What is tabular AI and why are organisations building a strategy around it?

Tabular AI applies machine learning to structured data or the rows and columns across enterprise data sources that power most business decisions. Organisations are building a strategy around it today because tabular foundation models (TFMs) make it practical to generate accurate predictions at scale, without requiring dedicated model training for every use case. Tabular AI on SAP goes further, bringing TFMs and large language models (LLMs) together with orchestration, centralised management and governance through a single AI service.

How are tabular foundation models (TFMs) different from traditional large language models (LLMs)?

Tabular foundation models are AI systems pretrained on large, structured datasets to learn common tabular data patterns. Unlike LLMs, which are trained on text, tabular foundation models operate natively on rows and columns, fully understanding the rich semantics and relationships in structured data. As fully managed, pretrained models, TFMs eliminate the need for customised training, accelerating time to value while reducing cost and complexity.

What is the difference between a prediction, a predictive insight and a tabular AI strategy on SAP?

A prediction is the statistical output of a tabular AI model. A predictive insight adapts that prediction to prompt users to act with increased precision, based on a complete understanding of the semantics and relationships in your structured data. Tabular AI on SAP goes further by operationalising predictions and insights with specialised capabilities, such as tabular orchestration, context selection and registry, that increase the contextualisation and reasoning of the AI models based on your business data. It helps you to unify your AI practice, cut costs and deliver the management, governance and observability needed to scale.

What is SAP-RPT and how does it help organisations make predictions from structured business data?

SAP-RPT (relational pretrained transformer) is a fully managed, table-native foundation model that uses predictive capabilities such as regression and classification and contextualisation options like in-context learning, to deliver accurate, explainable predictions from structured business data. It works across business use cases, such as finance, supply chain, sales and HR, without requiring any model training or fine-tuning. In comparison with other tabular model alternatives, SAP-RPT are trained with real world-business data from decades of experience across domains and with the business process knowledge that only SAP can provide. Explore SAP-RPT

What is SAP AI Core and what role does it play in scaling tabular AI workflows?

SAP AI Core is the fully managed, scalable AI runtime on the SAP Business AI Platform for deploying and operating AI at enterprise scale. For tabular AI, it brings large language and tabular foundation models, orchestration, context registry and selection, targeted retrieval and batch inference together in one service. It balances cost, performance and accuracy through model choice and automatically surfaces the most relevant prediction examples from large datasets, processing data efficiently and asynchronously.

How can I access tabular foundation models?

Tabular foundation models are available through the SAP AI Core model library, including the latest versions of SAP-RPT and TabPFN from PriorLabs. Developers can build customised AI applications with them, while models like SAP-RPT already power SAP solutions and Joule Agents, embedding intelligent predictions directly into business workflows.

Is Prior Labs an SAP company?

Yes, SAP completed its acquisition of Prior Labs in July 2026, bringing one of the world’s leading tabular foundation model (TFM) research teams into the SAP family. Prior Labs will continue to operate as an independent entity, with SAP having previously committed to invest more than €1 billion to scale it into a globally leading frontier AI lab for the structured data that underpins the world’s businesses.

Tabular AI

Turn structured data into accurate, explainable predictions that your team and AI agents can act on.

Try SAP-RPT playground

Try SAP AI Core now

Zgradite stvari, ki so pomembne. Naučite se ustvariti sisteme, ki uporabljajo umetno inteligenco in pretvarjajo namen v izvedbo, 27.–29. oktober 2026.

Raziščite dogodek

What is tabular AI on SAP?

Tabular AI on SAP is purpose-built AI for structured data. It combines SAP and third-party benchmark-leading tabular foundation models with the orchestration capabilities in SAP AI Core to deliver faster, contextualised and more accurate predictions at scale, without custom model training or fragmented pipelines.

Unify tabular foundation models and capabilities for predictive insights at scale:

Access pretrained, fully managed SAP and third-party tabular foundation models built for structured data.
Extend the reach of tabular AI with SAP models and tools to data outside SAP systems.
Uncover powerful insights through advanced what-if and additional analysis, such as causation behind business outcomes.
Orchestrate tabular workflows across models and predictive capabilities, even when data is changing, incomplete or missing.
Monitor and govern tabular AI performance across your organisation.

