About this transcript: This is a full AI-generated transcript of Oracle AI World Tour London 2026: Opening Keynote from Oracle, published June 10, 2026. The transcript contains 13,835 words with timestamps and was generated using Whisper AI.
"Let's go. We reduce the time almost five minutes per patient. We're giving you back almost two hours to your day. We literally had providers that were tearing up. I can be home to see my kids' games. I can be home to do homework. We have navigated through a pandemic, a supply chain crisis across..."
[00:00:00] Let's go.
[00:00:30] We reduce the time almost five minutes per patient.
[00:00:39] We're giving you back almost two hours to your day.
[00:00:41] We literally had providers that were tearing up.
[00:00:45] I can be home to see my kids' games.
[00:00:47] I can be home to do homework.
[00:00:49] We have navigated through a pandemic, a supply chain crisis across the world,
[00:00:53] trade wars, and we have even navigated real wars that are taking place.
[00:00:56] The speed we're moving at innovation, it surprises us.
[00:01:00] We have more than doubled in size.
[00:01:03] We've become the first in-market multimodal AI that can understand videos.
[00:01:07] Oracle really gave us exactly the product that we needed.
[00:01:11] What I cannot do with any other application is get AI so close to data.
[00:01:16] Our data is in Oracle. Our metadata is in Oracle. Our tools for Oracle.
[00:01:20] Oracle Climate Instructure is a great platform to do AI tools.
[00:01:23] The relationship between us and Oracle, it was beyond the commercial aspects.
[00:01:27] They treat the investment as if it's their own.
[00:01:30] The tech offering is great. The people are better.
[00:01:32] We are truly a partnership. When we win, they win.
[00:01:41] Please welcome to the stage, Siobhan Wilson.
[00:01:49] Good morning. Good morning, everyone.
[00:01:52] And a very, very warm welcome to AI World Tour London 2026.
[00:01:56] It's great to see you all here.
[00:01:58] All of our customers, partners, media, and analysts all joining us here today.
[00:02:04] I know I speak for everyone at Oracle in saying we can't wait to talk to you all about how Oracle AI changes everything.
[00:02:13] It's been a year since we were all together, and I think we can all attest to the speed at which AI innovation has accelerated in the past 12 months.
[00:02:24] Here at Oracle in the UK, we've announced new innovation in healthcare with the launch of Oracle Health Clinical AI agent, Clinic Note.
[00:02:34] We've added new genitive AI capabilities into our UK sovereign cloud to support the UK government's vision of AI.
[00:02:44] And recently, we renewed our partnership with Oracle Red Bull Racing, bringing AI to the racetrack.
[00:02:51] I must flag to all of you that the simulator and the replica car are here.
[00:02:57] So I would really recommend that you all go and see that and see the amazing work that the partnership is bringing.
[00:03:05] But it's the advantages that AI is bringing to every aspect of business that we are going to really focus on throughout the day.
[00:03:13] We're going to be hearing from Mike Cecilia this morning's keynote, our CEO, about how new AI capabilities are taking shape fast and how customers can use them.
[00:03:27] We're also here from Oracle leaders across Oracle cloud infrastructure, cloud applications, AI database, and the new AI data platform,
[00:03:37] who are all here having exciting announcements to make on stage here in London.
[00:03:42] And they will explain how Oracle is helping its customers harness AI to transform the way business gets done today.
[00:03:50] On top of this, we will also be hearing from the NHS SBS, which is transforming finance function of the National Health Service,
[00:04:01] while also preparing for what AI will enable next.
[00:04:04] We'll also be joined by Zoom on stage to discuss how they are harnessing the power of AI to help us all keep connected.
[00:04:14] After the keynote, there is a whole host of general and learning sessions, product showcases, hands-on labs, and exciting demos in the hub.
[00:04:25] All of these are focused on helping you understand how AI can help you drive business change and translate your AI vision into operational results.
[00:04:37] With all that said, it is now my honor to introduce to you Oracle CEO, Mike Cecilia.
[00:04:45] Mike, thank you for joining us today.
[00:04:47] It's fantastic to have you here, and a pleasure to hand the stage over to you.
[00:04:52] Thank you, Mike.
[00:04:52] Please welcome to the stage, Mike Cecilia.
[00:04:59] Good morning, and thank you for being here.
[00:05:02] For decades, customers have come to Oracle when their business challenges were quite significant.
