Thanks to SAP Joule, the power of generative AI will be able to surge into your IS. A robust, secure, multi-LLM cloud offering, which SAP RISE users will soon be able to deploy in their SAP solutions, or implement within their specific ones, through SAP BTP.
At the end of September, SAP presented Joule, a natural-language searchable co-pilot based on generative artificial intelligence techniques. This tool will gradually be integrated into all SAP cloud solutions. It will first make its debut on the SAP SuccessFactors HR platform, before being integrated into the SAP S/4HANA ERP in public cloud mode.
SAP Joule is not a direct replacement for the SAP CoPilot digital assistant (based on the Recast.AI offering). It is based on new artificial intelligence technologies from SAP’s extensive partner ecosystem. The chosen approach is to cover all existing solutions, enabling us to offer the very best in generative AI.
A multi-LLM strategy
SAP Joule is intended as a gateway to the various LLMs (Large Language Models) offered by its partners. Technically, users will be able to switch from one offering to another at will, while retaining the same proxy for their queries: SAP Joule.
With SAP software packages, it’s up to the editor to choose the LLM best suited to a given use case, as not all offer the same reasoning capabilities or the same cost. The aim here is to maximize users’ ROI by always implementing the most suitable offering, in terms of both power and cost.
This multi-LLM approach will also make it possible to adapt seamlessly to regulatory constraints, as the deployment of an SAP solution in Dubai or Beijing may lead to certain technological choices better suited to local regulations or culture.
And, if necessary, SAP can combine generative AI with other technologies. Just Ask from SAP Analytics Cloud can be used to add an analytical component to the intelligent assistant’s responses.
Technology under control
Generative AI promises to revolutionize the world of conversational agents. However, we must beware of the problem of hallucination: in the absence of sufficient reliable data, the answers given by the chatbot may lose their relevance, even if the explanation given seems plausible.
To avoid this, it is essential to feed the AI with business data, whether from third-party data sets or contextualized data supplied by the company. Retrieval Augmented Generation (RAG) will significantly improve the quality and reliability of the answers provided by generative AI.
Note that for each of the LLMs used, strict contractual rules have been laid down to ensure that data originating from the company is not used as training data for generative AI.
More generally, SAP is committed to making AI a safe and reliable technology (” Relevant, Reliable, Responsible “). To this end, an ethics committee was set up back in 2016 to define the rules for using this technology. These rules have been integrated into the SAP Product Standards, which apply to all developments at SAP. Before a solution can be delivered, it must comply with these principles.
A fully accessible technology for SAP BTP
Generative AI is set to be deployed in full force in companies. Those who adopt it will benefit from productivity and efficiency gains that will enable them to score points against the competition. It is therefore highly advisable to evaluate the potential benefits that this technology (and others, such as the cloud) can bring to your organization, without delay.
SAP Joule is based on the SAP Business Technology Platform (SAP BTP), which means it can be used in all areas of a company’s information system. Not just ERP and HRIS, of course, but also customized services and applications, thanks to the ability to call on generative AI from any specific code deployed on the SAP BTP. This enables companies to develop innovative use cases based on this emerging technology. Most uses of generative AI have yet to be imagined…
Applium keeps a close eye on Joule’s potential
Joule is a relatively young co-driver in the SAP ecosystem. Designed to improve user experience and operational efficiency, its forthcoming integration into SAP S/4HANA Cloud Public remains a decisive point. As soon as it is made available by SAP, Applium’s Cloud and BTP teams will be quick to implement it on demonstration environments.
We’re convinced that Joule offers significant potential for a more efficient ERP, benefiting from the latest technologies. However, its current maturity does not yet allow us to make a decision. The quality of integration with the ERP system and the support we provide to ensure its long-term success will be decisive factors for the future. As soon as we have implemented SAP Joule in our environments and are using it, we’ll let you know in a future newsletter about our initial experiences with this promising tool.
