CloudWorld 23 - Steve Miranda on why the financial supply chain matters to healthcare, and what's next for generative AI

I'm about to jump into the Las Vegas breach, and get my Oracle CloudWorld 23 immersion experience. Before I hit the road, I did my quarterly online session with Oracle's EVP of Application Development Steve Miranda.

As expected, Miranda had a lot to say about generative AI at CloudWorld. But his early take on the show also surprised me - in a good way. 

For anyone who was under an enterprise rock, last year's CloudWorld included boldly-stated global healthcare goals from Larry Ellison - yes, bold even by Ellison standards (Last year, my diginomica colleague Derek du Preez took on those issues in Oracle has a vision for national and global healthcare databases to improve patient care).

Oracle's health care news - your cloud ERP is only as good as your industry know-how

Some of the goals Ellison outlined, including unification of healthcare databases globally, can't happen in a year, and, let's face it, maybe not in ten. But as Miranda looks back, he sees progress on multiple fronts (the Oracle Newsroom contains a slew of CloudWorld news updates). For this piece, I want to hone in on the financial supply chain news, and how it impacts healthcare specifically.  As Miranda told me, this is about horizontal cloud applications, tailored to the industry needs of customers - several of which should be joining Oracle on the keynote stage.

So we've got a brand new financial supply chain, a finance and supply chain for healthcare. We've got a set of updates around procurement, replenishment planning, recall management. As per normal, we build horizontal applications generally, but very tailored towards a very specific healthcare use case, which will integrate with Cerner, and/or integrate with the EMR of your choice. So we have a lot of Epic customers. In fact, on stage with me will be Providence Health, who happens to be an Epic customer, running in our supply chain.

No matter how you structure your horizontal/vertical ERP application stack, the days of ERP solving only generalized problems are over. Customers expect more. They want to solve healthcare workflow problems, not just manage payables, receivables, and the other generalized things ERP used to focus on. Miranda:

In addition to that, we've got the workforce scheduling. Our HCM integrates tightly with the EMR for really two purposes: one for better scheduling of actual caregiving, surgeries and doctor's appointments, especially in areas of high demand, like nursing scheduling, which seems to be perpetually in high demand. Also, we integrate so that the actuals from the EMR - what's actually happened in appointments, gets fed back for accuracy - and, frankly, a lot less work in payrolls.

Miranda says Oracle's Cerner customers informed these offerings:

What we found with our existing Cerner customers is: particularly these high demand workers are spending a lot of time entering/correcting their hours to get paid correctly. It's kind of a shame, especially in an area like healthcare where you've got a labor shortage, to have the caregivers spending time on that. So we announced that last year, and delivered it going forward.

The impact of mature Fusion Cloud customers - the underrated story to watch at CloudWorld 2023

Of course, you can't call a keynote complete without some generative AI talk, and we'll surely get that this week (My last discussion with Miranda took us in that direction, Can AI make supply chains smarter?). In fact, I am vigorously pursuing my own generative AI questions; Oracle won't be able to escape my own AI agenda.

But we shouldn't let the intense generative AI interest/questions obscure the significance of what Miranda is talking about here:

1. Tailoring vertical ERP to horizontal apps - moving away from generic approaches to cross-industry cloud ERP.
2. Building out nuanced industry cloud applications, driven not by an "industry cloud" promotional agenda, but by hands-on knowledge of customer requirements, which are then productized, rather than lost in one-off custom code misadventures.
3. Moving beyond cloud ERP go-lives into a more focused discussion of what customer success should look like, particularly in ERP. ERP is notoriously inward-facing; that's not going to cut it anymore.

I am hammering this because if we're not diligent, the influx of generative AI demos will derail the progress we are making talking about extracting long term ERP value in a customer-first world, and serving your own customers better. Yes, we do need to hear stories from generative AI beta customers as soon as possible, but in general, enterprise applications vendors spend way too much PR bandwidth talking up new customers and new deals.

So for those who stream these CloudWorld keynotes live or on replay, check the stories with this narrative in mind - how mature are these customers? Let's forget generative AI for five minutes - what value are customers extracting now? By now, Miranda knows my stump speeches. He anticipated this one:

I've given these talks: 'Fusion is coming. Fusion is here. Here's some early adopters. Hey, we've got customers going live. Hey, we're expanding the supply chain.'

Now, almost every conversation I'm having is, 'We've got it up and running. We've made the journey to the cloud. You guys are giving us these quarterly releases - GenAI, or B2B payments or something [Oracle has notable B2B payments news at this show]. How are we changing? What are the business metrics that we're measuring? And how are we improving?

Amen - these are the cloud ERP conversations vendors should be having, instead of stuffing inboxes with customer wins that haven't even gone live. Miranda:

During my last talk with JP Morgan, their CHRO Robin was all about, 'Hey, we're getting these features; how many are we adopting? Where's our dashboard for how we're using to take advantage of the system, and what are the benefits?'

There's a [consumer goods company] that's already adopted our B2B payments through JP Morgan. They implemented the connection to the bank in a day. Normally that takes a week. Testing the end-to-end connection between hosts is usually around a 12 week process; they save 92% of that time going forward.

Miranda wasn't finished: he told me to expect business value conversations this week.

Providence Health talks about the improvements on their close process - a new system. DP World's another one; they've expanded; they've reduced their close time, visibility into procurement they never had before. Hearst Corporation is using our early payment discount capability that he co-designed with us.

 I think you're going to have a lot of customers talk to you as far as, 'Fine, you went live; fine, you moved to the cloud. Forget all that. Let's talk real business impact.'

CloudWorld and generative AI - a quick preview

If Miranda is right; if those are the types of applications results we're going to hear about this week, it should be a winner of a show - whether or not generative AI is even mentioned. But, of course, it will be. Miranda explained Oracle's generative AI approach: "Over the next two to three releases, we'll be introducing about five dozen Gen AI use cases across the board."

By 'across the board,' Miranda is referring to everything from CX (sales and marketing) to back office finance to HCM. You can see CloudWorld's generative AI announcements live in Oracle's news section now; for more context on how Oracle is approaching generative AI for HCM, check my prior piece, Can generative AI make HR more productive - and less biased?

Miranda acknowledges that for this phase of generative AI, human-in-the-loop is part of a responsible AI design:

In HCM, things like job description, so in job posting, recruiting - you post a job that has certain requirements, years of service, etc. And again, using Gen AI to produce that. All the Gen AI is human reviewable at this point. As the technology evolves, that may or may not change, but it's all human reviewed. None of it is using customer information to train the model - and no PII information is ever sent to the model.

I think that sets the stage; now let's get some boots on the ground in Vegas, and see what we learn.