Since the dawn of time, history has been defined by serious and world-changing innovations. These groundbreaking moments were not always met with the appropriate levels of enthusiasm.
Well, that's the end of the raw food industry.
It's called a sale.
I know. We've always been more of a rowing-centric organization. We have... I call it inoculation. Clever. But we have a lot invested in leeches right now. I mean, leeches.
That's the end of their careers. No doubt. No more mainframes.
Well, that's the end of big data.
What is this called?
The internet.
I do not like it. They're calling it software as a service. You think people are going to rent software?
SAP HANA. It's the end of legacy enterprise tech stacks with siloed ERP and analytics. I don't even want to imagine a world without legacy enterprise tech stacks with siloed ERP and analytics.
So we're an autonomous enterprise now. You know what that means?
It's the end of busy work, bottlenecks, and silos, and the beginning of faster, simpler, better.
Welcome to the Autonomous Enterprise. It's not the end of anything. It's the beginning of better.
Please welcome to the stage Christian.
Wow. Hello everyone. Welcome to Sapphire 2026. Welcome to all our customers, partners, media, and analysts, and of course to my best friend. It's great to be back in Orlando. Sapphire is such a fantastic opportunity to connect with people from all around the world and to feel the energy of the wonderful SAP community. Especially in times of incredible change and speed, Sapphire is more important than ever.
Some time ago when we started to prepare this keynote, I asked myself the question: will SAP actually be a software company in the future? Are you scared by this question? I'm not scared. For me, the time right now is the beginning of something even better. And today, I want to show you why. So, let's come back to this question at the end of the keynote.
There is no doubt that AI is changing everyone's life. Starting with my own life. As a CEO, I use AI for all kinds of tasks. I use AI to bounce around ideas for our next strategy refresh or to analyze the market sentiment before earnings and to prepare my customer meetings or this keynote today. And as a dad, I use AI to draw pictures with my kids. And AI does certainly a much better job than I could ever do. But take a closer look at this unicorn. The unicorn is not perfect. It actually has three ears. My daughter told me that at least it doesn't have two horns. So for her, an 80% accurate drawing of a unicorn is still good enough to make her happy.
In business, it's different. If AI agents run your payroll, your financial close, your demand and supply chain planning, 80% accuracy is just not good enough. When you run the world's most mission-critical businesses, your AI agents, they should not guess. They should deliver accurate, reliable, and compliant outcomes.
According to a recent Stanford AI survey, almost every company is now using AI, but many see only little value. Why do we face such huge challenges with AI in business? At the top of this iceberg, visible to everyone, is that large language models are getting better and better at tasks like generating text or images or in specific domains like writing software. All of these use cases are related to publicly available content the models are trained on. But if you go below the waterline, beyond the level of sales tables and into the real business world, you're going to find out that none of these models are trained on your business data and processes. These AI agents also don't naturally adhere to governance requirements like your security compliance framework, your data privacy requirements, or to your company's identity and authorization rules. All AI agents, including Joule, have faced these challenges until now.
Let us now show you how SAP will solve this challenge. The path to the solution starts with the brain of every company. Do you know which brain I'm talking about? 100 points. The brain of every company is the ERP. For over 15 years, we have been developing an ERP with incredibly deep process and data domain know-how. On top, all your governance requirements and customer-specific extensions are stored in the ERP. The ERP is the trusted system of execution running your company. So, how can we now infuse this deep business domain know-how sitting in your ERP into our AI agents? Let me illustrate this with the following prompt. First, I'm asking Joule to write me an update on my total year financial forecast with the latest sales and pipeline data plus supply chain data. To understand my question, the prompt is processed by a large language model. So far so good. But going forward, the LLM now has access to the brain of every company, your ERP, to first find the right process from thousands of business processes in your company. Then with the knowledge graph, select exactly the right data from over 7 million data fields stored in your ERP landscape. And finally, before the outcome is shared, we check all your identity and authorization rules to ensure the outcome is not only accurate but also compliant.
