SAP is extending its embrace of generative AI to the developer community.
At SAP’s TechEd event hosted last week in Bangalore, India, the company announced innovations for generative AI application development and new vector database capabilities, along with AI-centric learning opportunities for developers.
Chief among the announcements:
- A new vector database for SAP HANA Cloud, making it easier for AI models to manage unstructured data and enhancing interactions between large language models (LLMs) and organizations’ mission-critical data.
- The AI Foundation for SAP Business Technology Platform (BTP), described as “a new one-stop shop for developers to create AI- and generative AI-powered extensions and applications on SAP BTP.”
- New “semantic onboarding” capabilities for SAP Datasphere, aimed at helping customers deliver data with business context and logic intact.
- New generative AI capabilities and deeper integration between SAP Build Process Automation, SAP Signavio, and SAP Integration Suite, to accelerate integration, automation, and monitoring of business processes in heterogeneous environments.
- New SAP Integration Suite Edge Integration Cell runtime, enabling hybrid deployments and giving customers in heterogeneous environments the flexibility to design integration flows in the cloud and then manage them on-premises.
Collectively, the innovations will enable enterprises to build their own applications and improve their productivity in working with generative AI and large language models—a major step forward for tech professionals seeking to develop in integrated hybrid cloud landscapes.
“Today’s dynamic technology and business landscape means every developer needs to be an AI developer,” said Juergen Mueller, Chief Technology Officer and member of the Executive Board of SAP SE, who'll deliver the opening keynote at ASUG Tech Connect this week.
“The innovations we’re launching at SAP TechEd, from AI-infused pro-code tools to a one-stop shop to create generative AI extensions and applications on SAP Business Technology Platform, support the developers at the heart of the AI revolution and provide them with resources they need to transform the way businesses run.”
SAP Build Code
The new tools will also offer deeper interoperability with SAP’s proprietary ABAP Cloud development model and programming language through a side-by-side extensibility model, enhancing the ability of pro-code developers to collaborate with low-code or no-code developers working in SAP Build.
“Our philosophy is that everybody can be a developer,” said JG Chirapurath, Chief Marketing and Solutions Officer at SAP, in an interview with ASUG. “If you look at SAP applications, we have applications for every room of the house—finance, HR, marketing and sales, supply chain—and to all of those users, our message is that they can pick up SAP Build and be productive. They don't have to wait for central IT to assign them resources to get started.”
Bringing together previously disparate design and run-time services, alongside a unified application lifecycle management software with built-in governance capabilities, the suite is meant to enhance collaboration between professional developers, citizen developers, and IT departments.
SAP Build Code will be made available to select customers and partners initially before becoming generally available in Q1 2024.
Vector Database Features for SAP HANA Cloud
SAP also announced that it is enhancing SAP HANA Cloud, its database management system, to manage unstructured data (such as text, images, or audio) through a vector datastore.
“With this announcement, we get closer to the holy grail of generative AI where you can combine the strength of large language models with accurate real-time company data,” Mueller said on an analyst call.
Retrieving data and context in the native language of LLMs, vectors (long arrays of numbers that give value to a large number of attributes or dimensions) will allow SAP HANA Cloud users to gain more context on their business data. With this functionality, the SAP HANA Cloud database can be leveraged to continuously train LLMs underneath organizations’ generative AI applications with up-to-date, relevant business information. This, according to SAP, will simplify interactions with LLMs and encourage developers to build generative AI into their own applications.
“When you train large language models, it’s not enough to just train them with transactional data or analytical data that's sitting in a relational database,” said Chirapurath. “You have to be able to train, store, and associate all types of data to train your LLM. Databases that can do that—vector databases—employ a method called “vector similarity search,” which asks, if you search for a type of document, what else can you find inside this corpus of data that looks like it?”
The vector search functionality, available in Q1 2024, will also be available in all three main hyperscalers’ clouds, for those developing outside of SAP applications.
AI Foundation brings together various tools to empower developers to start building business-ready AI applications and extensions. Providing ready-to-use AI services, access to LLMs, vector database capabilities, and AI runtime and lifecycle management, AI Foundation is another incentive for developers to build their solutions on SAP BTP.
Under the banner of AI Foundation, SAP plans to offer AI services, generative AI management, AI workload management, access to business data and context, and lifecycle management based on SAP HANA Cloud and different SAP BTP features, such as the SAP AI Core runtime and SAP AI Launchpad service for managing AI use cases or scenarios.
