BERLIN — One of SAP’s key themes during its TechEd 2025 event in Berlin this year was “openness.” While it’s not an open source company in the traditional sense, over the last decade, SAP has regularly been among the top 10 open source contributors in terms of active contributors, and it’s doubling down on this. But SAP has also shifted to being more open in other ways, which became clear in a number of conversations I recently had with some of its executives.

This shift to being more open takes several forms. It includes opening up the SAP developer ecosystem to more tools like its embrace of VS Code and various agentic coding agents and platforms like Claude Code, Cline and Windsurf, as well as frameworks like Crew.ai and LangGraph. Among other projects, SAP is also backing the Agent2Agent protocol to help agents talk to each other, no matter the framework they were written in or the platform they run on.

An Open Data Ecosystem

But maybe even more importantly, SAP is looking to make its data ecosystem more open as well. Sometimes, that takes the form of new partnerships with the likes of Databricks, Google and, most recently, Snowflake. But it’s also about offering Model Context Protocol (MCP) servers to allow third-party agents to access tools and data on its platform.

“This is a shift,” Michael Ameling, SAP’s president of its Business Technology Platform and a member of its extended board, told me. “Ten years back, or five years back, I would say it was still different, and especially in terms of data lakes and the like. But we have shifted the strategy in that perspective — also simply due to the reality that a lot of especially big customers started to create their data lakes in Databricks, in Snowflake and with other vendors. And so this is where they are, and it would be insane to force them now to move all their data back.”

Acknowledging this reality, SAP decided to partner with these companies and focus on adding value on top of these integrations for the joint customers, Ameling explained. The company says it’s open to working with other databases, even as it obviously remains committed to its core database products like SAP HANA Cloud.

“Our strategy clearly is continuing to make sure we make HANA Cloud the best and the most robust database, not just for customers out there, for their use cases, but certainly for all of our SAP stack from applications to data to AI,” Muhammad Alam, who heads up all of SAP’s business software applications, noted in a press conference during TechEd.

Similarly, Bharat Sandhu, the chief marketing officer of the company’s Business Technology Platform, offered a similar assessment of why SAP is going this route. “We just want to do what we do really well, which is letting customers create solutions to run their businesses at the end of the day. And for us, at least on the Build side, one aspect of that is providing really good context, like we’re doing through our MCP servers, whether it’s business data or, increasingly, even tooling.”

In this new world, where many of the large enterprise software companies are now trying to position themselves as the place for building and hosting AI agents, there is also an increasing expectation that these agents will be able to use the tools and data from various sources. Context is, after all, what makes these agents do their best work — and multi-agent systems that talk to agents from multiple vendors may soon become the norm. In that world, open data ecosystems aren’t a choice but a necessity.

Besides the needs of AI systems, though, SAP’s customers clearly want to be able to use more SAP and non-SAP data for the specific business needs, and I wouldn’t be surprised if we saw more direct integrations with more data platforms soon.

An Open Developer Ecosystem

All of this is also true about developer tooling in the SAP ecosystem. In multiple conversations I had with SAP executives, the topic of developer choice came up. Developers are picky about the tools they use, but in the world of AI developer tooling, preferences keep changing rapidly.

“New things come in and come out,” Sandhu told me. “So it’s best to be open to serve our audiences, which are very diverse around the world.”

When it comes to models, SAP is also taking the open route with the release of an open-weight version of its RPT-1 relational foundation model, which aims to make predictions on tabular data with the accuracy of specifically trained machine learning models (though at a slower speed). Though there is a difference between the open and the closed models, which were trained with an enriched data set, that’s a step in the right direction, especially given that these kinds of models are still very nascent.

Ameling, similarly, noted that with the AI ecosystem moving so quickly and new tools and models appearing at a rapid clip, flexibility is key.

“I strongly believe you need this flexibility, because if not, which horse would you bet on? I think it’s like looking into a glass ball,” Ameling said. “So this is definitely where we say, ‘let’s have an open strategy, like we have for the LLMs, like we have for the data centers, and also all the hyperscalers and the like.’ As soon as something new comes, just talk to them. APIs are there. Let’s integrate it and we can hook up this innovation. This is what I mean by always bringing a future-ready architecture to the customer.”

In that context, it makes sense for SAP to want to partner with a variety of AI-centric tools like Claude Code, Cursor, n8n and Windsurf (and other coding agents will be able to use these tools, too, even without a formal partnership), all while also bringing more of its own tooling directly into Visual Studio Code with the help of a new extension.

SAP is even bringing support for ABAP, its proprietary high-level programming language, to Visual Studio Code. Traditionally, Eclipse has been the main development environment for professional coders.

SAP’s Sandhu told me that the extended Visual Studio Code support was driven by customer demand.

“At the end of the day, what we want to do is help customers build their solutions faster,” he said.

TRENDING STORIES