The enterprise software landscape is experiencing a fundamental shift as vendors move from AI copilots to autonomous AI agents. In a recent conversation with analyst and strategist Charlie Mitchell, we explored how SAP and Oracle are approaching this transformation—and the strategic differences between their offerings could reshape how enterprises think about automation.

Watch the interview here:

From Copilots to Autonomous Agents

The evolution has been rapid. We’ve moved from AI copilots that support human workers to AI agents capable of reasoning, adapting, and executing tasks independently. But here’s where it gets interesting: the past year’s custom-built agents proved too time-consuming and complex for most organizations to develop in-house. Now, the responsibility has shifted back to vendors to deliver pre-configured, role-based agents embedded directly into their software.

Both SAP and Oracle recently announced role-based AI agent capabilities, but their architectural approaches differ significantly—and those differences matter.

SAP’s Integrated Intelligence Approach

At its recent SAP Connect event, SAP unveiled new “agents at work” across its SaaS applications, repositioning Joule as an agent orchestrator that can call other AI agents to execute common tasks. SAP’s strategy emphasizes deeply embedded AI within integrated business processes, powered by contextual data from its Business Data Cloud.

The architecture operates as a flywheel: applications generate context-rich data, that data fuels embedded AI, and the AI enhances applications. The SAP Knowledge Graph serves as the connector within this system, providing agents with deep understanding of complex end-to-end business processes.

SAP offers a two-tiered pricing model. Joule Base is free with all SAP cloud subscriptions, providing foundational AI capabilities. Joule Premium extends these capabilities using tiered pricing based on AI units measured per user per month.

Oracle’s Unified Cloud-Native Strategy

Oracle has released over 50 role-based AI agents embedded within its Fusion Cloud applications, spanning supply chain, marketing, sales, service, finance, and HCM. Perhaps more intriguingly, Oracle launched an AI Agent Marketplace where customers can deploy validated partner-built agents directly within their enterprise environment.

Oracle’s architecture keeps enterprise data within the native Fusion environment, with AI built directly on top of integrated applications. The Fusion suite combines CRM, ERP, human capital management, and supply chain within one composable architecture powered by Oracle Cloud Infrastructure (OCI). This enables agents to follow optimized cross-departmental workflows with functional awareness baked in.

On pricing, Oracle takes a simpler approach: there’s no additional cost for access to generative AI, predictive AI, and AI agents as part of the core Fusion offering. The AI Agent Studio is also available at no extra charge.

The Architecture Question

As Mitchell pointed out, Oracle’s unified architecture may provide advantages for the next generation of AI workflows—coordinated multi-agent systems that accomplish cross-application tasks. “If we think about how humans work, they don’t just work within one application,” he noted. The cross-domain data integration limits operational work that often slows agent projects.

SAP’s architecture is more complex, but its embedded intelligence approach marks a potential shift away from dashboard-based enterprise intelligence toward embedded decision loops that help agents self-reflect and adjust behavior.

Beyond the Marketing Numbers

Both vendors tout impressive agent counts, but there’s marketing spin involved. Oracle claims 600 AI agents and assistants—though not all qualify as true autonomous agents. Some represent AI-powered features or use cases that have existed for years, now rebranded for the agent era.

The real question: what constitutes an agent versus a feature? True AI agents can perceive environments, make decisions, and execute multi-step workflows autonomously. A use case that provides an AI-powered insight isn’t equivalent to an autonomous agent executing complex processes.

The Competitive Landscape

Neither SAP nor Oracle operates in isolation. ServiceNow’s role-aware agents and control tower approach resonates with IT departments already familiar with the platform. Salesforce is positioning Slack as the collaboration hub for human-AI interaction. Microsoft’s Azure AI strategy leverages its platform dominance.

The vendor that wins won’t just have the most agents—they’ll have the right architecture for cross-system coordination, the clearest value metrics, and the lowest barriers to adoption.

As enterprises evaluate these platforms, the choice won’t be binary. It depends on existing infrastructure, strategic priorities, regulatory requirements, and cloud maturity. But one thing is certain: role-based AI agents are here, and they’re fundamentally changing how we think about enterprise software.

 

This article was originally published on LinkedIn.

 

Read more of my coverage:

AWS Doubles Down on European Digital Sovereignty with New Governance Structure

Veeam Data Platform v13: What MSPs and MSSPs Need to Know

Gen AI Moves From Experimentation to Everyday Work—But the Human Factor Remains the Challenge