SAP shifts focus to autonomous enterprise with new AI platform
SAP announced a strategic pivot at its annual Sapphire conference this week, moving away from building feature-rich applications toward automating business processes with AI agents and assistants. The shift centers on a new Business AI Platform designed to let workers act as controllers of AI workflows rather than manual task executors.
Richard Grandpierre, VP of product management for SAP's Business AI organization, said the change represents a fundamental rethinking of how users interact with enterprise software. "The user becomes a different paradigm for how people engage with SAP software," he said. "They become the controller of a process that orchestrates a fleet of assistants and agents that do the work for them."
Joule Work becomes the central interface
SAP introduced Joule Work, a new workspace that serves as the single connection point for all user interaction. Unlike the previous Joule copilot focused on natural language queries, Joule Work functions as a configurable dashboard where users can prompt the assistant, fetch data, surface insights, and execute transactions in real time.
The interface works across desktop, web, and mobile devices. Grandpierre described it as a way to consolidate work that was previously scattered across multiple applications.
RPT-1 model gains context expansion and explainability
SAP upgraded its RPT-1 (Rapid 1) foundational model for tabular data prediction to version 1.5, adding three capabilities that address current limitations.
The upgrade introduces retrieval-augmented prediction (RAP), which allows the model to ingest datasets of any size-a significant expansion from previous context window constraints. The model now includes explainability features that use large language models to describe how it reached predictions and express confidence levels. Users can chat with the model in a playground to understand key influences on predictions.
SAP also added support for multi-modal tabular models, allowing organizations to orchestrate between models from Gemini, OpenAI, AWS, and other vendors. The company announced last week it acquired Prior Labs, a German startup specializing in tabular AI models, to expand this capability.
Knowledge Graph moves into production
SAP's Knowledge Graph-a semantic representation of data structures, APIs, and database tables-is now embedded in Joule Work. The addition helps the assistant navigate complex data landscapes and retrieve accurate results across S/4HANA public and private cloud, Ariba, and SuccessFactors.
Grandpierre said the Knowledge Graph addresses a previous limitation where Joule performed inconsistently depending on whether SAP had trained it on specific entities. "In the past, Joule was a bit limited," he said. "For others we didn't train it for because the Knowledge Graph wasn't in place, it didn't work so well."
On-premises customers get limited AI access during migration
SAP announced that customers committed to migrating to S/4HANA can access certain AI features on their legacy on-premises systems while the migration proceeds. The program requires organizations to have converted more than 50% of their maintenance to the cloud to qualify.
Grandpierre acknowledged the technical barriers. Cloud-based AI agents can access predictable data endpoints, but on-premises systems-particularly older ECC installations with custom modifications-require substantial engineering effort to support the same capabilities. "It will require a lot of heavy lifting from us and the customer to deliver certain AI capabilities to on-premises customers," he said.
The move does not signal a retreat from the 2027 S/4HANA migration deadline. SAP remains focused on cloud delivery, where it can provide AI features reliably without extensive customization work.
Smaller companies can implement AI without external consultants
Grandpierre outlined three paths for mid-market companies with under $1 billion in revenue to adopt the new AI capabilities. Organizations can work with consulting partners, build in-house teams with AI expertise, or use SAP's managed deployment model where the vendor handles provisioning and integration.
SAP is shifting more of the AI infrastructure into its managed services to reduce implementation burden. The company created a Regional Implementation Group to support customers building capabilities internally. "We have a strong interest in making this easy and seamless so we can get adoption of our AI capabilities and learn from that," Grandpierre said.
For product development professionals evaluating enterprise AI strategies, these announcements reflect how vendors are moving beyond feature expansion toward process automation. The architectural choices-federated model support, knowledge graphs, managed deployment-address real constraints that development teams face when building AI systems at scale.
Understanding how SAP is structuring its platform choices can inform similar decisions in your own product roadmap. See AI for Product Development for resources on building AI into enterprise products, or explore the AI Learning Path for Product Managers to deepen your knowledge of AI strategy and implementation.
Your membership also unlocks:
