, to be living in a data-driven world.

Amanda Razani: On the second issue then, cost, what is your advice as to, when business leaders are looking to do these digital transformation initiatives, what first steps should they take so that cost is kept in mind? Then, how do they monitor their return on investment?

Yaad Oren: Yeah. First of all, we need to define very clear goals, like how success will look like for what we are building. We need to have in mind what other options we have to reach there, because sometimes if you take the AI world, and 90% of today world, and I’m quoting some of the greatest mind in AI, can be solved with what we call supervised learning, a very basic, down-to-Earth machine learning from 10 years ago. You need to have a very strong aspiration and a business goal, and then experiments a different way together that really justify to even consider such advanced technologies, and a little bit vulnerable technologies in the first case.
Once you have this, I really recommend experimenting with different technologies within the generative AI. In SAP, we have a great what we call an AI Hub, where we give access to all the models out there. Whether it is OpenAI, cloud, bread and so forth. So developers and customers can structurally evaluate the different options and benchmark in order to reach the most satisfying business need they need, and also with the optimized cost.
This is very important, and the last point is pretty simple. Many vendors are now providing cheaper options. We are going something similar to what’s happened with blockchain technology, that started as a very expensive technology, consuming 2% of the world’s electricity. Then, they changed the proof of concept, the proof of stake, the model to be much cheaper. We see here now, companies providing cheaper options to run the algorithms on better infrastructure.

Amanda Razani: That being said, this technology is advancing quite rapidly from when it first came into the public view over a year-and-a-half ago on such a large scale. All the business leaders are trying to harness this technology. It is more affordable now. What do you envision is the future of this technology, say a year from now?

Yaad Oren: Yeah. A couple of points. One of the things where I see a lot of progress and a lot of hope is the multi-modality of the technology. At the moment, if I over-simplify, we’re all familiar with OpenAI and ChatGPT. Both those runs more in scale on text, but gradually, we could start consuming every type of information, including videos, including audios. We are seeing the first springs on those kind of things, but eventually this technology will grow into consuming any digital media, any information. Combining it together to provide really holistic results, not only text-centric.
Another thing, which is definitely first over priorities, anything about the security. I will add also here, Amanda, the compliance. We have a lot of regulations coming up. If you ask me for my deepest concern, is not the vulnerability and risk in the technology itself, it’s about how evildoers can take this technology and run it to do malware and phishing attempts. This is actually very, very scary to some extent. Here, the good news, we’re seeing a lot of progress, especially in the US lately, with regulation. We had the President Biden AI Act in October. Of course, we work in European companies, so we are pretty advanced from the standardization already, as the EU is very advanced with regulation too. But the point I’m trying to make is definitely we’re going to see both security and reliability improvements, together with compliance and regulation, that’s going to limit the risk farther.
Yeah. The last point, that definitely this technology will progress, I would say, is the efficiency. We talked about the cost, but we haven’t discussed yet about the skills you need to build a gen AI application. We see a lot of design time, of the development time, of the development experience being simplified with gen AI. But still, it’s a pretty complex work to build advanced AI application. I definitely see, with many innovation around pro-code and low-code developers, and low-code and no-code developers even start to touch this kind of AI world. But definitely, we’re going to see a better developer experience and more streamlined approach to building applications.

Amanda Razani: There are some tools to help with that. But in general, what are your suggestions to help business leaders in staffing? Do you suggest that they offer some in-house training? Just what advice do you have?

Yaad Oren: Yeah. I think, first of all, you need an executive champion for this, a sponsor. Because otherwise, this can go to different direction and anyone call pull the blanket to a different direction of what you do with AI. The organization really needs to have dedicated cause, “Okay, we would like to improve our productivity of customer service in 10%. We would like to examine to reduce the cost in supply chain in 20%,” and so forth.
Then, we definitely need to start upskilling the workforce. I think it’s pretty bold saying this, but this is also SAP assumption and the industry, that if you’re not part of the AI game, you’re probably going to be leapfrogged by competitors who are going to take those different arrows in the quivers and start addressing new targets. Upskilling and the training you mentioned is very important. I think for both the developer force, but something important, Amanda, also for the end users. Another potential risk is now that we are opening this treasure trove of data and insight for any role in the organization, anyone can generate reports. You don’t need to be an analyst to really generate great report and insight. If you touch this data, also end user needs to understand when something is hallucinated, how to really send forward the reports. Both for end users and developers, definitely upskilling, and be ready for these new IT and business reality is a must.
Yeah. The last point is what we mentioned before, is having really strong security and reliability guidelines. I will a thing, Amanda, the ethics. Corporates really needs to look into ethics and bias consideration as well, and put some guardrails that will make sure that this technology will be harnessed to create a better world, not only a better business.

Amanda Razani: Absolutely. Well, if there’s one key takeaway you can leave our audience with today, what would that be?

Yaad Oren: I will definitely encourage enterprises to start upskilling the workforce into the AI revolution. This is important, Amanda. It doesn’t mean that anyone needs to be a builder and develop applications, but you need to start understanding we’re going to live in a data economy and in a data world. The value change of building applications, the value change of having solution in enterprises is going to change. The difference … SAP has hundreds of certification and learning as well on these kind of roles, but this is something that we are heading to a paradigm shift in how we build application, how we consume application, how much we rely on them and how we identify them. Have a really strong sponsorship and a plan for the learning and upskilling your organization will help to prepare them to this business reality.

Amanda Razani: All right. Well, thank you for sharing your insights with us today. I look forward to speaking with you again soon.

Yaad Oren: Thank you, Amanda. It’s my pleasure.