[00:05:08] When the existing way of doing things simply wasn't going to scale, these weren't edge cases, these weren't experiments.
[00:05:16] They were core systems containing mission-critical data.
[00:05:20] The parts of the business where failure simply is not an option.
[00:05:25] The solutions that we built at Oracle didn't happen in isolation.
[00:05:30] They happened because customers like you trusted us with real problems.
[00:05:35] You asked us to partner, to innovate, and you expected the technology that worked in the real world
[00:05:41] when the competition threatened, when regulatory pressures loomed, when economic wins grew turbulent,
[00:05:48] and at global scale with no room for shortcuts.
[00:05:54] That imperative shaped everything we built first.
[00:05:59] Resilient infrastructure and trusted data.
[00:06:02] Systems that had to be secure and speedy enough to run the business every single day.
[00:06:09] And before anything could become predictive or even adaptive, it had to be dependable.
[00:06:15] And once it was dependable, it could then become intelligent.
[00:06:20] So some of what you'll see today is powered by AI, and some of it is the groundwork that makes AI possible.
[00:06:29] But all of it is delivering real impact now.
[00:06:36] And that's why we're here.
[00:06:37] To show you what's already making a difference, how customers are already doing this today,
[00:06:43] and how that progress is already setting the stage for what comes next.
[00:06:49] Because just like your continued success drives us to build more capable systems,
[00:06:55] your own customers are driving you to operate at an entirely different level.
[00:07:00] Expectations in the age of AI haven't simply risen.
[00:07:04] In fact, they've reset entirely.
[00:07:07] And remarkable things that felt out of touch just a few years ago, a few months ago,
[00:07:12] and in some cases just a few weeks ago, are now the new normal.
[00:07:18] We have customers right here in the UK already doing that.
[00:07:22] You know, for example, a tech company training multimodal AI 15 times faster at half the cost on the Oracle Cloud.
[00:07:29] This is speeding time to market for their phone-based customer service.
[00:07:34] An engineering and construction company cutting time to hire from 45 days down to 21 days with AI embedded in Oracle Fusion HCM.
[00:07:44] And a government agency providing users with on-demand access to data and achieving well over 99% uptime with Oracle's analytics platform.
[00:07:54] All of that's running on Oracle.
[00:07:56] And as powerful as those examples are, they're just the beginning of what's possible in the age of AI.
[00:08:02] So the real question then is, well, what comes next?
[00:08:06] Well, with Oracle, we'd like you to decide.
[00:08:09] We're not here today to define your future with AI.
[00:08:13] We're here to provide you with the tools that you need to shape it.
[00:08:19] We're here to take care of what's weighing your business down so you can focus on what lifts it up.
[00:08:25] What you do with it comes from only what you can imagine.
[00:08:28] And maybe that'll be sparked by some of the conversations that we have in this very room today.
[00:08:33] We've got some ideas, but more importantly, we've got the tools to help you turn your own ideas into reality.
[00:08:41] I like to think about aviation shift from propellers to jet engines as a comparison.
[00:08:47] When that happened, airlines didn't just change their mission.
[00:08:52] They're still moving people and goods.
[00:08:54] But what changed was capability.
[00:08:57] Jet engines expanded what was possible.
[00:08:59] Faster routes, longer distances, entirely new markets.
[00:09:04] Because the underlying technology could finally keep up with rising expectations.
[00:09:09] And that's exactly what AI is doing and will do for the enterprise.
[00:09:15] It's not here to replace expertise.
[00:09:18] It's here to elevate it.
[00:09:19] By taking on the invisible complexity underneath, like analysis, coordination, and prediction,
[00:09:25] AI can move human judgment way up the stack.
[00:09:29] Leaders then spend less time reacting and more time deciding.
[00:09:33] Teams stop managing processes and start shaping outcomes.
[00:09:37] And this is why AI feels different right now.
[00:09:41] We're not just now experimenting for novelty.
[00:09:44] We're using it to solve real constraints.
[00:09:47] Whether you're hiring faster, seeing financial risk earlier, preventing supply chain disruptions,
[00:09:54] or resolving issues in real time.
[00:09:56] And that transformation isn't just a technology transformation.
[00:10:00] It's a human transformation.
[00:10:02] It transforms how we all work.
[00:10:04] AI changes how work gets done so people can focus on strategy instead of administration.
