A couple of weeks ago, SAP held its Sapphire 2026 conference in Orlando, Florida. As SAP’s premier annual event, it once again delivered a mix of big announcements, ambitious vision-setting, and plenty of industry buzz.
Of course, Sapphire season is never just about the event itself. The other part of the story is watching customers, partners, analysts, and the broader enterprise technology community try to interpret what it all means in real time.
Within minutes of every keynote, LinkedIn fills up with hot takes, prediction threads, and confident declarations about the future of enterprise software. Customers start wondering what matters, what is marketing, and what actually requires action today.
Sapphire 2026 felt especially active in that regard because many of this year’s themes touched on areas already reshaping the market: agentic AI, business data, workflow automation, and the growing connection between applications, analytics, and AI-driven experiences.
To SAP’s credit, the company presented a fairly cohesive vision for how these pieces fit together. At the same time, it's important to separate the wheat from the chaff and make sense of what all this means for customers living in the real world.
One of the things we like to do around here after a major industry event is publish a "knee-jerk reactions" post with a few early takes while the conversations are still fresh on our minds. Some observations may age well. Others may not. You're welcome to keep receipts.
If We Only Had a Brain...
SAP has long positioned their ERP at the center of every company's universe. It's a myopic worldview as old as the company itself, immortalized in campaigns like "The Best-Run Businesses Run SAP", etc. SAP at the center (or core), SAP as the intelligence, SAP as the reason everything works.
And in some respects, they're not wrong. Depending on who you talk to, anywhere between 60 and 77% of the world's transaction revenue touches an SAP system. That's a staggering number and just goes to show how integral SAP systems are to day-to-day operations for companies around the world.
In the opening keynote, Christian Klein, CEO of SAP, took things a step further, positioning ERP as the brain of every company. This is a somewhat bold statement, but one we're hearing from most ERP vendors these days as they look to reinvent themselves in the world of AI.
Figure 1: SAP's Position on ERP Being the Brain of Every Company
While the "ERP is the brain" analogy has merit, it doesn't necessarily hold up under scrutiny. ERP systems only take on some of the brain's primary functions. And when you look closely at which ones, a pattern emerges: ERP systems are really doing the work of long-term memory storage. For all you amateur neuroscientists out there, we're talking about the brain's long-term memory regions:
Hippocampus: This is the part of the brain responsible for forming both semantic and episodic memories. In ERP parlance, we're talking about the recording of facts (i.e., documents and transactions), events, and transaction flow.
Neocortex: This part of the brain is where long-term memories live. Think of this as being kind of like the ERP system database.
Cerebellum & Basal Ganglia: This part of the brain is responsible for procedural memory. For example, how to ride a bike, type, or throw a ball. In ERP terms, we're essentially talking about process knowledge. ERP systems natively know how to process transactions and, with the right UX, can make these rote processes relatively easy to perform.
These are all critical functions. But notice what's missing: the parts of the brain responsible for thinking and reasoning. Throughout the remainder of the keynote presentation, Mr. Klein and his colleagues expand on this concept by illustrating how SAP has mapped other brain functions to new dimension products/features such as the SAP Knowledge Graph, SAP BDC, and so forth.
🤔 Let's continue to unpack this, shall we?
Meet SAP's New AI Platform...Same as the Old One?
One of the bigger announcements at Sapphire 2026 was the introduction of the SAP Business AI Platform. The branding suggests an entirely new foundation purpose-built for the next era of enterprise AI. However, as is the case with many of SAP's renaming/repackaging adventures, the reality is a bit more nuanced.
At its core, the SAP Business AI Platform is a strategic packaging exercise. SAP has taken three existing portfolio pieces — the SAP Business Technology Platform (SAP BTP), SAP Business Data Cloud (SAP BDC), and AI Foundation — and consolidated them under a single, AI-centric umbrella. Most of what was announced already existed before Sapphire. What changed is how SAP is now presenting it.
Figure 2: Introducing the New SAP Business AI Platform
Branding aside, this shift does reflect a broader industry pattern. For example, Microsoft has done something similar by weaving together an AI tapestry based on Foundry, Fabric, and Copilot Studio. Salesforce is running the same play with Agentforce. It seems like every major enterprise software vendor is trying to establish a coherent story for how data, applications, automation, and AI services fit together. SAP might be playing a little bit of catch-up here, but the story is the same.
With that being said, we have had several conversations with customers confused about the positioning of all this. Part of this confusion stems from the fact that both SAP BDC and AI Foundation run on top of SAP BTP. We also find that some customers struggle to understand where SAP BTP ends and where host hyperscaler platforms (e.g., Azure, AWS, and GCP) begin. Add in strategic partnerships with AI model providers like Anthropic and OpenAI, and the lines become extremely blurry.