Well, the team delivered. Building nothing less than a new SAP. We are bringing together LLMs with 50 years of business know-how stored in our ERP by merging our AI foundation with our business data cloud and business technology platform. And today I'm super proud to launch our new SAP Business AI platform, which forms the basis for our vision of the future of business, the autonomous enterprise, where agents run the business and you can focus on what truly matters.
Let me start with the foundation. The heart of this new platform is the rich context layer. Here we infuse deep ERP business domain know-how into the AI agent. Through our knowledge graphs, our AI agents have now a compass, a map to find the right process and data in your ERP universe. And to provide the agents even more context, we are also introducing our new SAP domain models. They have been trained on SAP's code to even better understand the business logic of your company. But we don't stop here because you run your business not only with SAP solutions. Actually, why not? Our AI agents have to also understand non-SAP data. That's why we included our SAP Business Data Cloud in the context layer to build one semantic data layer across SAP and non-SAP. No more silos. No spaghetti data squall because no AI agent can compensate for a poor data model.
Let's move on to the build layer of the new platform. Here we are introducing Joule Studio 2.0. Our new build layer is open and agnostic. You can bring your own LLM or you use SAP's own models. You will be able to develop agents faster than anywhere else. Because Joule Studio 2.0 is integrated with the context layer and has direct access to all of your business logic. And finally, in the governance layer, we want all SAP AI agents for you so you don't have to deal with the complexity and governance requirements sitting below the waterline of the iceberg. Via our AI agent hub in LeanIX, we also invite you to actually maintain and provide together all the agents running in your company and we will govern non-SAP agents for free. So let's take a closer look into our new SAP foundation, the SAP Business AI platform. Sure, can you help me introduce the next speaker?
Of course, Christian. Please welcome to the stage Philip Herzig.
Good morning and thanks Christian. We are so excited about Business AI platform, a platform that is designed to close the adoption gap by delivering outcome, speed, enterprise readiness, and of course all the context you need. It's the place where you build, contextualize, reason, and govern AI end to end. And let's have a deeper look into build. In fact, before we actually start building anything, it would be great to know what to build, what has the highest impact on the business. In the past, you had to do time-consuming research or do this thing we proposed called AI discovery workshops. Now it is as simple as typing, 'show me my top three business challenges.' Now in this case we are a pharma distributor. We move a lot of products around and need to align prices between manufacturers, hospitals, and purchasing organizations. Complex supply chains, razor-thin margins. Now we have this new thing, the process consulting agent from SAP Signavio, that goes deep into the actual business processes of that pharma distributor to show it what has the highest impact. In this case we have a huge margin leakage, this year alone 24 million at risk. A great use case where AI can help and the proposed solution, a sales pricing validation agent that helps to resolve those inaccuracies.
Now where do we build this agent? Of course, in Joule Studio. It's the place where we extend, build, and integrate AI experiences like never before. Since its launch, customers and partners have built phenomenal things with Joule Studio already, like ABB that now handles more than 15,000 RFQs automatically every year, saving them millions. These are just a few great outcomes, but we heard your feedback loud and clear. You all wanted more. And today we're super excited to introduce Joule Studio 2.0 to build agents for business outcomes. Now what is this new Joule Studio? First, it's intent-based so it really understands your business challenge and it's deeply contextualized in your business data and business process know-how. It of course comes with out-of-the-box connectivity to all your apps. And it guides you through full-fledged product requirements documents and technical specifications. Also super important, it's model-based, so you can actually guide in a data-driven way your AI coding assistance. And it's SAP managed and comes with all enterprise qualities baked in. In addition, Joule Studio is open by principle. You can choose from any third-party AI model. We have them all on the platform already. Of course, also SAP's own AI models and you can build in the environment of your choice. No lock-in and of course your style and develop low-code, pro-code, everything in between.
Now, let's build something together with one of our amazing partners. And for this, I'd like to invite KPMG on stage, Rob Fischer from KPMG, a global vice president right here. Rob.
