Sap Expands Its Ai Stack For Developers

BERLIN, GERMANY — SAP Build is nan German endeavor giant’s exertion improvement and process automation solution. Increasingly, it’s go nan location to nan company’s efforts to bring much AI-powered devices to developers. At its TechEd arena successful Berlin this week, SAP is expanding this level pinch a immense array of caller AI features and capabilities, connected apical of expanding its Joule supplier platform.

Here are immoderate of nan apical announcements from nan event.

More Integration With More Developer Tools

One of nan main themes for SAP astatine this year’s TechEd is “openness,” and that is reflected successful today’s announcements — particularly successful position of making it easier for developers to usage nan devices they are already acquainted pinch to prosecute pinch nan SAP platform.

In practice, this intends developers who want to usage agentic coding devices for illustration Claude Code, Cline aliases Windsurf tin now usage SAP Build improvement frameworks pinch section SAP Model Context Protocol servers. SAP argues that these MCP servers will let these coding agents to supply much applicable and aligned devices for nan developers successful its ecosystem.

For those developers who are particularly heavy successful nan SAP ecosystem and usage SAP’s proprietary ABAP (Advanced Business Application Programming) language, SAP will besides motorboat an ABAP-based MCP server that will expose features of nan ABAP Cloud and ABAP AI features to these coding agents.

Visual Studio Code users will get entree to a caller hold that will bring SAP Build features straight into their IDE. That hold will soon travel to Open VSX Registry — nan unfastened root replacement to Microsoft’s Visual Studio Marketplace — truthful different IDEs tin merge it, too.

SAP is besides partnering pinch n8n, nan celebrated AI business automation service, to let agents built successful SAP’s Joule Studio situation and n8n agents to activity together.

What’s New successful Joule Studio?

Talking astir Joule, SAP’s AI copilot that focuses connected endeavor business data, nan institution is besides expanding Joule Studio pinch a number of caller capabilities. Introduced earlier this year, Joule Studio is SAP’s low-code AI supplier and skills builder. It will go mostly disposable successful December 2025.

With this update, Joule Studio is getting caller features for illustration system-triggered agents, caller supplier monitoring devices and support for nan Agent2Agent protocol, which Google introduced earlier this twelvemonth to make it easier for developers to link AI agents to each other, nary matter what model they were built on.

And while SAP is putting a batch of accent connected its low-code supplier tooling, it is besides launching a fewer caller partnerships for developers who want to build their agents purely successful code. The SAP Business Technology Platform (BTP), nan company’s level for building AI-supported applications successful nan enterprise, will now support 2 celebrated unfastened root agent-building frameworks: Crew.AI and LangGraph.

Since ABAP remains halfway to SAP’s developer portfolio, it whitethorn besides not travel arsenic a astonishment that nan institution is bringing ABAP AI-assisted improvement to Joule, but possibly much importantly, SAP besides announced that it will merchandise “ABAP ample connection models (LLMs) trained connected ABAP codification and specialized for ABAP improvement will beryllium released adjacent year.”

MCP Gateway

It wouldn’t beryllium an arena focused connected agents without an announcement astir caller Model Context Protocol (MCP) capabilities. One caller instrumentality for developers is SAP’s MCP Gateway, which is portion of nan company’s wide SAP Integration Suite.

In summation to a number of caller API guidance features, including automatically fixing communal API issues for illustration latency surges, nan MCP Gateway now allows SAP customers to expose civilization API and integration flows for usage by agents successful Joule Studio. This will let developers to supply these agents pinch discourse from third-party devices and bequest SAP systems.

While AI agents are getting progressively tin astatine moving pinch modular APIs, it’s still easier to expose these devices done MCP, which, aft all, was custom-made for supplier use.

MCP and Memory for HANA Cloud

HANA Cloud, SAP’s managed database-as-a-service, is besides getting MCP support now.

“This allows agents to beryllium grounded successful afloat information context: navigating relationships crossed customers and suppliers, knowing geographic limitations done spatial data, and performing semantic searches done vector embeddings — each wrong a azygous in-memory engine,” Michael Ameling, nan president of SAP’s Business Technology Platform and personnel of nan company’s extended board, explained successful this week’s announcement.

In addition, HANA Cloud tin now besides usability arsenic nan representation for agents, giving them a semipermanent representation that developers tin entree crossed sessions.

Snowflake Partnership

All of these AI devices are only arsenic useful arsenic nan information they merge with, of course. After announcing a caller partnership pinch Databricks astatine nan opening of nan twelvemonth and an integration pinch Google’s BigQuery only a fewer weeks ago, SAP is now besides announcing a caller business pinch Snowflake. Similar to nan Databricks integration, nan thought present is to bring information and AI products person together. With this caller integration, SAP and Snowflake will present zero-copy sharing betwixt nan SAP Business Data Cloud — SAP’s afloat managed SaaS information level — and Snowflake.

Daniel Yu, SAP’s main trading serviceman for its Business Data Cloud, told maine that it’s important to statement that this integration is what SAP calls a “solution extension.” “It is thing wherever we endorse nan company, but we besides sell,” he explained. “So our sellers will now waste SAP Snowflake. As a customer, you could really spell into nan [Business Data Cloud] cockpit and proviso Snowflake.”

Bringing together SAP and non-SAP data, nan 2 companies argue, will let enterprises to bring much business-ready information to carnivore connected their information engineering and AI and instrumentality learning workflows, arsenic good arsenic make it easier to crushed their AI devices successful organizational knowledge and, by extension, make their AI agents much reliable.

Group Created pinch Sketch.