This is a superb conversation about the trends and directions in which AI is evolving, especially in business applications.
Sramana Mitra: Let’s start by introducing our audience to SAP’s artificial intelligence activities. What trends are you seeing? What trends are you leading?
Mike Flannagan: The most significant thing that we’re seeing is that people don’t expect to interact with artificial intelligence, machine learning, advanced analytics as a side card to their core business processes. When we think about how to apply machine learning, we’re thinking about how to apply it in the context of existing workflows and how to enrich business processes that are used commonly rather than letting it be something that sits off to the side.
I’ll give you a couple of use case examples of where SAP is doing that. This is where we see the power of machine learning and artificial intelligence. Fieldglass, which is an SAP application that is concerned with workforce management, has a huge amount of data across our entire customer base about which job codes are easy or difficult to hire for and in which geographic markets, and what contract rates contingent workers would typically receive in each of those markets.
If I’m trying to hire, for example, a data scientist, I may have the option of hiring that data scientist in New York, Los Angeles, or Chicago. What I really want to understand is where’s the best location to hire that person so that I can get the very best talent at the very best price.
SAP has taken years of data from Fieldglass customers, aggregated and anonymized that data, and is now providing, what we call, SAP Fieldglass live insights back to customers inside the applications. It sits in the process of the workflow of their opening a new requisition for contingent workers. That gives them guidance on the availability of talent, the price that they would expect to pay for that talent, by the market that’s relevant for them. The power there is about being able to take aggregated anonymized data and apply machine learning to it to provide the ability back to a customer to get information in context of the workflow.
Sramana Mitra: In this particular use case, you are training your algorithm with data across multiple customers. How do the customers feel about that? Is this a discussion that customers are having with you?
Mike Flannagan: Of course. They’re quite excited about us being able to provide these kinds of insights. Every customer has questions and concerns any time that you talk about aggregated and anonymized data. In this case, we’ve been very successful in working with our customers to assure them of the security and privacy of all of that data. Once you set aside those security and privacy concerns, customers are really excited about getting these kinds of insights that help them make better decisions.
There’s a lot of complexity in the way that data is structured from company to company. The ability to apply machine learning to harmonize job titles to industry standard job codes is something that would take a person an inordinate amount of time to do. By the time you are able to get intelligence from the data, it would not be worth the effort that you have to put in. That’s really where the power of machine learning comes in.
Sramana Mitra: The algorithm is also trained to provide recommendations on how to adjust the job codes and so forth based on this kind of harmonizing.
Mike Flannagan: That’s right. I might call it Data Analyst. You might call it Business Analyst. When you look underneath that, the system is intelligent enough to understand that it’s the same set of five skills. The job title is not the relevant piece. It’s the skills.