AI Agents in ERP: Why SAP, Microsoft and Oracle Are Betting on Autonomous Processes

7 min read

SAP has merged Joule with Microsoft Copilot, Microsoft is transforming Copilot Studio into an autonomous agent, and Oracle has launched 22 agent-based applications in a single release. Within three months, every major ERP provider has integrated AI agents into their systems. Gartner predicts that 40 percent of all enterprise apps will include task-specific AI agents by the end of 2026. For mid-sized businesses, the question is no longer if, but when their ERP landscape will run on agent-based intelligence.

What are AI agents in an ERP system?

Traditional ERP systems are passive databases. They store orders, inventory and financial data, but they don’t act. When an order arrives, a human must check stock, confirm availability and trigger shipping—every step needs a person.

AI agents flip this model. They detect patterns in transaction data, make decisions within defined rules and carry out actions independently. In supply chains, for example, an AI agent spots an impending material shortage, evaluates alternative suppliers, compares prices and generates a purchase recommendation—in seconds, not minutes.

The difference from copilots is decisive: a copilot answers questions and suggests. An agent executes. SAP, Microsoft and Oracle have completed this paradigm shift in just months—and ushered in a new era of process automation.

SAP Joule: From assistant to autonomous process manager

On January 14, 2026, SAP announced its most far-reaching integration to date: Joule, the company’s in-house AI assistant, is merging with Microsoft Copilot. The practical upshot? A project manager in a Teams meeting can ask Copilot to query Joule for the current stock level of a specific component via natural-language chat—and receive the answer directly in the conversation.

Yet SAP’s strategy extends beyond simple chat integration. New Joule agents for supply-chain management autonomously verify production-order prerequisites, flag material shortages, and propose solutions. The game-changer: a Joule agent can instruct a Microsoft Copilot agent, and vice versa. SAP and Microsoft are thus building a multi-agent ecosystem in which specialized agents collaborate.

For mid-market S/4HANA customers, the upshot is clear: the AI arrives not as a separate product but as an integral part of the existing system. Those already on S/4HANA Cloud receive Joule agents as part of their licensing model. Users still on the legacy ECC platform now have an extra incentive to accelerate migration.

Microsoft Copilot Studio: Dynamics 365 becomes an AI-powered workplace

In April 2026, Microsoft took the next step: Copilot Studio is evolving from a chatbot builder into a platform for reasoning-based AI agents. These agents operate autonomously, interact with applications, execute complex business processes, and make decisions with minimal human oversight.

Microsoft’s goal is unambiguous: Copilot Studio should no longer merely respond—it should act. The updates transform the basic chatbot into a full-fledged, autonomous AI employee for business processes.
Based on Microsoft announcement, April 2026

For Dynamics 365 users, that means AI agents can autonomously qualify customer inquiries in CRM, prepare quotes, and schedule follow-up meetings. In the finance module, agents automatically reconcile invoices and escalate only exceptions to human clerks. According to a recent analysis, 15 million people already use Microsoft Copilot, though only 3.3 percent pay for the premium version. The new agent capabilities are likely to shift that willingness to pay.

Microsoft’s strategic edge: the agents operate seamlessly across Microsoft 365, Dynamics 365, and Azure. A mid-market company already invested in Microsoft’s stack gains AI agents without additional infrastructure.

Oracle Fusion: 22 agent apps unleashed at once

On 24 March 2026 Oracle launched its most aggressive move yet: Fusion Agentic Applications are a brand-new class of enterprise software that does not rely on individual copilots but on coordinated teams of specialist AI agents. Twenty-two agent apps span finance, HR, supply chain and customer service.

Oracle’s approach is fundamentally different: instead of bolting AI onto existing software, Oracle has embedded the agents natively inside the transaction system. The agents tap directly into corporate data, workflows, approval hierarchies and permissions. With the AI Agent Studio, companies can also build their own agents without resorting to traditional application development.

Real-world use cases from the first release include automated headcount planning and salary optimisation, AI-driven supplier discovery to cut costs, autonomous receivables collection to improve working capital, and cross-sell agents that surface revenue potential in existing customer accounts.

What stands out is Oracle’s speed: 22 production-ready agent apps in a single release is a market signal. While SAP and Microsoft are rolling out agents one by one, Oracle is going all-in. Existing Oracle Fusion Cloud customers gain immediate access. For Germany’s mid-market, however, the advantage is less pronounced: Oracle is far less common in the DACH mid-market than SAP or Microsoft.

What does this mean for SMEs?

Three trends are converging on SMEs right now: the skills shortage is worsening, business processes are growing more complex, and ERP vendors are finally delivering AI features that go beyond gimmicks. AI agents in ERP systems are no longer a pipe dream—they’re here.

The tangible benefits fall into three areas. First, routine relief. Agents take on tasks like invoice matching, inventory checks and appointment scheduling that currently tie up skilled staff. Second, faster decisions. Instead of waiting for the monthly report, agents deliver decision-ready insights in real time. Third, fewer errors. Agents work rule-based and consistently—they never skip an approval step or overlook a price variance.

At the same time, legitimate questions remain. How much autonomy should an agent have? Who’s liable if an agent makes the wrong call? And what happens to employees whose routine tasks disappear? Business leaders need to answer these now—not once the agents go live. The EU AI Act already provides the regulatory framework.

A realistic scenario for a 300-employee mid-size firm: the invoice-in agent automatically matches incoming invoices against purchase orders, flags price variances and posts entries when everything aligns. The inventory agent monitors critical materials and reorders before stock hits the minimum. The HR agent flags open leave requests, checks coverage rules and approves under policy. Three agents that together free up roughly two to three full-time equivalents of routine work—not through layoffs, but by redeploying capacity to higher-value tasks.

How to prepare your ERP for AI agents

1. Verify data quality. AI agents are only as good as the data they work with. Clean master data, remove duplicates, and update process documentation. Without clean data, agents deliver inaccurate results.

2. Ensure cloud readiness. All three vendors deliver AI agents primarily via their cloud platforms. SAP Joule requires S/4HANA Cloud, Microsoft’s agents run on Dynamics 365, and Oracle relies on Fusion Cloud. On-premises systems are excluded.

3. Define a pilot process. Don’t automate everything at once. Choose a clearly defined process—such as invoice receipt matching or inventory monitoring. Measure results, then scale.

4. Establish governance rules. Which decisions can an agent make independently? At what amount must a human confirm? Define these rules before rollout, not after.

5. Involve your team. AI agents don’t replace employees; they change their tasks. Clerks become process managers who steer agents and handle exceptions. Communicate this shift early and plan training sessions.

The technology is ready. The question is whether your internal conditions are in place. Companies launching a pilot project now gain a 12- to 18-month head start over those waiting for the next licensing round.

Frequently Asked Questions

Source of cover image: Pexels / Jakub Zerdzicki (px:36598855)