Announcing Kumo’s Partnership with SAP: Bringing Relational Foundation Models to the Heart of Enterprise AI

At SAP TechEd this year, SAP publicly announced a new partnership with Kumo, calling it “one of the most exciting parts” of the event. We are thrilled to share this with our community. This collaboration brings together SAP’s leadership in managing the world’s most critical business data with Kumo’s next-generation Relational Foundation Model (KumoRFM), enabling a new class of predictive and generative AI capabilities for enterprise systems.

A Shared Vision for Enterprise AI

Businesses run on structured, relational data. From ERP to CRM, financial operations to supply chain networks, nearly every mission-critical system revolves around tables, relationships, and business semantics that require precision. Yet companies have historically needed different machine learning models for every predictive task. One model for demand forecasting. One for late payment prediction. Another for churn, sales scoring, or supply chain risk.

This fragmentation slows AI adoption and drives up cost.

SAP recently introduced a bold new paradigm: a table-native foundation model, sap-rpt-1, that brings in-context learning to structured business data. Instead of maintaining hundreds of narrow models across the business, companies can use a single foundation model to solve many predictive problems instantly by showing it a few examples from the task at hand. SAP has demonstrated the power of this approach across finance, supply chain, HR, and customer operations.

Kumo’s Relational Foundation Model was built to solve the same challenge. Our research and production systems enable universal prediction across any relational dataset with state-of-the-art accuracy, full explainability, and no task-specific training. This alignment in vision is what makes the partnership with SAP so compelling.

What SAP Announced at TechEd

During the SAP TechEd keynote, Richard Grandpierre (Head of Product Management, Business AI, SAP) introduced the partnership live to the global audience:

“One of the most exciting parts is announcing a brand new partnership. We are partnering up with Kumo. Kumo is a leading startup that built their own relational foundation model, in a very similar approach to what we have taken at SAP. They have innovated in quite a few domains with groundbreaking work, especially in explainability. Once you have run your analysis, explainability is fully built in. You can ask the model to explain the reasoning behind the patterns you are seeing. It can dig deeper and analyze the levers behind the predictions. We are very excited about this partnership. It fits our vision of being an open ecosystem where developers can choose the right model for each use case. You will be able to try Kumo’s models in the SAP platform very soon.”

This recognition from SAP reinforces the shared belief that foundation models for relational data will transform how companies build predictive intelligence into their core business systems.

Why Relational Foundation Models Matter for ERP and CRM

AI adoption in enterprise systems has always been constrained by three barriers: long development cycles, the need for specialized teams, and the difficulty of maintaining hundreds of custom models across large business environments.

Relational foundation models remove these barriers completely.

Instant Predictions for Hundreds of ERP and CRM Tasks

A table-native foundation model can use a few example rows to solve any prediction task across the business. Useful examples include:

• Finance: Assigning cost centers, tax codes, or ledger mappings
• Supply chain: Identifying shipments at risk
• Sales: Scoring leads or predicting opportunities
• Customer service: Forecasting support escalations
• Procurement: Predicting supplier delays

Instead of requiring a new AI project each time, teams can move from question to insight in minutes.

Accuracy That Outperforms Narrow Models

SAP’s published benchmarks show sap-rpt-1 cutting prediction errors by up to half compared to task-specific models. KumoRFM has demonstrated similar gains across real-world enterprise data sets. The reason is simple: foundation models learn the structure and semantics of business data at a global level, so they can generalize across problems more effectively than one-off trained models.

Fully Embedded Explainability

Kumo has pioneered explainable relational AI, so every prediction comes with a reason. Users can ask:

• Why did the model predict late payment?
• What factors are driving supply chain risk?
• Which attributes influence customer churn?

This level of transparency builds trust and enables quick iteration on business logic.

A Single Model Instead of Hundreds

By shifting from a landscape of separate models to one universal engine, companies reduce operational overhead, increase speed, and democratize access to predictive AI for business teams.

What This Partnership Means for SAP Developers and Customers

SAP has long been the steward of the world’s most important business data. Kumo was founded to bring predictive intelligence to the same type of relational data, without the heavy lift of training pipelines and model maintenance.

Together, the partnership will:

• Give SAP developers access to new predictive AI capabilities powered by KumoRFM
• Provide SAP customers with more choice in foundation models for ERP, CRM, and operational systems
• Accelerate the adoption of universal prediction for real-time business decisioning
• Extend SAP’s open ecosystem with another purpose-built model for structured data

SAP announced that developers will be able to try Kumo’s models in their platform soon, alongside SAP RAPID ONE and other AI capabilities in the Business AI ecosystem.

Looking Ahead

We are excited about the work ahead. The combination of SAP’s global reach and Kumo’s breakthrough research on relational foundation models opens the door to a new era of predictive intelligence across every business function.

Kumo will be sharing more updates soon as the integration progresses. We look forward to collaborating closely with the SAP community and helping customers unlock faster, more accurate decisions from their enterprise data.

Stay tuned for more.

Join our community on Discord.

Connect with developers and data professionals, share ideas, get support, and stay informed about product updates, events, and best practices.

Join the discussion