Q4 2026. That’s when SAP plans to make its new Joule agents for supply chain generally available. They include a Supply Chain Planning Assistant that SAP says will help planners “sense, simulate, and resolve exceptions in a single workspace” (SAP News, 8 October 2026).
SAP announced the agents at SAP Connect in October, as part of what it calls the “autonomous supply chain”. For people who run demand and supply planning every week, the practical question is what changes in their own work.
What SAP announced for planning
For planners, the announcement comes down to three things:
- Exceptions in one place. The Planning Assistant is meant to bring signals together, so a planner can see an exception, try out options and check the effect downstream without switching between screens and spreadsheets.
- Root causes and suggested fixes. Takeda, one of SAP’s early customers, says it will use the assistant for “AI-detected root causes and AI-suggested remediation”, plus automated tracking of planning performance.
- Agents that can act when allowed to. According to SAP, an insight in one area can inform the next and, “when authorized, trigger coordinated action.”
None of this is live yet. These are planned features, due in Q4.
What changes for a planner
Most planners start the week with a long list of alerts. They then spend hours working out which ones matter and why. Why did the forecast for that product family jump? Is the shortage at the plant real, or is the master data wrong? Who needs to know?
The assistant is aimed at that work. If it does what SAP describes, planners will spend less time finding problems and more time deciding what to do about them. Monday starts with a list of exceptions that already have a likely cause and a suggested fix.
SAP calls this “decision velocity”: how quickly a team can spot a change and agree on the response. Devesh Mishra, who leads product for SAP Supply Chain Management, puts the problem this way: each decision could be rational, but “the sum of them was not.”
What stays with your team
An agent works from the data and rules it is given. Three things don’t change:
- Data quality. Root-cause analysis on bad master data gives wrong answers that look convincing. Lead times, lot sizes and product hierarchies need to be right before an agent works from them. We wrote about this in Why Supply Chain AI Needs a Reality Check.
- Decision rights. SAP says leaders need to “define what it is authorized to complete” and “know who is accountable.” Your own team has to decide which fixes an agent may apply by itself and which need a planner to sign off.
- Judgement. A promotion, a call from a key customer, a supplier who often delivers late: planners know things that aren’t in the system, and the agent doesn’t.
How to prepare before Q4
There are three things you can do now, whatever SAP ends up shipping:
- List your ten most common exceptions and how planners resolve them today. This tells you what to give the agent first, and gives you a baseline to measure it against.
- Agree the decision rights. Decide which changes can go through automatically, which need a planner, and which go to S&OP. The business needs to make this call. IT can’t set it for you.
- Start with one task. When we built Ask Lily with Evergreen Garden Care, we picked one recurring piece of research and reporting work. It gave each demand planner 6.5 days a month back, and led to forty new AI ideas from other parts of the business.
Teams that already know which decisions they want help with will get value from these agents sooner. For the wider business case, see our post on the real ROI of SAP IBP.
Planning for AI in your SAP landscape?
Our consultants work in SAP IBP every day, and we’ve built AI agents that planners already use in their daily work. If you want to know what the new Joule agents could do for your team, get in touch. Talk to a planning expert →