Experts are questioning whether Salesforce’s Dreamforce demonstration of an AI agent working with SAP would breach SAP’s API policy if used in production. Salesforce showed its agent onboarding a supplier through SAP while users operated from Slack, raising concerns that Salesforce could take control of the user experience while SAP is reduced to a back-end system.

The dispute centres on whether Salesforce used SAP APIs or merely interacted with the software’s interface through service accounts. SAP’s policy restricts autonomous or generative AI systems from planning and executing API calls, as well as scraping or replicating application behaviour, unless they use SAP-endorsed architectures. Salesforce says the restrictions do not apply because the demonstration used SAP’s user interface; independent SAP consultants argue the approach may still bypass recommended architecture and best practice.