These details matter for customers trying to understand what they're buying, where workloads actually run, and which components introduce additional licensing or operational complexity. We'll attempt to peel back the layers on all this in just a moment.
The Fine Print on Openness
Throughout the keynote, SAP emphasized the fact that the SAP Business AI Platform is "open by design". While this is technically true in most respects, there's another way of looking at this: a significant portion of the platform capabilities are not SAP-developed IP.
Underneath all the branding, customers are mostly consuming services from Microsoft, Databricks, Anthropic, OpenAI, and others. SAP is acting as an orchestration and packaging layer, adding governance, business context, workflow integration, and enterprise controls on top of those foundations.
There's genuine value in that. Most enterprise customers do not want to stitch together twenty different AI and data services from scratch. And, they want to have access to a wide array of AI models — a sort of "bring your own model" (or BYOM) approach.
But this flexibility comes with some notable costs. Not only are you paying a premium for repackaged services, you're also losing direct access to the native flexibility, release cadence, and developer tooling of the underlying providers. Proprietary abstraction layers can make advanced functionality much harder to reach than working directly with the source platforms.
None of that makes SAP's approach wrong. It does reinforce one thing: understanding where the actual platform boundaries exist is way more important than where the marketing diagrams suggest they do.
SAP Business AI Platform Layers
Strip away the branding and the SAP Business AI Platform starts to reveal itself for what it actually is: a layered architecture designed to organize how AI solutions are built, grounded in business context, and governed across the enterprise.
At a conceptual level, the structure makes sense. SAP divides the platform into three primary concerns: building AI-powered experiences, contextualizing those experiences with business data and reasoning, and governing the resulting agents and automation landscape.
Where it gets interesting is when you examine which technologies SAP considers strategically important within each layer.
Build Layer
The Build layer focuses on the tools and services used to create AI-powered applications, workflows, and agents. Much of this centers around technologies customers already associate with SAP BTP, including Joule Studio 2.0, SAP Build, and the SAP Integration Suite.
Figure 3: Introducing SAP Joule Studio 2.0
This is the part of the platform aimed at enabling developers and business technologists to assemble AI-enabled business experiences without building everything from scratch. In many ways, the strategy mirrors what we're seeing elsewhere in the market with platforms like Microsoft Copilot Studio and Salesforce Agentforce, where the emphasis is shifting toward orchestration, workflow composition, and rapid assembly of AI capabilities into operational business processes.
This is also an area where I think something more significant is happening beneath the surface. Earlier we talked about ERP as the brain of the enterprise and the gaps that have always existed in how ERP systems make decisions and reason through complexity. This Build layer is where we start to see that intelligence shift. Not out of the ERP core entirely, but outward into a tapestry of intelligent agents and workflow processes that can sense, reason, and act in ways traditional ERP systems never could. The core still holds the data and the process logic. The agents are increasingly where the thinking happens.
SAP Integration Suite plays an especially important role here because agents are only useful if they can interact with the broader enterprise landscape. APIs, events, business transactions, document flows, approvals, external systems — all of it becomes part of the equation. After all, most enterprise AI initiatives eventually reveal themselves to be integration projects in disguise.
Contextualize & Reason Layer
This is arguably the most important layer in SAP's (re)imagined AI platform architecture. It's here where SAP attempts to solve one of the hardest problems in enterprise AI: grounding AI systems in accurate business context. This is where the SAP Knowledge Graph and SAP BDC become strategically significant.
The SAP Knowledge Graph is a foundational piece of SAP's long-term AI strategy. It provides semantic understanding of how business objects, processes, relationships, and transactions connect across the SAP ecosystem. In practical terms, this gives AI systems far richer context than simply dumping ERP tables into a vector database and tasking a large language model (LLM) to figure things out.
Figure 4: Working with the SAP Knowledge Graph
This distinction matters more than it might first appear.
Enterprise data is highly relational, highly contextual, and deeply dependent on business process semantics. Understanding the relationship between a sales order, delivery, invoice, vendor, customer hierarchy, material, production order, or cost center requires a lot more than generic language reasoning. It requires structured business understanding.
SAP isn't alone in recognizing this. Across the enterprise software landscape, there seems to be this stream of consciousness moment where vendors are independently arriving at the same conclusion: large language models alone are not enough to power reliable enterprise AI. What's missing is a structured intelligence layer sitting between raw data and AI reasoning models:
SAP is building that layer through the Knowledge Graph and tabular AI models.
Microsoft is pursuing a similar idea through graph-powered intelligence capabilities and ontologies that weave together organizational data, relationships, and semantic context.
Salesforce has its Data Cloud and metadata-driven semantic layer.
ServiceNow has its own graph-based process intelligence approach.
