SAP Expands Its AI Stack for Developers

Wait 5 sec.

BERLIN, GERMANY — SAP Build is the German enterprise giant’s application development and process automation solution. Increasingly, it’s become the home to the company’s efforts to bring more AI-powered tools to developers. At its TechEd event in Berlin this week, SAP is expanding this platform with a vast array of new AI features and capabilities, on top of expanding its Joule agent platform.Here are some of the top announcements from the event.More Integration With More Developer ToolsOne of the main themes for SAP at this year’s TechEd is “openness,” and that is reflected in today’s announcements — especially in terms of making it easier for developers to use the tools they are already familiar with to engage with the SAP platform.In practice, this means developers who want to use agentic coding tools like Claude Code, Cline or Windsurf can now use SAP Build development frameworks with local SAP Model Context Protocol servers. SAP argues that these MCP servers will allow these coding agents to provide more relevant and aligned tools for the developers in its ecosystem.For those developers who are especially deep in the SAP ecosystem and use SAP’s proprietary ABAP (Advanced Business Application Programming) language, SAP will also launch an ABAP-based MCP server that will expose features of the ABAP Cloud and ABAP AI features to these coding agents.Visual Studio Code users will get access to a new extension that will bring SAP Build features directly into their IDE. That extension will soon come to Open VSX Registry — the open source alternative to Microsoft’s Visual Studio Marketplace — so other IDEs can integrate it, too.SAP is also partnering with n8n, the popular AI business automation service, to allow agents built in SAP’s Joule Studio environment and n8n agents to work together.Image credit: The New Stack.What’s New in Joule Studio?Talking about Joule, SAP’s AI copilot that focuses on enterprise business data, the company is also expanding Joule Studio with a number of new capabilities. Introduced earlier this year, Joule Studio is SAP’s low-code AI agent and skills builder. It will become generally available in December 2025.With this update, Joule Studio is getting new features like system-triggered agents, new agent monitoring tools and support for the Agent2Agent protocol, which Google introduced earlier this year to make it easier for developers to connect AI agents to each other, no matter what framework they were built on.And while SAP is putting a lot of emphasis on its low-code agent tooling, it is also launching a few new partnerships for developers who want to build their agents purely in code. The SAP Business Technology Platform (BTP), the company’s platform for building AI-supported applications in the enterprise, will now support two popular open source agent-building frameworks: Crew.AI and LangGraph.Since ABAP remains core to SAP’s developer portfolio, it may also not come as a surprise that the company is bringing ABAP AI-assisted development to Joule, but maybe more importantly, SAP also announced that it will release “ABAP large language models (LLMs) trained on ABAP code and specialized for ABAP development will be released next year.”MCP GatewayIt wouldn’t be an event focused on agents without an announcement around new Model Context Protocol (MCP) capabilities. One new tool for developers is SAP’s MCP Gateway, which is part of the company’s overall SAP Integration Suite.In addition to a number of new API management features, including automatically fixing common API issues like latency surges, the MCP Gateway now allows SAP customers to expose custom API and integration flows for use by agents in Joule Studio. This will allow developers to provide these agents with context from third-party tools and legacy SAP systems.While AI agents are getting increasingly capable at working with standard APIs, it’s still easier to expose these tools through MCP, which, after all, was custom-made for agent use.MCP and Memory for HANA CloudHANA Cloud, SAP’s managed database-as-a-service, is also getting MCP support now.“This allows agents to be grounded in full data context: navigating relationships across customers and suppliers, understanding geographic dependencies through spatial data, and performing semantic searches through vector embeddings — all within a single in-memory engine,” Michael Ameling, the president of SAP’s Business Technology Platform and member of the company’s extended board, explained in this week’s announcement.In addition, HANA Cloud can now also function as the memory for agents, giving them a long-term memory that developers can access across sessions.Michael Ameling, President, SAP Business Technology Platform (Credit: SAP).Snowflake PartnershipAll of these AI tools are only as useful as the data they integrate with, of course. After announcing a new partnership with Databricks at the beginning of the year and an integration with Google’s BigQuery only a few weeks ago, SAP is now also announcing a new partnership with Snowflake. Similar to the Databricks integration, the idea here is to bring data and AI products closer together. With this new integration, SAP and Snowflake will introduce zero-copy sharing between the SAP Business Data Cloud — SAP’s fully managed SaaS data platform — and Snowflake.Daniel Yu, SAP’s chief marketing officer for its Business Data Cloud, told me that it’s important to note that this integration is what SAP calls a “solution extension.” “It is something where we endorse the company, but we also sell,” he explained. “So our sellers will now sell SAP Snowflake. As a customer, you could actually go into the [Business Data Cloud] cockpit and provision Snowflake.”Bringing together SAP and non-SAP data, the two companies argue, will allow enterprises to bring more business-ready data to bear on their data engineering and AI and machine learning workflows, as well as make it easier to ground their AI tools in organizational knowledge and, by extension, make their AI agents more reliable.The post SAP Expands Its AI Stack for Developers appeared first on The New Stack.