Yeah, well, first it's great to be here, Philip. And that's right, we're living AI transformation ourselves and we're learning from it every day. We're a Grow customer and we know what it takes to modernize core systems on S/4 public cloud while still keeping the business moving. We've got a final rollout with about 270,000 users and we moved early on AI. We deployed Joule for consultants across 34 countries and more than 3,000 of our consultants are using it already and we're co-innovating with you. Our teams have built many agentic AI solutions, including 20 agents on the new Business AI platform. And for one of the clients we're working with, we're targeting $120 million in reduced contract leakage. That is real tangible value.
Absolutely. It's amazing value. Hey Rob, KPMG is a global leader in AI as well and also of course very well recognized for its high quality advice. What is the number one thing you're seeing here with customers?
Yeah, I think it's this clear shift, right, from pilots to integrated AI agents. You know, customers seeing AI not just as workforce efficiency and cost savings, but as core to competitiveness and a driver for unlocking more business value. And where we're really seeing the leaders really separate from the pack is in the execution and the organizational adaptability. You see leaders pairing the AI investment with the workforce transformation and then embedding AI where the decisions and the accountability sit. And because most enterprises run on multiple platforms, that AI has to orchestrate across the ecosystem with trust built in from day one. And hey, that's where SAP comes in.
Rob, thank you so much. That was a great setting up the stage. And now we actually would like to implement the pharma distribution scenario of the margin leakage we just saw and gladly one of your colleagues Chris gladly accepted to do the demo. So we don't have to do the demo, you do the demo for us. So that's amazing. Thanks for the partnership. And Chris, let's get you right on the stage.
Thanks for being here. Hi everyone. I'm excited to show some magic and really build something powerful for an organization. Here we can see Joule Studio where I can extend, build, and integrate agents. And what I really love about it, it's fully integrated with the SAP ecosystem and I can build enterprise-grade agents without writing a single line of code. Joule has already taken the findings from the process consultant demo that we showed before and I can use them to build the new pricing validation agent. This is where we tell Joule what we need to achieve rather than how we actually do it. It's a game-changer for how we deliver value for our clients. And this is where we start to see Joule acting less like a tool and more like a colleague and co-worker where it checks and asks us critical clarifying questions such as what's the source of our pricing data or what really drives success in this scenario? And as I'm going through these answers, one of the things that I think is extremely powerful is it's fully integrated with our agent landscape where in this case we can see that it's already identified two existing governed agents that we have deployed that already do part of the job. This is automatically included in our solution. And from this generated intent, I get transparency into what is the problem statement, what is the business context, what are the relevant source systems and data domains that are included in this, as well as taking a look at what is the recommended solution for this agent.
From all of this context, what I'm able to do is automatically generate the product requirements document that have historically taken us weeks or days to generate this type of blueprint that we now have at a click of a button and get in just minutes. What we're able to do is within this product requirements document is verify that this is the intent, this is what we need to build this agent as well as if we need to make any adjustments such as adding a margin threshold of 1.5% to make sure it has that baseline. Now it's time for Joule Studio to start doing the building and we can build the solution. And that's kind of really all of the heavy lifting that's done on my side. And we're able to go and generate and I'm just going to move over here as we generate the actual solution. And as we're going through this and as we're building this, what we're able to do is jump to a completed agent already. And what from this completed agent, what we're able to do is take a look at all of the unit testing scenarios that were identified and built, as well as take a look at the underlying solution with the integration with Signavio to see how each of these three individual agents are orchestrated together to solve this complex problem and being able to see what that entire end-to-end workflow looks like as well as we're able to dive deeper into the design and take a look at all of the configuration and capabilities that are built into this solution. Everything from what are the tools and skills to what are the individual evaluations that are needed for this solution. This is based on our data, our landscape and really what is needed to check the success of this and this type of solution is exactly what it takes to reduce pricing disputes and protect margin at scale. This is really how we turn expertise into agents that run the business from insight to impact to value. Back to you, Philip.
And that's Joule Studio that gets you to the outcome so much faster and far more accurately than ever before. We tested building this agent and many more on comparable platforms. And of course, while we cannot do the code generation faster, in fact, because of all the context, a little bit slower, Joule Studio wins in every other dimension it has been designed for, bringing you to the outcome 10x faster. And it does it at a far higher accuracy because it has the understanding of the SAP domain and it has the understanding of all the latest APIs you got to use and so much more. And finally, developers love it. We have given this experience to a thousand developers who tested the alpha version of it. And the feedback speaks for itself. And that's the next generation of Joule Studio starting to roll out to first customers starting June.
