SAP argues that 100% accuracy in AI is existential for business, requiring strict governance to secure profit margins and stability.
In the world of consumer-grade artificial intelligence, a 90% accuracy rate is often celebrated as a triumph. However, in the high-stakes arena of enterprise operations, that remaining 10% isn’t just a margin of error; it is an existential risk. This is the core message from Manos Raptopoulos, SAP’s Global President of Customer Success, who argues that the transition from probabilistic “guesses” to deterministic control is the only way to secure profit margins in the age of AI.
The Accuracy Gap and Operational Risk
Raptopoulos emphasizes that the evaluation criteria for Large Language Models (LLMs) have shifted. It is no longer enough for a model to be creative; it must be precise. For a global enterprise, an AI agent that miscalculates supply chain requirements or misreads financial postings by even a small fraction can cause immediate, scalable damage to the bottom line.
While a student might use AI to draft an essay where “close enough” is acceptable, a CFO cannot accept a “close enough” tax depreciation calculation. According to SAP’s latest insights, the rise of “Agentic AI”, systems capable of autonomous reasoning and execution, demands a governance framework identical to that of a human workforce. Without it, companies face “agent sprawl,” a modern evolution of the shadow IT crises that plagued the last decade, where unmonitored AI agents make autonomous decisions that conflict with corporate policy.
Data Foundations: The Grounding Moment
The efficacy of AI is tethered to the quality of the data it processes. Raptopoulos identifies a “data foundation moment” where companies must resolve fragmented master data and siloed systems. Generic LLMs trained on internet-scale text are insufficient for corporate needs because they lack context. Instead, SAP advocates for “relational foundation models” grounded in proprietary data, such as invoices, orders, and historical logs.
Supporting this view, reports from MIT Technology Review suggest that organizations with clean, integrated data architectures see a significantly higher ROI on AI deployments. For SAP users, this means moving toward a “clean core.” By keeping the core ERP system free of heavy customizations, businesses ensure that AI agents aren’t “hallucinating” financial results based on outdated or messy spreadsheets, but are instead pulling from a single, unpolluted source of truth.
Designing Trust into the Interface
As software moves from static buttons to intent-based interfaces, where an employee simply tells the system to “prepare a briefing for my top customer,” trust becomes the primary currency. Raptopoulos notes that adoption depends on employees feeling confident that these digital teammates respect corporate boundaries and data privacy laws like GDPR.
Engineering these systems requires specialized “AI personas” tailored to roles like the CFO or CHRO. These personas must be embedded within existing workflows to prevent the friction that often kills new technology adoption. If an AI tool requires a user to leave their primary workspace to “ask a question,” it has already failed the efficiency test. The goal is “invisible AI” that works within the apps people already use.
The Strategic Layers of Success
To navigate this transition, SAP proposes a three-layered strategy for C-suite leaders:
-
Embedded Functionality: This involves immediate productivity gains via AI integrated into core applications, such as auto-generating job descriptions in HR or identifying payment risks in Finance.
-
Agentic Orchestration: This is the next frontier, where multiple AI agents coordinate across systems. For example, a supply chain agent might detect a weather delay and automatically trigger a logistics agent to reroute shipments and a finance agent to update budget forecasts.
-
Industry-Specific Intelligence: This involves deeply specialized applications for unique sectors, such as predictive maintenance in manufacturing or personalized patient journeys in healthcare.
By treating AI governance as a central operating layer rather than a side IT project, businesses can turn AI from a potential liability into a source of durable competitive advantage. The future belongs to those who can govern their agents as strictly as they govern their balance sheets.
About the Author
Jennifer Sakmufuwo Baba is a tech analyst and writer covering artificial intelligence, fintech, and emerging technologies at TechRegard. Based in Nigeria, she's passionate about translating complex tech developments into compelling, accessible stories for diverse audiences. Her work focuses on how technology shapes innovation across Africa and globally.
