Over the past few years, SAP’s Sapphire event has been primarily targeting the business-side of the SAP audience; with a focus on (strategic) decision makers and executive-level personnel. This focus would generally result in less directly interesting news for more hands-on people such as developers or consultants like myself (especially compared to SAP TechEd). However, with some really profound announcements and strategic moves that will very much impact the way we work hands-on with SAP technology at (and around) this year’s Sapphire, now there truly is a lot to unpack for everyone. Let’s dive in.
Setting the stage: the Autonomous Enterprise
Let me start by saying that it will be virtually impossible for me to do all announcements and developments justice in a single blogpost; it’s simply that much. However, I will try to summarize the key takeaways for the domain of data and analytics here as well as share a few of my initial thoughts.
Having said that, the backdrop to which we need to see everything is that of the Autonomous Enterprise. As we all know, SAP loves a good name change every now and then. However, this time, it’s more than that. When SAP first started moving development capacity from its traditional solutions to AI and increased the pressure on customers to go along with the AI ‘trend’, I was skeptical: SAP wasn’t an AI-native company, nor was it their primary business. However, with the rebranding of the Intelligent Enterprise to the Autonomous Enterprise and SAP’s continued investments in companies such as Conduct, Reltio, n8n and Parloa, I see a real strategic (paradigm) shift in SAP going from a rather traditional ERP software company to a business/AI-driven platform company. The continuous improvements towards Joule (and Joule Work) prove that SAP’s assistant will be embedded nearly everywhere and for all types of users (from end-users to consultants).
Long story short: SAP is rebuilding its architecture as the SAP Business AI Platform and converges its data and semantic governance around the new autonomous suite and Joule Work. In turn, this means that everyone who has a career in SAP, whether you’re a customer user or a partner consultant, really needs to start taking AI seriously. So, now that we have established our context, it’s time to truly turn our attention to the data and analytics developments.
SAP Knowledge Graph as the foundation for the future
Now we have established that the formal direction for SAP’s solutions is towards autonomous agents, there are also clear challenges to be addressed. The first and foremost of these remains the data foundation. To this day, many SAP customers want to move on to a more AI-driven IT landscape with the goal of efficiencies in mind, yet rarely are their contemporary data landscapes ready for this. Not only is there still the matter of semi-and-unstructured data, there is also a considerable volume of tacit, or ‘tribal’, knowledge that is rarely formalized. Think of that one colleague that knows how that one process works in detail, or custom exception-handling routines that reside entirely outside documented schemas. While there is no technical solution that can read people’s minds (yet, luckily!), there are ways to close this gap.
In order to try and do so, SAP introduced ‘Company Memory’, a highly advanced knowledge management layer designed to capture and structure this ambient operational intelligence from sources such as MS Teams, Outlook or Slack. This is where SAP Knowledge Graph comes in. The Knowledge Graph resolves the critical deficiency that prevents most LLMs from understanding existing ERP models and their many custom builds, by encoding fifty years of SAP ERP engineering into machine-readable semantic relationships (see image above). All this SAP ‘legacy’ information now becomes an asset to actually ground multi-agent reasoning and prevent hallucinations we currently see so often when agents are used within an SAP environment. Now combine this approach with the acquisition of Prior Labs, which is known for its Tabular Foundation Models (TFM), and SAP’s new SAP-RPT-1.5 TFM (essentially allowing agents to do way more complex stuff with your data, such as create ‘what-if’ scenarios), and you have a promising combination of ingredients in your data landscape.
Business Data Cloud
With the big architecture change out of the way, let’s have a look at our trusty Business Data Cloud (BDC), which will act as the ‘Knowledge Core’ (along with SAC) in SAP’s new architecture (see image below; which depicts the development of data landscapes over the past decades). In connectivity, we see a new partnership for BDC (through BDC Connect) with Amazon Athena. This will function similar to the existing/planned BDC Connect options by exposing SAP Data Products via Delta Share (with enrich/write back options). There are also various other interseting announcements for BDC, among which are native HANA Cloud in BDC (very welcome!), Reltio for master data management (MDM) with Model Context Protocol (MCP) support for event-driven data sharing also inside BDC (essentially creating an AI supported metadata catalog; how will this work with/against Collibra?), and both AI Core integration and Joule Agents directly available from within BDC.
Also, while not technically announced during Sapphire, remember that SAP acquired Dremio earlier this month as well. With this acquisition, BDC will become an Apache Iceberg-native enterprise lakehouse (in SAP’s own words) that facilitates both SAP and non-SAP data, as well as being a perfect fit for aforementioned Knowledge Graph and Prior Lab’s TFMs. Given that both Iceberg and Delta Lake are designed to solve similar challenges, what will this mean for Delta Share within BDC? Only time will tell but note that all of these features are currently planned to be rolled out (to not yet specified degrees) in the remainder of the year.
SAP Analytics Cloud
On the SAC front, we also see capabilities that shift the paradigm of enterprise planning from periodic, point-in-time forecasting exercises to continuous, agent-driven optimization workflows. Through the newly announced SAP Enterprise Planning initiative, autonomous Joule Agents are embedded directly into SAC. These agents are capable of autonomously detecting external market data signals, simulating the quantitative impact of these signals on KPIs, recommending strategic actions, and orchestrating downstream planning updates while being able to ‘explain’ all this through the actual numbers in your SAC system. This sounds very powerful, but until we see an actual demo I will hold off on final expectations (e.g. does it only work with seamless planning or also with import?).
While also not specifically Sapphire-exclusive, other valuable (Q2) updates to SAC bring the asymmetric reporting layout (especially useful for tables), a new data export API, delta calculations in the Job Monitor and finally a ‘replace model’-feature. This will basically help you out when changing your model underneath a Story; you will be able to (re)view the mapping of a new model to existing measures and dimensions used in a Story and then replace that model while the layout and configuration of your Story are preserved. Although not necessarily fool-proof, I am very happy that SAP is addressing this long-standing limitation.
Conclusion
If we look at the whole picture that was presented at Sapphire this year, we can say without a doubt that the bottom line is that AI, and perhaps better put, agents, are here to stay. SAP is not only investing in this technology, but also re-designing the whole SAP ecosystem architecture around it. With the acquisitions of Dremio and Prior Labs in particular, I personally see a long overdue step towards a truly more open SAP in terms of data. Delta Share brought some openness, albeit with significant limitations, but the adoption of Apache Iceberg as a ‘native lakehouse’ in BDC brings me renewed hope (and excitement!). Combine this with the TFM technology from Prior Labs and it is not hard to imagine the possibilities SAP customers could reap.
I say ‘could’ intentionally, because I first want to experience the results of these announcements in practice. While all these new acquisitions and technologies push SAP’s data landscape towards open standards, it also means that customers will be using these open standards within an ever more vertically integrated SAP ecosystem, from systems of record (S/4HANA) to reporting (SAC). And with more deadlines for classic SAP solutions approaching, the pressure to move to the Cloud and start adopting architecture principles and technologies that promote AI agency, the infamous vendor lock-in is far from history.
Let’s close out on the positive note that this Sapphire, for the first time in many years, actually brought some seriously interesting technological developments (/announcements) for SAP consultants like myself. With AI still developing at a fast (albeit a tad slower than previous years) pace, I can’t wait to actually go hands-on with all the new technologies that are waiting on the horizon.