Figure 5: Building an IQ Layer Across the Enterprise
What makes SAP's vision worth watching is the depth of process semantics baked into the Knowledge Graph combined with its investments in tabular models. While much of the AI market remains fixated on large language models, SAP is investing heavily in tabular data models that are able to reason their way through complex ERP system databases. SAP's own SAP-RPT-1.5 model reflects this focus, as does the recent acquisition of Prior Labs and its Tabular Foundation Models (TFMs).
If SAP can pull this off — combining knowledge graph semantics, structured business context, and tabular reasoning into a coherent whole — they could end up with something genuinely differentiated from the generic AI platforms flooding the market right now. After all, no other vendor is sitting on decades of encoded business process logic at this scale. That's either a significant moat or a significant complexity burden. Probably both to be honest. How well SAP can surface all of that through modern AI interfaces will be the real test.
Govern Layer
This layer is all about security and governance. Within the keynote, the main attraction was the new(ish) SAP AI Agent Hub. SAP AI Agent Hub is designed to provide a centralized 360° control plane for managing agents, particularly for organizations deploying agents across large numbers of business processes.
Once again, SAP's entering an increasingly crowded market as pretty much every major enterprise software player is vying for position as the go-to agent management platform. Some notable competitors include:
Microsoft's Agent 365 platform
Salesforce's Agentforce
ServiceNow's AI Control Tower
The concept makes sense. Most enterprises are not going to manage hundreds of AI agents through disconnected point solutions. Eventually, they're going to need centralized governance models similar to what emerged around APIs, integration platforms, identity systems, and cloud infrastructure over the last decade. That evolution took time. This one probably will too.
While it's too early to predict who will win this race, color us skeptical on SAP's ability to execute on this vision. SAP has a pretty lengthy history of building platforms like this that play very nicely with SAP solutions but struggle to keep pace with technology innovations outside of the SAP biosphere.
Introducing the SAP Autonomous Suite
The other major announcement at Sapphire 2026 was the introduction of the SAP Autonomous Suite. At its core, the Autonomous Suite represents SAP's vision for an enterprise application landscape where software becomes increasingly proactive, conversational, and capable of executing work with varying degrees of autonomy. Instead of users manually navigating transactional systems to complete every process step themselves, AI agents, copilots, automation workflows, and reasoning engines begin handling larger portions of operational work on their behalf.
Figure 6: Introducing the New SAP Autonomous Suite
Shifting Responsibilities
To understand why this matters, it helps to think about what ERP systems have traditionally been.
Historically, ERP systems have mostly been positioned as a very expensive system of record. They stored transactions, enforced process discipline, and acted as the authoritative source of enterprise data.
While these systems undoubtedly added value, the reality is that most of the hard work is still mostly carried out by users. Interpreting information, coordinating activities, chasing approvals, executing processes across multiple applications. These have historically been human tasks. Indeed, accountants, planners, buyers, coordinators, and operations managers built entire careers around knowing how to make these systems do what the business needed.
SAP's vision with the Autonomous Suite is to shift some of that responsibility onto a new AI-powered abstraction layer. In this model, applications are expected to surface insights proactively, coordinate workflows automatically, and participate directly in business operations alongside human users. SAP Joule, AI agents, the Knowledge Graph, BDC, and the SAP Business AI Platform all function as key enablers for this broader strategy.
Going Headless
There's a deeper implication here that's worth sitting with. If agents and automation are handling increasing portions of operational work, the role of the traditional application interface starts to change fundamentally. Users may no longer need to log into a procurement system to raise a purchase order, navigate a finance module to approve an invoice, or open a supply chain application to investigate a delivery exception.
Instead, an agent handles it and, behind the scenes, a workflow executes. In this scenario, human beings are only interrupted whenever something requires judgment or intervention: the so-called "human in the loop" model.
This is the emerging idea behind headless ERP (or appless as SAP likes to call it) experiences. In this model, line of business applications take more of a behind-the-scenes role, focusing on maintaining the data model, process logic, and transaction logs. Meanwhile, the user-facing interaction layer increasingly shifts towards a more conversational mode where notifications and autonomous actions drive the bus as opposed to transaction screens full of fields and menus.
Figure 7: SAP Goes Appless with SAP Joule Work
SAP Joule Work, SAP's redesigned user experience introduced at Sapphire, points directly at this direction. Here, once again, SAP is coming in a little late to the game and facing stiff competition:
Microsoft has been hard at work integrating Copilot across the collaboration tools many business users live in every day: Teams, Outlook, and Microsoft 365.
Salesforce has a similar play with Slack and Agentforce.
ServiceNow is repositioning its entire platform around autonomous workflows rather than ticket-driven interfaces.
At the end of the day, the real winners here are the end users. The traditional application screen — the form, the dashboard, the transaction code — may not disappear entirely, but its role as the primary interaction model for enterprise software is increasingly being questioned. So, if you're sick and tired of dropping what you're doing and having to fire up the SAP GUI, this is undoubtedly a welcome shift in the right direction.
