Artificial intelligence in ERP is moving beyond predictions, dashboards, and conversational search. AI agents can now help users carry out actual business processes by gathering information, evaluating conditions, recommending next steps, and executing actions with human approval.
A practical example can be found in manufacturing.
As of September 2026, SAP’s Production Planning and Operations Agent in SAP Cloud ERP helps production supervisors evaluate production orders without manually moving between multiple applications to check materials, capacity, scheduling, and other operational conditions.
Instead of making AI another tool employees have to manage, the goal is to embed intelligence directly into the work they already perform with the SAP Business AI Platform.
An AI agent in ERP is software that can evaluate business information, perform multi-step tasks, and recommend or execute actions on behalf of a user while operating within defined business processes and controls.
Traditional ERP automation typically follows predetermined rules: when one event happens, the system performs a specified action.
An AI agent can go further. It can gather information from different parts of the ERP system, evaluate that information in the context of a user's goal, identify an exception, and determine what actions should come next.
SAP’s Production Planning and Operations Agent provides a practical example of what this looks like inside SAP Cloud ERP.
Rather than treating AI as a separate application, the agent works within a production supervisor's existing workflow.
AI can help production planning teams by automatically evaluating materials, production capacity, scheduling dates, and production orders before those orders are released to the shop floor.
Consider the scenario demonstrated in our video.
A production supervisor arrives at 7:00 a.m. and discovers that 41 production orders need to reach the manufacturing floor that day.
Before releasing those orders, the supervisor needs answers to several important questions:
The answers already exist within the ERP system. The challenge is gathering and evaluating them efficiently.
Without the agent, the supervisor may have to work across several different areas of SAP Cloud ERP to get a complete picture.
That can include reviewing production orders, checking work center capacity, evaluating material inventory, and using material requirements planning (MRP) information to understand when missing components may become available.
The Production Planning and Operations Agent brings those activities together.
The Production Planning and Operations Agent evaluates the information needed to determine whether selected production orders are ready to be released.
As demonstrated in the video, the agent can help evaluate:
The supervisor begins in the Manage Production Orders application and selects the orders that require action.
Instead of opening and evaluating each order individually, the supervisor can ask SAP Joule to initiate the process.
The agent evaluates the parts, materials, and components required for the production order.
It can determine whether inventory is available to support production and identify situations in which a required component is missing.
Having inventory is only part of the production equation.
The agent also evaluates work center utilization, including available machine and labor capacity, to determine whether the organization has the resources necessary to execute the selected orders.
The agent evaluates the dates associated with the orders to determine whether production requirements can be fulfilled according to the prescribed schedule.
Together, these checks give the production supervisor the information needed to make a more informed release decision without manually assembling the data from several places.
One of the most important aspects of the demonstration occurs when the agent identifies a missing component.
Rather than simply stopping the process or making the decision autonomously, Joule presents the production supervisor with potential next steps.
Depending on the situation, the supervisor may be able to:
This is where the relationship between AI automation and human expertise becomes particularly important.
In the demonstration, the supervisor already knows from the procurement team that the missing component will arrive the following day. Based on that business knowledge, the supervisor decides to proceed with the order.
The agent performs the research and processing work. The human still makes the business decision.
No. In this example, the AI agent supports the production supervisor rather than replacing the supervisor's judgment.
This distinction is critical. The Production Planning and Operations Agent can perform many of the repetitive activities required to evaluate a production order. But when an exception occurs, the production supervisor remains responsible for deciding what should happen.
The workflow demonstrated in SAP Cloud ERP follows a human-in-the-loop model:
AI gathers information → AI evaluates conditions → AI identifies exceptions → Human reviews options → Human makes the decision → AI executes the approved action.
This approach allows manufacturers to automate repetitive work without removing business context, expertise, or oversight from important operational decisions.
In the demonstration, Joule provides the interface through which the production supervisor interacts with the agent.