AI Foundation will later this quarter include a generative AI hub within SAP AI Core, giving developers access to LLMs from different providers to orchestrate multiple models and take advantage of tooling and testing capabilities. This hub will also connect to the vector search capabilities within SAP HANA Cloud, allowing developers to incorporate contextual data into specific use cases.
‘Semantic Onboarding’ for SAP Datasphere
SAP rolled out new “semantic onboarding” capabilities for SAP Datasphere, aimed at helping customers deliver data with business context and logic intact. With these semantic onboarding capabilities, organizations will be better equipped to build and deliver their own business-ready data products. Expanded capabilities include:
- Import entities: bringing in business context from source systems such as SAP S/4HANA and SAP Business Warehouse, and automatically reusing semantical definitions and associations.
- Simplified data sharing: access to customer and partner-managed data products via APIs, in and outside of SAP Datasphere.
- Prebuilt business content: including ready-to-use templates, dashboards, and reports that aim to de-risk data and analytics projects.
“SAP Datasphere is going to become a key element for partners and customers in terms of rationalizing the data complexities they face,” said Joshua Greenbaum, Principal at Enterprise Applications Consulting. “Every company deals with this, regardless of size, industry, or geography. They have major problems with consolidating their different kinds of data into processes that can be leveraged effectively.
“You can’t do any AI without clean, well-governed data, and that applies for operations and transactional functions as well,” added Greenbaum. “The focus on data quality in SAP Datasphere, amid the integration of data that generative AI is creating, will have a positive impact.”
SAP Integration Suite
AI-enabled process recommendations, simplified user experience, and easier-to-implement process improvements are among generative AI capabilities and integrations announced between SAP Build Process Automation solutions, SAP Signavio solutions, and SAP Integration Suite.
Collectively, these innovations will allow customers to better integrate, automate, and monitor business processes across SAP and non-SAP applications. SAP Signavio will now connect directly to integration flows in SAP Integration Suite, allowing process analysts to analyze process performance across both SAP systems and software from other vendors.
With Joule, SAP Build Process Automation can now assist users in creating workflows and business rules, while increased interoperability with SAP Build Apps and access to prebuilt integrations and automations in SAP Business Accelerator Hub will make it easier to develop automations. Advanced event mesh capabilities for SAP Integration Suite can capture common business events, initiating workflows in SAP Build Process Automation and drawing support from process mining data from SAP Signavio.
A new Edge Integration Cell runtime is now generally available for SAP Integration Suite, enabling improved integration and API management in both customer-managed data centers and private clouds.
For those customers in heterogeneous environments, this will ensure greater flexibility and control over integrations and APIs while allowing them to deploy integration flows designed in the cloud in their on-premises environments; this hybrid deployment option allows customers to keep certain integrations local while benefitting from cloud design capabilities.
The runtime can support running integrations in private landscapes without routing data to the cloud, as well as remote and offline integration scenarios (such as systems in warehouses or on ships), while maintaining compliance with region-specific laws and regulations.
“SAP is selling itself short by only focusing on generative-AI capabilities in recent announcements,” said Greenbaum, who referred to the TechEd announcements as collectively telling a “strong story."
“With SAP Build Code and SAP Integration Suite, SAP is encouraging customers and partners to build and manage next-generation functionality across the board, whether we’re talking about generative AI or industry-specific activities,” he added. “And SAP is clearly looking to support development in integrated hybrid cloud environments.”
Two new knowledge resources—the SAP BTP Guidance Framework for designing, building, and deploying applications using SAP BTP, and a new developer’s guide for SAP BTP, focused on implementing business applications on SAP BTP using ABAP Cloud and SAP Cloud Application Programming Model (CAP)—are now available.
SAP Analytics Cloud, which already features a just ask feature akin to Joule, allowing users to quickly receive data-driven insights in response to their questions, will launch planning and analytics features to enhance data modeling and data integration, including a user-friendly data modeling experience and a new lightweight viewer to improve performance.
SAP also announced new role-based certification and free learning resources aimed at upskilling developers, particularly around back-end developers using the ABAP Cloud development model.
SAP announced that it has joined the corporate affiliate program at the Stanford Institute for Human-Centered Artificial Intelligence, connecting SAP researchers and engineers to the academic community to examine the role of generative AI in business.
And the company rolled out a new rewards program for RISE with SAP customers who run SAP S/4HANA Cloud, private edition, with an ERP clean core. As part of the program, RISE customers who maintain a clean core can earn annual credits through the SAP BTP Enterprise Agreement to pay for SAP BTP usage.
For more on these announcements, read SAP's official TechEd report.