SAP adds Amazon Bedrock into AI Core, streamlining generative AI use for regulated firms

Time’s almost up! There’s only one week left to request an invite to The AI Impact Tour on June 5th. Don’t miss out on this incredible opportunity to explore various methods for auditing AI models. Find out how you can attend here.

Enterprise application software maker SAP is deepening its relationship with Amazon Web Services through new AI-centric integrations. The first involves its AI Core platform: SAP is adding support for the Amazon Bedrock AI development service, creating opportunities for companies, especially those in regulated industries, to experiment with generative AI.

In addition, SAP is announcing that it’s doubling down on AWS chips to power its infrastructure. The company will upgrade its Graviton chips and plans to power its Business AI offering using AWS Trainium and Inferentia.

“AWS was the first cloud provider certified to support the SAP portfolio, and today, thousands of enterprise companies run SAP solutions on AWS to get the most out of their mission-critical applications,” Matt Garman, AWS’ incoming chief executive, says in a statement. “Now, AWS and SAP are making it faster and easier for companies to apply generative AI to their core business data to become more efficient, responsive, and sustainable.”

Bringing gen AI to more enterprise companies

Part of SAP’s Business Technology Platform, AI Core allows customers to manage their AI assets in a “standardized, scalable and hyperscaler-agnostic way.” The incorporation of Amazon Bedrock means those organizations can now access a litany of supported large language models, such as Amazon Titan, Anthropic’s Claude 3 Opus, Cohere’s family of Command R models, Mistral 7B and Mixtral 8x7B, and Meta’s Llama, Stable Diffusion, and AI21 Labs.

VB Event

June 5th: The AI Audit in NYC

Join us next week in NYC to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.

Request an invite

“We have a flexible and open architecture,” SAP’s chief AI officer, Dr. Philipp Herzig, tells VentureBeat in an interview. “We entertain a broad range of partnerships…we want to select the best technology for the right purpose.” He goes on to say that Amazon Bedrock possesses “some of the best models that our customers also demand,” offering a “broad range of models that we as developers in SAP, but also our customers, can choose from in order to build our own experiences, as well as their custom experiences.”

This means enterprise developers can leverage top-performing LLMs to improve key business processes built on SAP. For example, Amazon Bedrock could be used to enhance SAP’s S/4HANA Central Finance, streamlining the auditing process by matching bank statements to accounts.

“I think it’s fair to say that I certainly have not seen this level of excitement, investment, momentum and energy from customers around any technology, probably since the very earliest days of AWS itself,” Matt Wood, AWS’ vice president of AI products, remarks to VentureBeat. “Our enterprise customers are taking very deliberate steps. They’re moving generative AI into some advanced prototypes; they’re moving into production. And those production use cases are moving closer and closer to the core of their business.”

Though AWS gains another foothold for its AI development platform, the SAP integration is significant because of the types of companies that are using SAP’s enterprise resource planning applications, namely those in regulated sectors. It’s a space Woods admits may seem counterintuitive but is embracing gen AI faster than the average enterprise firm.

He attributes the compliance organizations such as the New York Stock Exchange, NASDAQ, Pfizer, Merck, the District of Columbia Department of Health, those in travel and insurance, life sciences and the public sector for driving the requirements “necessary to be successful with generative AI.” “They’ve got the data quality in place; they’ve got the data security in place; they have the right data approach to their data strategy. As a result, the use of generative AI is a much easier lift than with other organizations.”

Wood states that regulated firms are flush with large amounts of text data, which can be optimized with the help of AI, and have “a lot of use cases internally” where gen AI “can really make a dent in.” He believes these organizations have largely been left on the proverbial sidelines in their digital transformation journey, passed by the likes of gaming, media, transportation, hospitality, and travel. Gen AI may be viewed as a tool to “catch up with the Joneses” and overtake them.

SAP customers can use Amazon Bedrock within the company’s Business Technology Platform (BTP) or its managed cloud offering RISE with SAP.

Powering AI with AWS chips

Alongside its Amazon Bedrock tie-up, SAP is expanding its usage of Amazon chips to power its AI infrastructure. Specifically, its HANA Cloud, made for intelligent data applications, will be upgraded from Graviton3 to Graviton4 chips. In addition, SAP says it will use AWS Tranium and Inferentia chips to train and deploy future SAP Business AI products.

“Using Trainium and Inferentia2-based instances in a proof of concept, SAP engineers trained and fine-tuned generative AI LLMs in two days versus 23 days with comparable Amazon EC2 instances,” AWS boasts in a release. “By leveraging Trainium’s specialized architecture designed for efficient ML model training, SAP can accelerate the development process while maintaining high levels of accuracy and reliability.”

As Herzig explains, SAP is looking to dedicatedly designed chips to help it lower energy consumption, reduce costs and improve performance. He sees AWS’s Graviton4 as a “leap forward, in terms of price, performance, [and] for database and enterprise workloads.” The new chips will be used to support SAP’s BTP, Datasphere, Analytics Cloud and the company’s Cloud ALM solution.

Source link

About The Author

Scroll to Top