[00:10:11] Insight and maybe even foresight instead of reconciliation and relationships instead of doing things like rooting tickets.
[00:10:20] In other words, AI doesn't just make enterprises faster.
[00:10:27] It gives them the altitude to think bigger.
[00:10:31] And here's what we mean.
[00:10:32] When people talk about AI today, in the early days of AI, they often focus on very small familiar examples.
[00:10:39] Things like co-pilots are embedded into workflows or incremental productivity.
[00:10:43] And certainly they're useful.
[00:10:44] And certainly they present a real tangible game.
[00:10:49] But that is, to me, just like judging jet engines by whether they simply made the takeoff smoother.
[00:10:58] The real impact from AI is coming from thinking about the entire system and rethinking about entirely new capabilities.
[00:11:06] Here's what it may look like in practice in finance, for example.
[00:11:09] We've got a global enterprise planning 150 million euro expansion into Europe.
[00:11:14] They need to lean on their bank to identify how they can leverage existing cash flows and where they may need credit.
[00:11:20] The bank systems, in this case powered by Oracle AI agents across cash flow, balance inquiry, and document intelligence, generate a fast unified forecast, which identifies a need for a 50 million euro line of credit.
[00:11:35] And with quick analysis, the corporation is able to apply for the credit line.
[00:11:39] The bank systems are able to confirm the headroom, pull in a risk review, and of course, keep humans in the loop, elevating their decision making up the stack for final approval.
[00:11:51] This results in a very quick AI-powered credit proposal, which is ready for speedy execution.
[00:11:57] The enterprise then expands faster.
[00:11:59] With real-time visibility into liquidity, suppliers are on board for five, and suppliers are on boarded for supply chain.
[00:12:05] Financing without delay.
[00:12:08] Finance teams can stay focused on growth, not reconciliation.
[00:12:11] And at the bank, credit and onboarding move quickly without sacrificing control.
[00:12:16] This frees up, of course, the risk teams to focus on exposure, stress testing, portfolio health.
[00:12:21] What once took months, maybe even a year, can now happen in hours on both sides of the relationships.
[00:12:27] We have a company with a very large dealer network that needs to address rising dealer churn, and that's often difficult to detect early.
[00:12:37] There may be signals across many systems and formats, things like sales transactions, quote service history, product mix, engagement data.
[00:12:45] But it makes it very hard to see deteriorating relationships, given the sheer number of signals, before real revenue is actually at risk.
[00:12:53] But with applied AI, the company now treats this fragmented information in a single, unified view.
[00:12:59] A supervisor, dealer churn agent, orchestrates worker agents across sales processes, quoting processes, and service data.
[00:13:08] This correlates historical patterns, identifies early churn signals, scores at-risk relationships, and recommends targeted interventions.
[00:13:17] Commercial teams then can act earlier with focused outreach, which reduces attrition and protects revenue.
[00:13:26] In this new world, agents are talking to agents.
[00:13:30] Humans, of course, can stay firmly in control, but no longer need to do all of the glue work.
[00:13:36] And this is what enterprise operations can look like when systems are built for AI scale capability.
[00:13:43] Just as jet engines open routes that could never be reached before, AI can take you to new places,
[00:13:50] freeing you up for entirely new ways of operating across banking, healthcare, retail, and beyond.
[00:13:56] Freeing you up to make visionary changes beyond what was imagined even months ago in every industry.
[00:14:02] With AI, for example, we're helping move patients from biopsy to therapy in hours instead of weeks.
[00:14:09] We're helping speed up the journey from lab discovery to scale production, making drug development faster and safer.
[00:14:17] We're helping farmers run fields and greenhouses that continuously optimize themselves, boosting yields, using lost water, and improving margins.
[00:14:30] That is the art of the possible when AI isn't bolted on as an afterthought, but it's built in everywhere and built in everywhere by design.
[00:14:41] Jet engines didn't work because they were bolted onto old airplanes.
[00:14:46] Aviation had to rethink the entire system.
[00:14:49] Materials, maintenance, runways, air traffic control.
[00:14:53] And that's what we're doing with AI.
[00:14:55] We've done the re-engineering for you.
[00:14:58] So these AI capabilities should be as easy as possible to turn on and to turn into meaningful progress.
[00:15:04] And of course, it all starts with data, your data, because the best AI is going to be fueled by your trusted data.
