Forecasting just changed — and SAP is betting on it.
Data science has a new class of tool: pretrained foundation models for tabular data. Instead of fitting and tuning a model for weeks, you use one that already knows how to forecast — out of the box, with zero hyperparameters. SAP has agreed to acquire PriorLabs, the lab behind TabPFN, committing more than €1 billion to it. We bring that capability into your IBP today.
The Challenge: better forecasts, without a project
Classical forecasting means choosing models, tuning parameters, and maintaining them — a data-science effort most planning teams can't sustain.
Parametric models are also brittle: a window sized for one product misfires on another, new products have no history to fit, and volatile demand breaks the assumptions. The result is effort spent, and risk carried, for a forecast that may or may not hold up.
What if you could add a state-of-the-art model with none of that — and no risk?
How it works
We register a pretrained tabular foundation model as an external algorithm in your IBP forecast model, through SAP's own External Forecasting framework. When a planner runs the forecast, IBP calls the model, it scores every planning object, and the results are written straight back into the plan.
Everything stays where it belongs. Preprocessing, postprocessing, lifecycle and phase-in planning all remain inside IBP — the model simply returns a forecast, and IBP does the rest.
What you get
Pretrained Foundation Model
A model that already knows how to forecast tabular time series — no fitting, no historical data preparation, no training cycle to wait on.
Zero Tuning
No hyperparameters, no windows to size, nothing a planner can misconfigure. Robust where classical, windowed models turn brittle.
It Cannot Make Your Plan Worse
Added as one more algorithm in your model, IBP's own best-fit selects it only where it beats what you run today. The floor only goes up.
Native to SAP IBP
Delivered through SAP's own External Forecasting framework. No new tools, no retraining your planners — they run their normal forecast.
Works Everywhere
Planning-area, key-figure and periodicity agnostic — daily, weekly or monthly. Supports lifecycle and new-product planning.
Secure by Design
Runs on SAP BTP, connects with certificates — no password handed over. Stateless end to end; your planning data is never stored.
Getting started is simple
Onboarding is a one-time connect. After that, it is no-touch — planners just forecast as they always have.
Connect Once
A one-time communication-arrangement setup on SAP BTP — your IT, an afternoon.
Add the Algorithm
Register it in a forecast model and map your key figures — history in, forecast out.
Run Your Forecast
Best-fit decides where it wins, and the results appear in your plan. Nothing else changes.
Let your own system be the judge.
We don't ask you to trust a claim. Add it to your forecast model and let IBP's best-fit arbitrate — per planning object, on your own data. Worst case, it's never selected and nothing changes. Best case, your plan gets sharper.
Risk-free by design.
Why Integrated Business Planning Solutions
State of the Art, Instantly
Bring a modern foundation model into planning without a data-science project or a long implementation.
Risk-Free Adoption
Best-fit guarantees a reasonable baseline — the plan is always at least as good as today's.
No Change Management
No upskilling, no new way of working. Planners keep doing exactly what they do now.
Aligned with SAP Business AI
Built on SAP BTP and the technology direction SAP itself is investing in.