Benefits

Leverage

Leverage leading tabular foundation models built to understand business data.

Scale

Scale tabular AI with a single managed service.

Access

Access built-in predictive insights where decisions happen in business workflows, Joule Agents and customised AI applications.

Features

Centralised control

Unify orchestration, observability and governance through tabular AI.

Traceable predictions

Explain predictions with the tables, rows and columns that influenced results.

Repeatable tabular AI strategy

Standardise predictions across teams so every function consistently acts on the same trusted insights.

Reduced AI costs

Consolidate specialised and benchmark-leading tabular foundation models to deliver predictive value to multiple use cases.

Predict from any source

Run predictions on data wherever it lives: across distributed sources, historical or continuously updated records and tables of any size.

AI where it’s needed

Operationalise insights through customised AI applications or embed directly into Joule Assistants and Joule Agents.

See how customers are succeeding with tabular AI on SAP

Clariant achieves 90% tax determination accuracy

Learn how Clariant, a leading global speciality chemicals company, achieved 90% accuracy across invoices spanning multiple countries, jurisdictions and product classifications with SAP RPT at the core of its tax determination pipeline. The result: tax experts spend less time on routine determinations and more time focused on complex exceptions and strategic tax planning.

Frequently asked questions

: Tabular AI applies machine learning to structured data or the rows and columns across enterprise data sources that power most business decisions. Organisations are building a strategy around it today because tabular foundation models (TFMs) make it practical to generate accurate predictions at scale, without requiring dedicated model training for every use case. Tabular AI on SAP goes further, bringing TFMs and large language models (LLMs) together with orchestration, centralised management and governance through a single AI service.
: Tabular foundation models are AI systems pretrained on large, structured datasets to learn common tabular data patterns. Unlike LLMs, which are trained on text, tabular foundation models operate natively on rows and columns, fully understanding the rich semantics and relationships in structured data. As fully managed, pretrained models, TFMs eliminate the need for customised training, accelerating time to value while reducing cost and complexity.
: A prediction is the statistical output of a tabular AI model. A predictive insight adapts that prediction to prompt users to act with increased precision, based on a complete understanding of the semantics and relationships in your structured data. Tabular AI on SAP goes further by operationalising predictions and insights with specialised capabilities, such as tabular orchestration, context selection and registry, that increase the contextualisation and reasoning of the AI models based on your business data. It helps you to unify your AI practice, cut costs and deliver the management, governance and observability needed to scale.
: SAP-RPT (relational pretrained transformer) is a fully managed, table-native foundation model that uses predictive capabilities such as regression and classification and contextualisation options like in-context learning, to deliver accurate, explainable predictions from structured business data. It works across business use cases, such as finance, supply chain, sales and HR, without requiring any model training or fine-tuning. In comparison with other tabular model alternatives, SAP-RPT are trained with real world-business data from decades of experience across domains and with the business process knowledge that only SAP can provide.

Explore SAP-RPT
: SAP AI Core is the fully managed, scalable AI runtime on the SAP Business AI Platform for deploying and operating AI at enterprise scale. For tabular AI, it brings large language and tabular foundation models, orchestration, context registry and selection, targeted retrieval and batch inference together in one service. It balances cost, performance and accuracy through model choice and automatically surfaces the most relevant prediction examples from large datasets, processing data efficiently and asynchronously.
: Tabular foundation models are available through the SAP AI Core model library, including the latest versions of SAP-RPT and TabPFN from PriorLabs. Developers can build customised AI applications with them, while models like SAP-RPT already power SAP solutions and Joule Agents, embedding intelligent predictions directly into business workflows.
: Yes, SAP completed its acquisition of Prior Labs in July 2026, bringing one of the world’s leading tabular foundation model (TFM) research teams into the SAP family. Prior Labs will continue to operate as an independent entity, with SAP having previously committed to invest more than €1 billion to scale it into a globally leading frontier AI lab for the structured data that underpins the world’s businesses.