The Suite Life
From a portfolio perspective, this shift is not an entirely new concept for SAP. Twenty years ago, SAP unified many of their core enterprise functions — ERP, CRM, SCM, SRM, and PLM — into a tightly integrated operational platform known as the SAP Business Suite. The emphasis was on process integration and transactional consistency.
One could argue that the Autonomous Suite is the next evolution of that idea. The difference is what holds it together. The connective tissue is no longer just shared transactional workflows; we're now talking about shared business context, AI orchestration, semantic understanding, and agent-driven execution. SAP is not just integrating applications. It's attempting to build an environment where AI systems can reason across those applications and act on what they find.
Whether that vision is achievable depends heavily on foundational realities many SAP customers are still wrestling with: fragmented data landscapes, inconsistent process governance, incomplete cloud migrations, deep customization complexity, and integration sprawl. Autonomous systems are only as effective as the foundations beneath them. That's not a small caveat.
But the strategic direction is clear. SAP believes the future of enterprise software is not just cloud ERP. It's AI-coordinated business operations built on top of cloud ERP (and SAP's new Business AI Platform right...right?). The Autonomous Suite is SAP's attempt to define what that operating model looks like.
Real or Imagined?
At this point, it's worth stepping back from the architecture diagrams and keynote demos to ask a simple question: how much of this is actually real for customers today?
The honest answer is that it depends on what you're looking at.
The underlying technologies SAP discussed at Sapphire are real: SAP BTP, Joule, Business Data Cloud, AI Foundation, Integration Suite, and the Knowledge Graph. Customers are already building automation solutions, copilots, data platforms, and AI-assisted workflows on top of these platforms today. That part isn't vaporware. One could (and we have) made the argument that there are way too many proprietary entanglements in most of these products, but that's a conversation for another day.
What feels less real is the fully unified and highly autonomous operating model SAP presented on stage.
The Sapphire demos were carefully orchestrated scenarios built on clean integrations, governed data, modernized application landscapes, and well-defined business processes. The reality inside most enterprises looks considerably messier. For example, many SAP customers still haven't even started their S/4 migration project. Others are managing fragmented data estates spread across SAP, Microsoft, Salesforce, Oracle, Workday, and decades of accumulated custom integrations. And that's not to even mention the number of core business processes that still depend on spreadsheets, email chains, and tribal knowledge that never made it into any system at all.
The problem is that autonomous systems require mature foundations to stand on. AI agents cannot reliably orchestrate broken business processes. Knowledge graphs do not magically fix poor data quality. Autonomous procurement workflows still depend on governance models, approval structures, and integration reliability. Even the best reasoning models struggle when the underlying enterprise landscape lacks consistency and operational discipline.
None of that makes SAP's vision wrong. The direction makes sense. The broader market is clearly moving toward AI-assisted operations, agent orchestration, and increasingly autonomous workflows. But what SAP presented at Sapphire is less a description of where most customers are today and more a signal of where enterprise architecture is heading over the next several years.
That distinction matters, but the vision does not need to be fully mature today to be strategically important. In our estimation, the immediate challenge for customers is determining whether or not to bet the farm and trust SAP to deliver on their AI vision, taking both the good and the bad. There's no right or wrong answer here, but we would certainly encourage you to do your homework and shop around because, as we've observed, SAP's vision isn't unique within the industry, and certainly not new.
In the meantime, the real near-term opportunity is far more incremental: modernizing integrations, improving data foundations, introducing targeted AI-assisted workflows, and gradually reducing friction in how work gets done. No matter where you are in your SAP journey, these are improvement that you can start on today. And these less glamorous foundational investments may ultimately determine which organizations are actually ready when the more ambitious parts of this vision arrive.
Closing Thoughts
Stepping back from the individual announcements, SAP Sapphire 2026 felt less like a traditional product conference and more like SAP attempting to redefine how it wants customers to think about enterprise software altogether. The company is clearly moving beyond ERP as a transactional system and toward a future where applications, data, automation, and AI agents operate together as a broader intelligence layer.
Whether SAP can fully execute on that vision is an open question. Much of what was presented still depends on customers modernizing their architectures, improving data quality, and establishing stronger governance foundations. In many cases, the limiting factor will not be the AI. It will be the operational maturity of the environments those AI systems have to work with.
Still, it would be a mistake to dismiss these announcements as mere marketing hype. Even where parts of the vision remain aspirational, the broader direction feels increasingly settled. Enterprise software platforms are evolving from systems that record work into systems that actively participate in getting it done. SAP appears determined to position itself at the center of that shift. How much of this moves from keynote stage to operational reality over the next few years will be worth watching closely.
