SAP adds generative AI assistant to enterprise software portfolio

SAP has this week joined the generative AI digital assistant party with the announcement of Joule, its version of a natural-language, AI copilot that will be an integrated part of its enterprise software tools. 

According to SAP, Joule will help people get work done faster and lead to better business outcomes. 

 Users will be able to ask a question or frame a problem in plain language and receive a response drawn from business data across the SAP portfolio and third-party sources. SAP gave the example of a manufacturer asking Joule for help understanding sales performance better: Joule would be able to identify underperforming regions, link to other data sets that reveal a supply chain issue, and automatically connect to the supply chain system to offer potential fixes. In an HR scenario, it would help write unbiased job descriptions and generate relevant interview questions.

It’s by combining data from different sources that enables Joule to support these scenarios, Thomas Saueressig, Member of the Executive Board of SAP SE and SAP Product Engineering Lead, explained during a media briefing for the launch of Joule:

We need to have the real business context, the real business data. That's what we bring together with the real-time transactional data of the customer itself, which we can access, and SAP foundational model, which by default is different because only we have this wealth of data, and the LLMs, which we bring together in the reply. That's how we provide the best outcome possible for the user.

While Joule will pull in data from SAP and third-party sources for its responses to users, SAP will not be merging customer data with third parties to train Large Language Models (LLMs). Bharat Sandhu, SVP for AI and Application Development Platform at SAP, said:

One of the biggest things we have to make sure is we respect customers’ data privacy. So we do not use customer's data to train large language models that's available to other customers at all, and we will not be retraining an external large model with customer data.

The generative AI tech will be embedded throughout SAP’s cloud enterprise portfolio, from HR to finance, supply chain, procurement and customer experience, as well as into SAP Business Technology Platform – but there will be no availability for on-premise applications. 

  Joule will be available with SAP SuccessFactors solutions and the SAP Start site later this year, and with SAP S/4HANA Cloud public edition early next year. SAP Customer Experience and SAP Ariba solutions along with SAP Business Technology Platform will follow, with several other updates across the SAP portfolio to be announced at the company’s events in October and November.

Building 

SAP was keen to point out that Joule builds on its existing AI offerings, and that more than 26,000 SAP cloud customers already have access to SAP Business AI across multiple scenarios and partner solutions. 

The vendor has potential for success with its new generative AI play based on its wide scope of product and customer base – as CEO Christian Klein notes, the firm has almost 300 million global enterprise users working regularly with its cloud solutions. Despite the broad user base, Joule is taking a narrow rather than general approach, focusing on specific functions and job roles. 

This type of AI assistant is similar to what rival vendors have already launched: Salesforce is running a pilot of its Einstein Copilot this Fall; Freshworks’ Freddy Copilot is currently in use by 390 beta customers; and Microsoft is integrating its AI assistant, Copilot, with its employee engagement product, Viva.

During the press launch event, Saueressig noted that SAP started working on AI years ago, adding that the firm already has more than 130 AI cases in production with customers, and more than 360 partner apps with AI available in the SAP app store. But Joule is a major leap forward in how people will work in the future across all of the firm’s portfolio, Saueressig adds: 

You can ask all of the questions according to business, and you'll get the right insights, recommendations and answers, which we bring together with the wealth of business data, which we know. 

It will be easier and more democratized to access our solutions. It helps that you can guide users differently to the respective situations and solutions and bring all the context together. By default, this will improve the productivity of the user so that you can complete the tasks in minutes rather than the hours or days before.

However, to achieve these benefits, it’s no good working in isolation. Saueressig pointed out:

That's the reason why you also have seen the announcements throughout the course of the year that we team up with industry leaders to make this great innovation together. We already announced strategic investments in Aleph Alpha, Cohere and Anthropic, but also our strategic work we do with Google, Microsoft, and IBM. In this topic, we need to talk about an ecosystem.

SAP didn’t unveil specific pricing for the new technology. However, Julia White, SAP Chief Marketing and Solutions Officer, disclosed that the generative AI use cases of the applications that Joule calls upon will have a unique business model – AKA pricing - associated with them based on value provided. She added that SAP will be sharing more details about that in association with each of the applications Joule is available for.

My take

Joule isn’t generally available to customers yet, so it remains to be seen how well the technology works and how easy and useful it proves for employees. It will also be interesting to see the more detailed pricing for users, and how complicated that is to manage for customers.