Next in the platform, let's talk about contextualize and reason because Christian mentioned it already. Agents are only as powerful as the context they are operating on. But the problem still is we know it, the data is fragmented. It's trapped in proprietary formats and of course it's not semantically aligned. Now Business AI platform starts on a strong foundation, SAP Business Data Cloud, which acts as the business data fabric that gives you universal business context across your entire multi-cloud landscape both for SAP and non-SAP data. Business Data Cloud is one of the fastest growing products in SAP's history. Customers like Baker Hughes or Salando are getting amazing business benefits from it already today. For example, by using any of the 300 data products that are coming out of the box managed by SAP. So you don't have to do the heavy lifting. However, these data products are coming only from the SAP world. But of course there's so much more data that needs to get integrated to empower our agents like customers coming from Salesforce, employees coming from Workday or any other application. This is where Relio is coming in allowing you to create the single golden record for all your master data end to end. So AI agents are able to reason over consistent master data. And to help you with your data modeling efforts, we are introducing Joule agents into BDC as well. For example, our all-new data products generation agent takes the heavy lifting away and the same experience we saw in Joule Studio to entirely create new data products or derive data products in virtually no time.
BDC of course is open by principle. BDC Connect, introduced last year, allows you to integrate and share with a variety of data platforms. And today, we're excited to add Amazon Athena to the mix as well. But we didn't stop there. We're so excited about our latest intent to acquire Dremio, a high-performance data lakehouse with cutting-edge data federation across any cloud and any traditional legacy database. Once this acquisition is closed, BDC will 100% support the open table format Apache Iceberg for plug-and-play interoperability with any data store. And combined with BDC, this will build the foundation to let your agents reason across data from any cloud, any environment without data movement.
Now with this, we have a solution to build up tables and data on the fly. And this gives us the next stage which is predictions on the fly. Because the opportunity in business AI was never just large language models. They're fantastic, but they're just one part of the equation. It is AI for decision intelligence, reasoning on structured business data. Predictions like demand forecasts in retail or predictive maintenance in asset management. To do such things, we have to rely so far on what we call narrow AI models, one for each task. Now, if you want to solve for 10 of such predictive tasks and say for 10 company codes, you have to train easily 100 AI models, months of work, and you need a lot of data scientists, which often times you don't have, and it's a significant investment, hard to scale and adapt. So what we did is we basically put all of that into one giant model which we introduced last year called Rapid 1, which is the most performant tabular foundation model on SAP data. And today we're excited to introduce the next model generation with Rapid 1.5.
This does not only allow for far better accuracy on tabular data, but also introduces an all-new chat API that allows you for on-the-fly what-if analysis and predictive simulation. And a big focus was on explainability. So you can always follow what the foundation model is doing based on the rows and the columns that led to the prediction. All of that without any AI model training because it just works on the fly. You can check it out today on rapid.cloud.sap. But there's one more thing because we are taking the next leap to the finest category. We're excited to share our intent to acquire Prior Labs, a frontier AI lab for tabular foundation models and bringing the latest TAP PFN model family to life actually today on their platform. While Rapid 1.5 is best-in-class for the SAP data, their TAP PFN model family is best-in-class for non-SAP data, unlocking industry-specific use cases, for example, financial trading. And these are just a few highlights of what's new and contextualized and reason in Business AI platform.
Next up, govern. You know the thing, right? Agents are now everywhere. Some are great, some are not. And almost no one really has a consistent picture. No central governance. An AI agent hub is the place that changes that. It's one command center to discover, manage, and govern SAP and non-SAP agents, MCP servers, and everything else in your landscape end to end. SAP AI agent hub is built on SAP LeanIX that's already used by more than 25% of the Fortune 500 companies. And today, we're taking it to the next level. AI agent hub not only allows you now to discover all your agents in context, your landscape or your business processes, but once you identified each of those SAP or non-SAP agents, you can actually control their risk, define architectural boundary conditions or compliance rules. And once you're done with that, you can mark an agent as verified. And Business AI platform and AI agent hub are making sure that only verified agents can operate in your landscape to ensure highest levels of control.
