It was the sweet moment. Scott Russella member of SAP’s executive board, had begun by warning that “sugar is part of life, but it also generates incredible economic growth.”
It referred to the activity of the Royal Eswatini Sugar, a company that exploits two thirds of the sugar in Eswatini (formerly Swaziland) and also produces 35 million liters of ethanol as fuel. Still, your IT manager, Rob Coombepresents it as “a small company, with 6,000 employees, that manages seven villages, schools and clinics…”.
Coombe adds that they are in the process of digitization and accommodation in the public cloud, which is why it was one of the clients invited by SAP to its event. Barcelona Sapphireand concludes that they have achieved a 15% reduction in IT spending, with greater efficiency in their processes.
“It’s just that we have a very cheap technology“Russell bursts out with happiness, amidst applause from the audience.
These are things that happen in this type of massive meetings, with almost 5,000 attendees, when the organization manages to present grateful customers who, in addition, know how to gracefully recount the vicissitudes of living in remote places, 70 kilometers from the nearest store, and depending on trucks that occasionally require repair. The cloud seems to bring everything closer.
This may also be the case of a Spanish wine producer founded in 1835 in Jerez de la Frontera, such as González Byass. Your SAP project manager, Elizabeth Braza, tells how they have optimized, through 100% recycling of glass, plastics and paper, bottling and packaging processes. At the same time, it deals with the plastic tax, replacing it as much as possible in its packaging with bioplastics from sugar cane.
Your contribution to the event has to do with the management of the supply chain through RDP, which can also be applied to its production centers in Chile and Mexico. Sustainability and efficiency are the key words of the session.
Success stories
And so, other clients like Beyond Gravity, Coop or Kone, they recount their success stories, stimulating a business audience that was largely already aware of technological developments. SAP had presented them a week earlier at the global Sapphire in Orlando.
There, for example, the agreement of collaboration with NVIDIA to develop the software AI Enterprise and integrate into SAP software Omniverse programming software, from the chip manufacturer, to develop interfaces or APIs specialized in, for example, translating between different programming languages used by AI developers.
All of this, focusing on the main novelty, which is, of course, generative artificial intelligence. SAP announces that it “embeds and customizes” it in all its processes, applying the generative AIs existing in the market, through “Joule, which acts as a co-pilot for everything” in its business technology platform, as explained by the CEO Christian Klein.
DISRUPTERS posed three specific questions to Klein about JouleSAP’s own AI, whose role may at first seem somewhat confusing.
The explanation that the platform “integrates” other AIs, such as Copilot from Microsoft, Gemini from Google and so on, through its Joule generative AI, is it because Joule acts as a kind of interface?
“There are a lot of co-pilots out there. Every technology company says theirs is the best… and the difference is that when you’re working with the SAP system you don’t have to integrate the copilot core that you want to use. You can use the tool out of the box and you don’t need someone from IT to integrate it. But [Joule] It is not a mere interface: [la IA ‘externa’] is deeply integrated into our product,” Klein responds to the second question.
The first was about Joule’s ability to speak languages. Right now only expressed in English. Klein clarifies that “many other languages will be available in the third quarter. And starting in July, Spanish will be one of them.”
The third answer explains where the name that SAP has applied to its AI comes from. That Joule refers to the energy unit which in Spanish we call joule, which measures the amount of energy or heat. Which means that AI “releases energy,” adds the marketing manager. Julia White.
One of the talks held during the Barcelona event.
The use of the name in physics pays tribute to the British James Prescott Joulewho in the 19th century investigated and made great contributions in the field of thermodynamics.
Regarding the AI called Joule, the head of product engineering, Muhammad Alam, has something to add: “We have ready-to-use scenarios, but we continue to test to verify which large language model works best, to automatically select the most suitable one, the one that can produce the most reliable results.” It’s part of Joule’s mission.
Alam clarifies that the concept of “scenarios” has nothing to do with orchestrating a company’s specific workflow. “It’s very difficult to apply AI at that level,” he says. On the overall task he suggests the need to have a “holistic view.”
What they offer scenarios “they are steps”, concrete tasks for generative AI.
“Of the OpenAI models, we are working with Microsoft Azure, because it offers a lot security and privacy. We have to ensure that our customers’ data does not flow into the public model, or Microsoft’s. And with other models we are doing the same.”
Alam states that, in the development of APIs to connect its clients with different intelligences, the data isolation and the residency requirements they may have. “If some type of model needs to be generated with customer data, it is done on our site, without contributing to the public model.”
50 scenarios
“The nice thing is that we integrate the applications so the customer doesn’t have to worry about it. We make sure choose the most suitable AI in each case. The client only has to think about it if he is building his own scenario and needs to test for himself which model gives the best results. And he doesn’t have to make contracts with various entities,” adds Alam.
But, we ask, the client must then be very clear about what he or she intends to achieve. Which is not always easy.
“The scenarios we work with are validated by the first clients who have used them and have provided us with feedback. Now most can simply consume them“, he assures. Communication, through Joule, is done in natural language.
To clarify a little better what is meant by this concept of scenarios, Alam puts a simple example: “You must budget for a trip and to do so you need to visit various sites, compare how much flights may cost and other things. Now, you ask an LLM to make the estimate, on certain dates, and they solve it. The client does not need to think about it. “If the stage is worth it, you should just use it.”
“We have about fifty scenarios available, and at the end of the year it will be a hundred, for the direct use of our clients. If one fits you, you can use it. And if you need something unique, which is not among those hundred, you can build it for yourself in our application. We have hundreds of customers who are already using them,” he adds.
On the other hand, he also emphasizes that SAP is building “a foundational model” from AI, which by the way has no assigned name, “trained with data from clients” who have authorized it.
In the scheme that the company draws, of its technological platform with artificial intelligence, three layers appear. At the base is this foundational model and in the intermediate layer are the “embedded” generative AIs, with access through what is called Generative AI Hub. The top layer is occupied by the co-pilot Joule, as controller of everything.
Klein emphasizes that all this is included in a single contract, the one that each client signs with SAP.
Philip Herzighead of AI (the acronym of his position is curious: CAIO), specifies that what this structure aims to do is, first of all, give “a direct advantage to clients with a new tool.”
“Secondly,” he continues, “the first layer [Joule] does it is not necessary to have a data scientist to benefit from AI. “It is the only platform that gives access to all AI models, with the best and cheapest technology.”
Its third point is the promise to apply ethical and responsible criteria in this use of artificial intelligence. SAP assumes the Ethical principles for AI recommended by UNESCO.
