The countdown for legacy ERP systems is officially on. If your business runs on SAP ECC 6.0, you likely already know that mainstream maintenance for versions EHP 6-8 ends on December 31, 2027. This deadline is sparking a massive shift toward S/4HANA, with nearly 30% of companies moving to the cloud private edition in the last year alone. But as these migrations accelerate, many IT and operations leaders are realizing a hard truth: an ERP upgrade isn’t a silver bullet for better decision-making. It’s just the foundation.
While SAP and Oracle provide the essential transactional core for finance and procurement, they aren’t built to solve the messy, variable reality of a factory floor or a global supply chain in real-time. An ERP system acts as a giant filing cabinet, it records what happened and what you intend to do. However, it often lacks the mathematical depth to tell you the best way to do it when parts are late, machines break, or labor is short. To bridge this gap, companies are looking beyond the ERP to specialized planning layers that can handle the heavy lifting of mathematical modeling.
In this new era, the goal isn’t just to move your data to the cloud. The goal is to build a system of action where your ERP, your planning tools, and your execution systems talk to each other without friction. This requires a shift away from old-school custom coding toward a more flexible, API-driven approach. By decoupling your decision logic from your core ERP, you can improve your agility and keep your systems ready for whatever the market throws at you next.
The Shift to Clean Core and Side-by-Side Extensions
For decades, the standard way to handle unique business requirements was to customize the ERP core. If you had a specific way of scheduling your production lines, you’d write custom ABAP code inside SAP. This worked for a while, but it created a nightmare during upgrades. Every time you wanted to move to a new version, you had to test and fix thousands of lines of custom code. SAP is now pushing a strategy called Clean Core to end this cycle. The idea is simple: keep the ERP standard and move your unique logic to side-by-side extensions.
By using platforms like the SAP Business Technology Platform (BTP), you can build custom apps that sit next to the ERP rather than inside it. This is where advanced planning and scheduling (APS) software comes into play. Instead of trying to force a standard ERP module to handle complex constraints like sequence-dependent changeovers or specific labor qualifications, you connect a specialized solver. This solver pulls data from the ERP, finds the best schedule, and pushes the results back. This keeps your core ERP clean and your upgrades fast.
The benefits of this approach are measurable. Some companies have reduced their custom code from 9,000 developments down to fewer than 500 by moving to side-by-side extensions. This doesn’t just make IT’s life easier, it allows the operations team to update their planning rules without waiting for a massive ERP upgrade cycle. When you separate the system of record from the system of decision, you gain the freedom to refine your operational logic as fast as your business changes.
Breaking the Integration Bottleneck for AI and Planning
Data integration remains the single biggest hurdle for most businesses. Recent data shows that 80% of organizations struggle with integration when trying to adopt AI. If your planning tool can’t get real-time data on material availability or machine status, it’s basically guessing. Most companies are still stuck with “batch” interfaces where data moves once a day. In 2026, that’s too slow. If a supplier delays a shipment at 10:00 AM, your production schedule should reflect that by 10:05 AM.
Modern integration now relies on an API-led or event-driven approach. Instead of sending a massive file every night, the ERP sends a small notification (an event) whenever something changes, like a new sales order or a finished production run. This allows your planning engine to trigger a re-plan instantly. This is particularly vital for improving operations research practice, where the quality of the output depends entirely on the freshness of the data. If you’re working with stale data, even the best mathematical model will give you the wrong answer.
The financial case for better integration is strong. Research into the SAP Integration Suite shows a 345% return on investment with a payback period of less than six months. This isn’t just about saving developer time, it’s about reducing the cost of errors. When your ERP and planning systems are out of sync, you end up with excess inventory, missed delivery dates, and expensive overtime. By investing in a solid integration layer (often called an iPaaS), you create a nervous system that connects your brain (planning) to your body (execution).