Furthermore, once you then deploy these agents, of course, you want to see what they are doing. We bring all the telemetry, all the observability into one place and now you can navigate seamlessly into Cloud ALM from AI agent hub. So you can watch their health, their goal completion or the tool calling correctness. In addition, you can then also from the AI agent hub go over into agent mining from Signavio so you can watch their behavior and find further optimization opportunities against your KPIs. And finally, we're also mapping all your agents into SAP SuccessFactors so you can see the impact these agents actually have on your workforce end to end. The full governance lifecycle of AI agent hub will be available as generally available in Q3 this year and is now part of SAP Business AI platform out of the box at no extra charge. And this is the platform the autonomous enterprise is going to run on. With that back to you Christian.
Yeah. Thanks, Philip. KPMG is a great example of how our ecosystem is already embracing the new platform. To ensure our Business AI platform will be the most loved platform in the market, we are making three further exciting announcements. First, all our SAP AI agents are open and exchange data with third-party agents without any limits and at no additional charge. Second announcement, we are excited that AI development with Joule Studio 2.0 will be available for free and we will also offer the one-time for free until the end of this year. And third, we will invest over 100 million euro into our ecosystem to build agents and drive AI adoption.
All of these wonderful partners are already developing agents on the new platform and look at their fantastic feedback. So let's hear more from our partner Anthropic.
We build Claude, the AI that enterprises trust most. At Anthropic, we build frontier AI systems. And I've learned that when the stakes are high, enterprises reach for the AI they trust. We've built the company around the conviction that as AI becomes more capable, enterprises need to understand how it makes decisions and to know it will operate within the boundaries they set. Earning that trust has costs. We've delayed launches over safety concerns and published research even when it's controversial. These are not easy decisions in the fastest moving industry on Earth, but we believe they're the right ones to make. Which brings us to SAP. The world's largest enterprises run on SAP. And that's exactly where trusted AI belongs. And that's why Claude models are available to SAP's customers on the SAP Business AI platform. By combining Claude with SAP's depth and scale, we can help hundreds of thousands of organizations use AI throughout every part of their business. Inside Joule, Claude will power agents that take action across finance, HR, procurement, and the supply chain. This puts AI to work inside the systems you've already invested in and the processes your people already rely on. SAP is a great example of a company that is bringing AI into the enterprise in a way their customers can actually trust.
And Tommy is a great example of how we are working together with partners on our new platform. One AI customer of SAP is one of the oldest and largest financial institutions in the world. And in highly regulated industries like banking, accuracy and compliance matter even more. Let's talk about their AI transformation. Welcome on stage the CFO of JPMorgan Chase, Jeremy Barnum.
Jeremy, welcome to Sapphire. Look at all of these people. Is this your first time?
It is my first time indeed. A bit impressive and intimidating.
Every day JPMorgan Chase serves millions of customers and you're actually facilitating trillions of dollars in transactions. So clearly technology is the backbone of your company. How do you manage the tradeoff between fast innovation without compromising on risk and compliance?
Yeah. So let me give you a sense of that scale. We serve over 86 million US customers and 7 million small businesses. And we process about 12 trillion in payments across 120 currencies every day. And technology is the backbone of all of that. It's what allows us to serve our clients seamlessly and securely. But that scale brings complexity which can slow you down and introduce risk. So we're very focused on moving fast without compromising safety. And that means investing in common tools and platforms that are easily deployed and available firmly. So in that context, SAP has been a critical partner for us. And one particularly notable recent milestone in that partnership has been our decision to invest in a major upgrade of our general ledger to SAP's latest version in order to bring our entire financial ecosystem into one unified intelligent platform.
Yeah. So Jeremy, I mean when we talk about scale this is scale. So we are super proud to be your partner on this journey. Now this morning we talked about how to make AI successful in business. Now for you especially as a bank, 80% accuracy is not an option. Now how can SAP and JPMorgan, how do we work together to make AI more accurate?
