Steven Miranda of Oracle: De-siloing workflow and data is front and center in the Agentic Era

Paul Greenberg and I have been doing our CRM Playaz show for 17 years as of next month.  And for about that same amount of time we've both been going to Oracle's big annual user event AI World (before that it was Cloud World and even before that Open World).  And every time we attended the event we would be fortunate enough to be part of the analyst Q&A with Steve Miranda, their Executive Vice President, Applications Development.

Those gaggles (that's what we call those Q&As) last about 20-30 minutes. So if you add 15 or so years of 30 minute Q&As that adds up to 7.5 hours of gaggling. And while that's collectively a pretty fair amount time, it dawned on me that I hadn't really had spoken with Steve outside of the annual Q&As; and Paul was in the same boat, give or take a call or two.  So we decided we needed to get Steve on our turf - aka our show - so we could really get to no him a bit more. And even though it took a bit longer than expected, we were able to have Steve join us for the last CRM Playaz show of 2025.

We covered a lot of ground during our non-gaggle time with Steve and it was great.  We learned more in the 30 minutes he was on with us than in the 15 years of gaggles. And in this clip Steve explains that while technical integration is often seen as the primary hurdle, the real challenges lie in workflow and data; and how the focus on agentic AI shifts the emphasis towards a more comprehensive approach to achieving desired business results that weren't achievable before

Below is an edited transcript of the clip.  Click here to see the full episode. 

Edited Transcript

Brent Leary: I know we’ve had a couple of shows lately around the whole topic of what is CRM in 2026. And thinking back to I think it was when you bought RightNow, that was kind of like the foundational move to where you are today and all the different acquisitions. You guys made a ton of acquisitions, like within a span of 2 years that brought a lot of those perspectives and that kind of data into the mix.

And now when you’re looking at what’s going on with all the kind of - I don’t want to say hype, but the agentic focus that everybody is on. And what I think - and our buddy Michael Wu even talked about - a lot of folks are really hoping this is the era that we actually get to outcomes as opposed to focusing on the individual applications and being able to use agents not just from one company, but the - kind of - I don’t want to say Lord of the Rings or Lord of the Agents where you have one agent to rule them all or something like that.

But a lot of people are hoping that there’s going to be some good partnerships and integrations that allow them to actually focus on the outcome and not specific applications to hopefully get to that outcome. How do we make sure we get there this time? Because it feels like we’ve had opportunities because I think at the enterprise level most do not have just one vendor. They do have a number of vendors that they work through to try to get the outcomes they want, but they haven’t really gotten the outcomes they want. And it feels like this focus on agentic and agents, there’s another opportunity to have that happen. What do we need to do to get there?

Steve Miranda: So, let me see if this answers your question. This is the way I look at it. For years, you know, although we try to build a suite, clearly most - if the vast, vast majority of our customers use multiple different systems. And so they always ask the integration question. And I’d be careful to talk to a customer.

To me, an integration often gets thought of as a technical problem. It’s usually not a technical problem. It’s usually a workflow - like I can give you a bunch of ETL systems technology that Oracle sells or others sell that will connect Salesforce to Oracle or Salesforce to SAP or whatever combination. And people have done it.

So technically it’s not a - it’s a workflow and it’s a data problem. And I think what’s different this time - and as the name would imply, agentic apps - it is an agentic application with some form of workflow - or that name will get renamed - and some form of an agent that’s governing these workflows.

So I think from the start of the problem, what the technology has lent itself to is a better part of the solution. Whereas in the past, the focus of the problem has been on ETL tools or integration technology. It’s been on data formatting, back and forth, but it didn't talk about the workflows or the data.

That’s part one. The second part I would say is, if you look at what a lot of people are getting success at AI today, it’s things like natural language into queries. So now you can just talk to your system or give it a - you don't need to run a report, you don't need to run a query, you can just - just like a Google or just like ChatGPT, you’re doing natural language to a select process.

Well, what you find then is very quickly the most interesting questions you can ask really span these domains. Because once you start putting English language and I get my first level data, you get very interesting questions that span. And that’s where the pressure is on integrating the data part.

So I think both pieces that I’ve traditionally felt that people don’t focus on in integrations - the workflow and the data - it’s much more in front and center in terms of where the technology is going. So that’s - if there’s hope, I think it’s going to be better.

And I don’t think it’s - you know, I think Dr. Wu said it correctly in preceding me, MCP is a leading standard that will evolve. There’s AI standards. I don’t ever think it’s been about companies not wanting to share or agree on technical standards. I just think that it’s been lacking in terms of the comprehensiveness of that particular problem.