BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Over the last few years, AI and machine learning technologies have been rapidly adopted by software vendors across all fields, from IT to analytics to data management.
Oracle, one of the world's largest vendors in the enterprise IT space, is no exception. Last year, the vendor increased its AI tools with a major expansion to Oracle Adaptive Intelligent Apps, which sought to bring AI and machine learning capabilities to HR, finance, e-commerce and sales applications, among others.
By bringing AI to its various platform offerings and to the cloud, the Oracle adaptive intelligence software looks to deliver machine learning capabilities that are ready to go and easy to use across a variety of domains -- even though the vendor is playing catch-up in some regards with the technology, which it refers to more informally as Oracle AI Apps.
In this Q&A, Melissa Boxer, vice president of Adaptive Intelligent Applications at Oracle, discusses the evolution and advantages of Oracle AI Apps, as well the vendor's future plans for AI. The interview was conducted at The AI Summit in San Francisco, where Oracle representatives presented a keynote.
Review what Oracle presented at The AI Summit, and what's going on with Oracle AI Apps?
Melissa Boxer: Obviously, AI is the buzzword of the day. Everybody is trying to work on something, and there's always a project going on. Part of the issue is you can build whatever you want, but it takes time. You have to understand your use case, you need to understand the data, [and] you need the right platform tools. I think, sometimes, we lose track of the fact that people just want stuff to work and to drive themselves.
The reason machine learning and artificial intelligence is so powerful is the fact that you are, of course, leveraging compute and algorithms and data all together, but it's that [they] deliver what you want, when you need it and in the context of what you are working on.
What we're doing at Oracle, and what I talked a lot about at The AI Summit, is really looking across the enterprise and talking about how you can pragmatically infuse AI into enterprise applications and what some of the business benefits are of that.
At Oracle, we have a huge portfolio of AI applications, and there are more coming ... We're going to start to see our enterprise portfolio become more and more AI-enabled.
Where did the tools in Oracle Adaptive Intelligent Apps begin?
Boxer: We started with some simple stuff, some known stuff, like next-generation recommendations, really to drive our customer experience suite. So, [we were] looking at next-best product recommendations and offers.
In our Sales Cloud portfolio, we started to build use cases around having the machine predict win probability, so you have better forecast accuracy, and then your sales reps can really focus on what sales opportunities will close.
With that as an input, we looked at what the next-best action is for a rep to take within an opportunity to close the deal. The whole idea is that the machine is making recommendations. But with machine learning, you need some sort of feedback loop, so there is always human-in-the-loop feedback that is built into the system.
In HCM [Cloud], we've looked at candidate matching. But [we're] not just looking at traditional job descriptions and resumes, but trying to add another measurement to it, because when you're trying to hire, you have to think about what the 'right candidate' means. You want someone who is efficient and maybe someone who will stay around for a while.
So, we've started to look at performance data as a proxy, or looking at similar profiles of similar candidates who are employed and are higher performers, or looking at the lifecycle of a candidate career. We really try to look at these use cases in another dimension.
Why would someone choose to use Oracle adaptive intelligence tools, rather than build their own system or choose one that is more specifically tailored to their needs?
Melissa Boxervice president of Adaptive Intelligent Applications at Oracle
Boxer: I think if you try to build something from scratch, you're really looking at longtime horizons; you're looking at expensive resources, like data scientists; and you're looking at a lot of trouble harnessing data -- even if you have great platforms, great collaboration tools.
I think what differentiates us on that dimension is it works, and we already have the built-in domain expertise in the applications. You can't just build models for the sake of building models. It has to apply to a use case; you have to work with a domain expert that can help guide you as to what works and doesn't work. You can have the most beautiful ROIC [return on invested capital] curves, but are they actually driving results in your business?
With Oracle's ability to harness third-party data signals, harness data from across the enterprise and incorporate that into the machine-driven outcome in real time, that's really important. That's what separates us from building your own or from the competition.
As data becomes easier to access and easier to share, do you see a change in the Oracle AI Apps model?
Boxer: I don't think I foresee that right now. I think there's a lot of work to be done to even harness the data out there.
It's one thing to access commoditized data, but it's really about the proprietary signals, taking data and developing the right type of features and signals and investing in data lakes for these signals. That's a really hard thing to do, and I don't really see a lot of companies adopting that anytime soon. So, I think we will always have this data advantage.
What do you foresee for the future for Oracle adaptive intelligence, then?
Boxer: It's all about the new interfaces and new UX [user experience] paradigms that will change the way we work today, and that's all driven by these new model-driven applications. It's really going to change the way we work in a good way, I think, so we can leverage our skill sets to do things that should be manual for other areas where machine learning maybe isn't appropriate.
Not everything should be machine-learning-driven, but there is a large portion of enterprise apps that can be driven by that.
Editor's note: This interview has been edited for brevity and clarity.