Yeah, I agree. I have to say at JPMorgan Chase we like our unicorns with only two ears please, thank you very much, it's very important. But no, AI is the most significant trend in technology in a generation and frankly in society at large in a generation and as a result everyone, I think all of us are experiencing pressure to use it everywhere. But in a finance organization the risk of that is that you sprinkle AI on broken processes and you miss the opportunity to retire technical debt and modernize the entire ecosystem. So from our perspective, AI is only as good as the data and processes underneath it. It can't reach its full potential in a fragmented legacy environment. So we're very focused on end-to-end process re-engineering and through Rise with SAP, we're embedding intelligence directly into the workflows. We're already starting to integrate agents to make the day-to-day management of the ledger more efficient. One example is an upstream agentic solution that proactively catches issues with systemic feeds so we can correct them before anything posts to the ledger and over time it learns patterns so we can fix root causes permanently. Importantly, the agents that we've built are not inventing their own business rules. Those rules rather come directly from SAP's embedded control framework and every AI-driven intervention is logged and fully traceable giving us the transparency and audit trail that we and our various stakeholders expect.
Sounds like music to my ear. JPMorgan Chase and SAP, you are not only a customer to us, we are also partners. We are partners in the payment space. We are embedding banking services of JPMorgan Chase into SAP's workflows for our joint customers. So what is your view on our partnership and any ideas for the future?
Yeah, well the partnership between JPMorgan Payments and SAP combines enterprise software and AI expertise with an extensive set of integrated payment solutions at real scale. And what this means in practice is that clients can seamlessly access JPMorgan payment solutions directly within their SAP system. Everything from trade finance and real-time reporting to global payment rails without leaving the workflows they already operate in. We're continuing to develop new solutions together that enhance digital experiences and help our clients grow. And this partnership is only getting deeper. We're now beginning to work with the SAP team on exploring agentic capabilities to create a much more automated and intelligent treasury environment for our joint customers.
Could not have said it better. Jeremy, if you could summarize our partnership in a few words.
Sure, Christian. I can do that. I'm going to use three words. Scale, speed, and trust.
Perfect. Jeremy, first of all, thanks a lot for the trust, for the partnership, the AI co-innovation, and I wish you and JPMorgan Chase only the very best. Thank you very much. Thanks a lot.
The power of the new AI platform enables us to transform our software application layer and as a result we are reimagining how your businesses will run. Our ERP and industry applications, or let's call them systems of execution, have generated incredible value for our customers. I hope you can agree to that. From R/2 to R/3 to ECC to S/4 to our modular cloud suite. I spent the first 15 years of my career working with finance transaction codes like PC02 or FB50 and I was a heavy power user of the BW analytic store. Still I am. 10 years ago I performed right here on this stage a digital boardroom demo. Look at this picture. I'm not only got older, a couple of things have changed, including the CEO.
The digital boardroom, I can remember these times really well, was a revolution at that time. Still, it was always about the end users to tell the system what to do or to analyze the figures provided by the analytical solution. But now this is changing. Today we are very proud to announce that our system of execution will transform into SAP's autonomous suite. The new autonomous suite will span across five domains. Autonomous finance, spend, supply chain, HCM, and CX as well. We are building on our deep industry expertise to launch industry AI. Why do we consider this to be the biggest evolution ever in the history of SAP's application business? Because the autonomous suite is changing the game in five key dimensions. The autonomous suite is role-centric with a tool assistant for every persona in your company always with the human in the loop. It's outcome-based. Our agents listen, understand, steer, and manage the system of execution. What we just have heard from JPMorgan Chase. It's audit-ready. We provide full traceability for every action performed by an agent with the highest compliance standards. And it is extensible by design. You and the ecosystem can extend the autonomous suite layer with new agents and skills every
And our autonomous suite is open by principle. Our free of charge orchestration layer will ensure SAP and non-SAP agents work hand in hand. This is the beginning of something better. Let's hear from one of our key partners how we are making enterprise software more powerful than ever.
