SAP today added a series of artificial intelligence (AI) agents to its portfolio of business applications along with a tool that enables its customers to build and deploy their own AI agents.
Additionally, SAP in partnership with Databricks unfurled SAP Business Data Cloud platform to streamline access to the data organizations need to drive AI applications.
Those capabilities combined will make it simpler for organizations to build agentic AI applications capable of driving an end-to-end process without losing semantic context, says Muhammad Alam, head of product and engineering and a member of the executive board for SAP.
In the absence of a platform such as SAP Business Data Cloud to manage and govern data, it isn’t practical for organizations to infuse AI into business processes, he adds. “Without good data you basically don’t have AI,” says Alam.
Initially, the SAP Business Data Cloud will be made available on the Amazon Web Services (AWS) cloud, with support for Google Cloud Platform and Microsoft Azure to closely follow. SAP also plans to provide access to external data using zero-copy tools that make use of data virtualization technologies to provide access.
The AI agents developed by SAP are extensions to the Joule copilot the company created to provide a natural language interface to large language models (LLMs) that SAP trained. AI agents take that capability to the next level by invoking the reasoning engines in those LLMs to automate a specific task.
It’s not clear how SAP plans to make it possible for organizations to orchestrate multiple AI agents that will be performing tasks alongside end users of SAP applications, but the potential to conspicuously improve productivity is clearly significant. However, SAP has already developed its own SAP Knowledge Graph that will be extended to include integrations with AI agents.
SAP has long been making a case for relying on packaged applications it develops to manage business processes versus each organization having to develop, for example, its own accounting software. In fact, many organizations have been waiting to see to what degree providers of applications are going to infuse AI agents and copilots into applications before determining where best to apply their own more limited IT budgets to deploying an AI application in a production environment.
Arguably, one of the most challenging issues with AI is determining what projects to move forward when providers of applications such as SAP might provide a similar capability to every customer. In effect, SAP is already turning AI agents into a set of table stakes that every organization will eventually invoke simply to remain competitive.
That doesn’t mean organizations won’t need to customize existing AI agents or build their own, but it does mean the range of instances where they might be required to make that effort is already starting to narrow as application providers start embedding their own AI agents across their entire portfolios.