After the supervisor selects the relevant production orders, Joule can initiate the agent and guide the user through the process.
The supervisor does not have to manually navigate each underlying application to collect the necessary information.
Instead, the agent performs that work and brings the relevant information and decisions back to the user.
When an exception such as a missing component is discovered, Joule presents the available actions. After the supervisor makes a choice, the system can continue the workflow and ultimately release the production orders.
This is an important shift in how employees interact with ERP. Rather than asking users to find every piece of information themselves, AI can increasingly bring the information, context, and next steps to the user.
The immediate benefit is time savings, but the potential value goes beyond reducing clicks.
Production supervisors can spend significant time moving between applications and evaluating individual orders.
An agent can perform much of that background analysis automatically.
Instead of spending the beginning of the day compiling information, supervisors can focus on the exceptions and decisions that require their expertise.
The agent considers multiple operational factors, including materials, capacity, and dates, before presenting the user with next steps.
That gives the supervisor a more complete picture of each order.
Routine orders do not necessarily require the same level of attention as orders with missing materials, capacity constraints, or scheduling problems.
AI can help identify those exceptions so employees can focus their attention where it creates the most value.
Perhaps most importantly, the AI capability exists within the business process itself.
Production supervisors do not have to leave their ERP environment, export data into another AI tool, or build their own prompts from operational information.
The intelligence is embedded into SAP Cloud ERP and the work employees are already performing.
Traditional ERP automation performs predefined tasks based on established rules, while an AI agent can evaluate context across multiple steps and help determine what action should happen next.
Automation is not new to ERP. Organizations have been automating approvals, transactions, calculations, and workflows for decades.
AI agents build on that foundation.
In the production example, the system is not simply being told, “Release this order.” It first has to evaluate the conditions surrounding that order.
Are the necessary components available?
Is adequate capacity available?
Do the dates work?
Is there an exception that requires human input?
The agent is helping orchestrate a broader process rather than automating a single isolated transaction. That difference is one reason AI agents have the potential to change how employees interact with ERP systems.
For manufacturers, the value of agentic AI may come from hundreds of small improvements to everyday processes rather than one dramatic transformation. Production scheduling is a good example.
Checking inventory, evaluating capacity, reviewing dates, identifying exceptions, and releasing orders are routine responsibilities. They are also necessary.
When AI can perform more of the repetitive evaluation behind those responsibilities, experienced employees can spend more time responding to exceptions, coordinating production, and making decisions that require business judgment.
That is where AI in ERP becomes tangible. It is not simply an AI strategy for some point in the future. It is technology being applied to actual operational work.
There has been significant discussion about what artificial intelligence could eventually mean for ERP systems.
The more important question for businesses now is:
What work can AI actually help our employees perform today?
The Production Planning and Operations Agent provides one answer. A production supervisor can select orders requiring action, allow the system to evaluate materials, capacity, and scheduling information, respond to exceptions, and authorize the appropriate action.
Instead of navigating multiple applications and manually performing every check, the supervisor can focus on the decisions that require human knowledge.
That is a much more practical definition of AI-powered ERP. And production planning is only one business process where this model can be applied.
As additional AI agents become embedded into core business functions, organizations using SAP Cloud ERP will have opportunities to evaluate where repetitive work can be automated, where exceptions can be surfaced faster, and where employees can spend more of their time making higher-value decisions.
Talking about agentic AI is one thing. Seeing how an agent works inside a real ERP process makes the concept much easier to understand.
In our demonstration, Navigator Business Solutions Solution Advisor Zack Good walks through the Production Planning and Operations Agent and shows how a production supervisor can use Joule to evaluate and release production orders while maintaining human oversight of important decisions.
Watch the full original demonstration on YouTube.
Want to understand where AI agents could create value in your manufacturing processes? Navigator Business Solutions can help you explore how SAP Cloud ERP, embedded AI, automation, and modern business processes can support a more efficient way of operating.