It's great to be here at Sapphire. What SAP and NVIDIA are building together is one of the most important platforms in enterprise AI. NVIDIA supply chain is incredibly complex. Millions of parts, hundreds of partners and factories, all connected through SAP. But what's changing is not just how enterprise systems are managed. It's how work actually gets done. We're moving from hand-coded software to AI that can understand, reason, and act. AI no longer simply answers questions. It works for you. And enterprise systems are where work happens. Finance, supply chains, procurement, and every workflow in between. SAP is the foundation of enterprise and now they're building the agents that sit on top of it. Trained on proprietary data with the skills to act. Soon every company will have a workforce of agents. These specialized agents will not replace enterprise software. They will make enterprise software more powerful than ever. And with Nvidia, agents can be built, evaluated, deployed, and governed. We created OpenShell so agents can act inside a company safely and with clear boundaries. And this work is happening as an open-source project. So everything we learn goes back to the community. SAP is embedding OpenShell across its entire platform to secure enterprise AI at scale. Agents can take action, but only what they're allowed to do. They can access data, but only what they're permitted to see. They can execute workflows, but every action is traceable. This is how we make AI safe and useful. It understands the business. It follows the rules. It uses the tools. And it moves fast. Thank you, Christian, Philip, and the SAP team. The autonomous enterprise is no longer a vision. It starts now.
Please welcome to the stage Muhammad Alam.
A big welcome to Sapphire and thank you for being here. So we didn't just build a new business AI platform. We built it and then we used it natively to really completely change the portfolio of products we have into a truly autonomous suite. Underpinning the autonomous suite are out of the box agents. Hundreds of agents cutting across all core business processes. These agents come together into what we call assistants or Joule assistants. We've mapped these assistants to roles across the core processes of an organization because we know that the first step in realizing value from AI is to empower your people to do more, do it better, or do things that just weren't possible to be done before. These assistants can be triggered by humans. They could be system triggered with humans in the loop or ultimately set up to run autonomously collaborating across multiple assistants once you have built the right trust and confidence in the results of their execution. They're also designed with outcomes as a core objective. Each assistant has a defined set of KPIs that you can expect it to deliver. The outcomes are then tracked in the SAP AI agent hub that Philip showed to ensure full transparency and accountability to you and your business. Because ultimately what you want is positive business impact rather than just count of agents deployed as the measure of success for your business. We also know that no two businesses are alike. So not every agent is going to work completely out of the box for you. And for this reason, we've made extensibility a core design principle. You can extend any of these agents by adding tools, workflow steps, and even code to the same simple experience that you saw in Joule Studio earlier. This allows you also to connect them to your non-SAP application because we know you're going to have to do that. You can of course also add completely new custom agents to any of these assistants to complete the autonomous execution for your enterprise. But we didn't just stop there. We know that the traceability and auditability of the agents is critical for you to build your confidence and leverage them at scale. And that is why we're proud to announce that we follow a SOX audit compatible ISO certified development process for building these agents. All our agents are built on this certified compliance framework developed in coordination with our own auditing firm as well as an external audit partner to ensure audit readiness at the agent level particularly for the audit relevant agents. Multiple assistants then come together to form an autonomous domain. In autonomous finance for instance, we have assistants for AP, for AR, for treasury, for controlling and so on. An assistant for every role, each with its number of agents and with its uniquely defined outcomes. We have built these autonomous domains again not just for finance but across the entirety of your core business processes. So let's go through them. In autonomous spend we have assistants from supplier management to sourcing, contracting and buying. For autonomous supply chain we have assistants spanning the entire make to deliver processes. For autonomous HCM, we have assistants like recruiting, payroll, and career development that cut through the entire HCM set of processes. And finally, for autonomous CX, we have assistants covering sales, service, commerce, and marketing processes. All of these domains then come together to help you become an autonomous enterprise, allowing you now as an organization to do things better, more efficiently, and unlock value that you just weren't able to unlock before. Across all of these domains, we're proud to be launching 224 agents and 51 assistants, a number that will only grow monthly from here on out. So, we built all of these systems and we made extensibility a first class principle. But we realized that there was still something missing, something big. Your company context, the organizational memory of your company that sits in the minds of people that have executed those processes for years and years. This