ERP Integration Approaches: 2026 Comparison
| Feature/Criteria | Point-to-Point (Custom) | iPaaS / API-Led | Event-Driven (Pub/Sub) |
|---|---|---|---|
| Speed of Setup | Fast for a single link | Moderate (requires platform) | Slower initial design |
| Maintenance Effort | High (brittle) | Low (governed and visible) | Low (decoupled) |
| Real-Time Capability | Very Limited | High (via synchronous APIs) | Highest (instant updates) |
| Upgrade Impact | High (often breaks) | Low (uses stable contracts) | Lowest (fully independent) |
| Best Use Case | Small, temporary projects | Enterprise-wide connectivity | High-velocity manufacturing |
AI Agents vs. Mathematical Planning: Knowing the Difference
Oracle and SAP are both heavily promoting AI agents and copilots. These tools are great for summarizing data, explaining why a shipment is late, or creating a draft of a procurement contract. However, there’s a common misconception that these AI agents can replace the need for mathematical planning. They can’t. AI agents are typically based on large language models or predictive algorithms that are excellent at spotting patterns, but they don’t understand the hard constraints of a factory floor.
If you ask an AI agent to “make the best schedule,” it might give you something that looks reasonable but is actually impossible to execute because it didn’t account for a specific machine’s cooling time or a worker’s certification level. This is where data science and OR teams create real value. They combine the predictive power of AI (to forecast demand) with the prescriptive power of mathematical solvers (to decide exactly what to do). The winning combination is using a copilot for the user interface and a mathematical engine for the actual decision logic.
Think of it this way: the AI agent is the advisor who tells you “we might have a problem with labor next week.” The planning engine is the engineer who calculates the exact shift roster that covers the workload while minimizing overtime and respecting labor laws. In 2026, we’ll see more companies using these two technologies together. The AI agent handles the “what” and “why,” while the mathematical solver handles the “how” and “when.”
The Role of Process Intelligence in ERP Migrations
One of the biggest risks in moving from an old SAP ECC system to S/4HANA is “paving the cow path.” If you have inefficient processes today, moving them to a faster database won’t help. This is why process mining and process intelligence have become such a big part of the migration conversation. Tools like SAP Signavio or Celonis help you see how work actually happens in your current ERP, often revealing hundreds of variations for a single “order-to-cash” process.
Once you identify these variations, you can decide which ones to keep and which ones to eliminate. More importantly, you can identify the real-world constraints that your ERP isn’t capturing. For example, if the data shows that production always takes two days longer than the ERP says it should, that’s a signal that your master data is wrong. You can’t improve your plan if your underlying assumptions about lead times and capacities are off. Using process intelligence during a migration allows you to clean up your data before you plug in your advanced planning tools.
Finally, consider the long-term risk of vendor lock-in. As seen in recent industry disputes over data access, relying on proprietary extractors can be risky. When building your integration strategy, focus on open APIs and standard data contracts. This ensures that you can always get your data out of the ERP and into the tools that help you make better decisions. Your ERP should be an open source of truth, not a walled garden that prevents you from using the best planning technology available.
Frequently Asked Questions
Should we choose SAP or Oracle for our supply chain?
The choice between SAP and Oracle often depends on your existing footprint and industry. SAP has a deep history in complex manufacturing, while Oracle Fusion is often cited for its user-friendly cloud interface and finance-first approach. However, for supply chain excellence, the integration layer and the specialized planning tools you connect to the ERP are often more important than the ERP brand itself.
What does a “Clean Core” strategy actually mean for operations?
A Clean Core strategy means keeping the standard ERP software as close to the original version as possible. Instead of building custom features inside the ERP, you build them as “side-by-side” extensions using APIs. For operations, this means you can implement advanced scheduling or workforce planning tools without worrying about them breaking every time the ERP is updated.
How do AI agents differ from planning and scheduling tools?
AI agents are great at summarizing data, answering questions, and automating simple tasks using natural language. Planning and scheduling tools use mathematical solvers to find the best possible answer among millions of options while respecting hard constraints like capacity and labor rules. AI agents help you interact with the data, but mathematical engines solve the actual problem.
What data is most critical for integrating ERP with a planning engine?
To run an effective planning model, you typically need five key data types from your ERP: Bill of Materials (BOMs) and routings, current inventory levels, open sales and purchase orders, resource capacities (machines and labor), and master data like lead times and changeover rules. The more real-time this data is, the more accurate your plans will be.
