Our goal is for our customers to achieve added value with AI solutions. When companies cite factors that prevent them from doing this – for example Complexity in integrationthe Regulatory Compliance, AI sovereignty or Data quality – let’s make it our mission to remove these obstacles.

That’s why in Q4 2025, we introduced new models and offerings for a sovereign cloud, updated numerous products, and announced new and deeper partnerships to enable customers to better use AI. These are described in detail below.

SAP Business AI for IT and developers

The Agent Builder in Joule Studio
General availability

Users and IT professionals can now use the Agent Builder in Joule Studio to create powerful AI agents for automating highly complex business processes. The tool enables the development of agents that can plan, logically analyze and dynamically coordinate multi-step workflows in both SAP and third-party systems. This means that workflows can also be effectively automated in scenarios where standard automation falls short.

With Joule Studio Agent Builder, companies can reduce the time spent on common tasks by up to 40%* and the time spent on creating and deploying custom agents by up to 35%*. This allows them to make decisions much more quickly and make processes more efficient.

Further information can be found here.

Extensions for the Generative AI Hub in AI Foundation
General availability

SAP-RPT-1, the first relational foundation model for corporate data

SAP has introduced SAP-RPT-1, its first Relational Foundation Model for enterprise data, which is provided together with a no-code test environment.

SAP-RPT-1, unlike Large Language Models (LLMs), is a base model that represents a new category of AI models designed specifically for relational and structured business data. SAP-RPT-1 is pre-trained, so customers benefit from significantly less time and cost of training models that specialize in very narrow application areas. With answers based on verified business data, the model provides reliable, fact-based forecasts, providing the accuracy and reliability that matters in critical business operations.

When performing analysis and forecasting tasks, SAP RPT-1 applies contextual learning. By providing sample datasets directly in the API call, users can perform classification and regression on tabular data. The model can be used as a ready-to-use solution and integrated into applications and business processes. The first version supports common forecasting scenarios, including binary classification and multiclass classification as well as numerical regression.

With the SAP RPT Playground, SAP also provides an interactive, web-based test environment in which customers can test the contextual learning functions free of charge and without programming using their own data or sample data sets provided by SAP.

SAP-RPT-1 is available for production use in the Generative AI Hub in AI Foundation. Two options are available: SAP-RPT-1-small for very fast forecasts and high throughput and SAP-RPT-1-large for the greatest possible accuracy.

Test SAP-RPT-1 for free and read the documentation.

The basic model SAP-ABAP-1

In order to enable customers and partners to develop their own AI-supported use cases to increase developer productivity, the basic model SAP-ABAP-1 is now available in the Generative AI Hub. Trained with over 250 million lines of ABAP code, 30 million lines of CDS code and extensive technical documentation, SAP ABAP-1 is specifically designed to efficiently understand, explain and quickly access knowledge, best practices and current innovations in the area of ​​ABAP code.

Customers can test the new model for free as part of the Generative AI Hub trial. Additional features will be available in 2026.

Further information can be found here.

The prompt optimizer

Developed in close collaboration with Not Diamond, Prompt Optimizer helps automate and accelerate the creation of effective AI prompts in leading models. This allows users to adapt the prompts to any model for their use cases without having to manually recreate them.

Further information can be found here.

The orchestration registry

Developers can now manage the configuration lifecycle for orchestration workflows, such as backing up, versioning, and deleting orchestration configurations.

Further information can be found here.

New models are available

New models are supported including Perplexity Sonar, Sonar Pro, Anthropic Claude 4.5 Sonnet, Anthropic Claude 4.5 Haiku, Cohere Command A Reasoning and Gemini 2.5 Flash Lite.

For more information on new and discontinued models, see one SAP note.

Try Generative AI Hub free for 30 days.

Extensions for SAP Document AI
General availability

Extracting information using computer vision

Schema administrators can now choose between Plain text extraction or advanced text and image processing (view) choose. When enabled, documents are processed using a multimodal model that analyzes visual elements such as hazard pictograms, stamps, signatures, logos, diagrams, and labels along with text. This enables more precise data capture for documents with many visual elements (e.g. safety data sheets or declarations of conformity). The feature improves the completeness and accuracy of extracted data, reduces the burden of manual labeling and verification through automatic extraction of visual elements, and optimizes cost and performance through schema-based control.

Further information can be found here.

Processing email attachments

Users can now process email attachments together with the email body or separately. You benefit from optimal flexibility and improved data extraction.

Find out more about Channels and Workflows.

Document workflows

Users can now quickly and easily define multi-step workflows to process documents according to their individual requirements. The new workflow function combines basic functions of SAP Document AI to automate and optimize complex tasks.

Workflows can be triggered automatically across input channels without the need for additional tools or integrations. Alternatively, users can upload a file and start the workflow manually. Workflows go beyond extraction and classification to help users with tasks such as email processing, content-based routing, and automated processing.

Find out more about workflows.

Integrated transport management

With an integration with SAP Cloud Transport Management, SAP Document AI enables users to export and import their schemas for SAP Document AI service instances (e.g. development, quality assurance, and production) using the transport feature. This ensures consistent schemas and workflows across all instances and enables better collaboration between teams and systems.

Further information can be found here.

AI-supported generation of requirements in SAP Cloud ALM
General availability

Consultants can now automatically generate high-quality business requirements directly from Fit-to-Standard workshop protocols. The new function in SAP Cloud ALM analyzes conversation content and then fills out a predefined template. It integrates SAP best practices to propose solutions. This means consultants need less time for manual transcription and can focus on strategic review and optimization.

This automation reduces the time spent creating requirements by up to 50%* and then writing user stories by up to 20%*. In this way, project documentation can be significantly accelerated.

Further information can be found here.

AI-powered user training and change management in the SAP Micro-App Hub
SAP Early Adopter Care program

Training professionals and content creators can now streamline the entire SAP user training lifecycle, accelerate training adoption, and reduce costs through the SAP Micro-App Hub. Connected to SAP Signavio and SAP Cloud ALM, the feature can analyze project scope, identify training needs, and automatically create tailored training content for each user role that aligns with business goals.

In this way, not only can up-to-date materials for various authoring and training tools be created within a very short time, but the time required for the initial analysis of training needs is also reduced by up to 60%*. Content can be created up to 50%* faster, ensuring faster onboarding and more accurate alignment of user training with SAP updates.

Further information can be found here.

Philipp Herzig is CTO of SAP SE.

*Disclaimer: This article describes the expected benefits for companies. All calculations are estimates based on SAP customer case studies, SAP benchmarks and other research. Actual benefits may vary and may be affected by additional factors not considered in this article. The information is provided without warranty of any kind, either express or implied. SAP assumes no responsibility for any damages associated with the use of this article. On the side https://www.sap.com/legal-notice Please see “Legal Notice” for the terms of use, disclaimers, mandatory disclosures and restrictions applicable to this material.