knowledge often referred to as decision traces often sits as unstructured data encapsulated in process models, policy documentation, even chats and long emails, making it effectively invisible to any agent on any agentic platform. So we built something to address this tribal knowledge gap, something we call the company memory. Company memory is a knowledge management layer. You can call it a context graph for your company specifically. It's built on the same foundation as SAP Signavio. So it's already natively aware of your process models and your process insights. You can additionally feed it your policies and procedures document relevant content from teams and slack channels and even long email approval chains or exception chains from which it then extracts what we call process atoms which are small structured units of business process knowledge. Each one can guide an agent on what to look for, what to do and more importantly what not to do. When an agent makes a call, it also maintains traces for full transparency and control. When it runs into an exception, it adds to the company memory on how it dealt with that exception. You of course have the power to directly update a rule, review a rule, a process atom manually, and once done, all agents adapt to it instantly. This is what allows us to bring the best of the public LLMs, SAP's unique context layer and your own company context together to really deliver the results that you can rely on. You can clap. I think there's a couple people that like to. Now, I know what some of you may be thinking. You know, you love the assistants and the agents, and you love the company memory story, and you can't wait to deploy in your landscape, but you still might have a long way to go before you complete your modernization journey to our latest cloud applications. However, your business of course expects you to deliver value today. Well, we've thought of that, too. And we're really excited to announce that we are enabling a significant percentage of these Joule assistants and agents to work in hybrid landscapes with the ability to connect to your S/4 on-premises and ECC landscape. I thought I would need to prompt an applause for this one. That turned out to be true. This will be available for our customers that have started their modernization journey on Rise. Of course, we've done this so you can start generating value from AI today on the SAP business AI platform while we're modernizing your estate. Finally, as you would expect, we're also proud to reiterate our commitment to openness and enabling A2A across all of these agents and assistants because we know that some of you may have unfortunately selected other agentic platforms for your organization. So instead of having to rebuild all of these four business process agents from the ground up, you can use A2A to connect these with your agentic platform of choice to get the benefit of the outcome-driven out-of-the-box certified Joule agents and assistants. Okay, so we've talked enough about these assistants. Now let's take a look at one of our more popular finance assistants in action. Please join me to welcome Sophia Levens onto the stage. Sophia,
Thank you, Muhammad. It's incredible to be here with all of you. You've heard about the autonomous enterprise and what I get to do is show you how it feels to work in this new way. Today, we're looking at the financial closing assistant. And in an era where companies can run supply chains in near real time and answer customer questions in seconds, the close can still take a week, sometimes longer. Not because the work is hard, but because the work is fragmented. All right, let's dive in here. What we're going to start with is looking at last quarter and how that went. What you'll see is this company has 12 legal entities, four ERP systems, six different currencies, hundreds of different reconciliations that depend on a human in one country waiting for a human in another. The close is not slow because financial teams are slow. It's slow because the work is person-to-person. What I'm going to show you is a new model, not automation. Automation runs scripts. What you'll see here is that financial closing assistant hard at work coordinating across all the different agents. It's also brought to my attention the 18 things that need a bit of approval. We're going to start here with the accounting approvals agent that has four proposals for me to review. You can see these have the details we need and a confidence level that's been set by my company. I can see this professional services approval is just below the confidence I set. I approve and the system remembers. Now over in intercompany reconciliation which is typically the hardest part of close you can see most of these are completed and working on it. You also see we have some variances each with the root cause already done. You can see timing cutoff and the occasional genuine dispute. One click matches the records on both sides in both currencies. What used to take weeks of emailing back and forth now happens in a single click. A pattern was recognized because the system has that company memory. It learns how your controllers think. Every approval, every override, every comment becomes part of how the agents behave next period. The system doesn't close the books. It learns the company that's closing them. Let's look back to what changed. So, I can easily go back to Q1 and ask the system to compare with what we just saw in Q2. And what we're going to see here is that in the Q2 close, it was faster with less variances with less items for the human to deal with. Each close will teach the next one. The system gets sharper. The exceptions get rarer. The close stops being something you survive and becomes something that can run itself. Okay, Muhammad, back over to you.