[00:15:14] And at Oracle, we've been the custodian of the world's most important data for nearly 50 years.
[00:15:21] We're providing access to AI models of your choice in a cloud tailored to your needs, using data residing in any cloud, in any format.
[00:15:30] And later, we'll show how this all comes together in our AI data platform.
[00:15:35] With infrastructure designed to handle massive AI workloads, delivering performance, security, and scale by design, it's no secret that at Oracle, we are building the world's largest AI cloud.
[00:15:52] And we think that this will all show up where the work actually happens.
[00:15:58] And that is surfacing the AI directly in the applications and workflows that you're already living in, right at your fingertips.
[00:16:05] So whether you're in finance, HR, manufacturing, retail, healthcare, financial services, and beyond, when you put it all together, you can enable AI that isn't just incremental.
[00:16:16] It's woven into workflows, it's tidying up tasks, it's offering a built-in automation step function.
[00:16:23] We think that's very valuable, but there's even more.
[00:16:28] True transformation requires data that's ready for AI.
[00:16:34] True transformation requires making the use of AI very easy.
[00:16:40] True transformation requires massive, reliable compute operating at scale
[00:16:45] that most enterprises have never had access to.
[00:16:50] So today, we are announcing new AI capabilities across our stack.
[00:16:56] AI built into the database so that it can work on live business data securely.
[00:17:03] 22 new AI-driven fusion application experiences that help teams get to outcomes faster with the right approvals and tracking.
[00:17:13] And a new OCI platform that makes it easier to build, deploy, and run AI at scale.
[00:17:20] In a moment, I'm going to bring up our product leaders to tell us more about these announcements and how those foundations of trust, choice, and scale make the entire stack AI ready for customers.
[00:17:33] So you can simply then focus on leveraging and using AI, not trying to figure out how to make all of it work.
[00:17:41] We hope that this will help you start reaching places that you've never been before.
[00:17:46] And that's the difference when your technology partner didn't just arrive for the AI moment, but in fact, has been preparing for it for decades.
[00:17:56] So with that, let's bring up the product engineering team leaders and hear what's new.
[00:18:01] Please welcome to the stage, Juan Loeza, Nathan Thomas, Steve Miranda, and TK Anand.
[00:18:21] All right, thanks for being here.
[00:18:24] Steve, I'm going to start with you.
[00:18:26] As I just mentioned, 22 new fusion experiences that we're announcing.
[00:18:32] With those new experiences, what's the shift in how work gets done, and what impact should our customers expect to see first?
[00:18:40] So thanks, Mike.
[00:18:41] So what we're announcing today is 22 brand new agentic applications.
[00:18:44] So they're agentic applications across the fusion suite of applications from CX, supply chain, HCM, financials.
[00:18:52] And the reason why they're brand new is it's taking on what we've built and what you talked about, OCI providing us all of the large language models, the best ones in the world, the database having AI features.
[00:19:03] So we've taken all that.
[00:19:04] We've built agents, as we've discussed in the past.
[00:19:06] But now with these agentic applications, we allow our end users to really interact with the AI, not just for process automation, but actually for decision making.
[00:19:15] So in today's world, if you're in a supply chain, managed supply chain, you have to deal with rising or certainly changing fuel costs.
[00:19:22] You're dealing with interesting transportation challenges, perhaps.
[00:19:24] You're dealing with supply chain shortages around the world.
[00:19:27] And, of course, you have increased customer demand.
[00:19:30] So it's beyond just entering transactions into your inventory system, your supply chain.
[00:19:35] You actually have to do some planning as a decision.
[00:19:36] So what an agentic app allows our customers to do is interact with the system, give real-time business objectives.
[00:19:43] And because the agents are embedded into Fusion and because we're aware of all the data that flows through in your transactions, we can present to the user some recommendations, some choices.
[00:19:55] The agent, rather, can present that.
[00:19:56] And so the agentic app not only allows a customer to interact with the system via business objectives, but then we give them scenarios that they can execute.
[00:20:06] And then, of course, the agents that we built behind the scenes go ahead and execute on their behalf.
[00:20:10] So we've done two things, really automate completely the system of record, Fusion applications, and change it from a system of record into a system of innovation using the new agentic applications.
[00:20:21] Where would you expect to see the fastest wins?
[00:20:24] For sure, I think supply chain and in finance and then perhaps service as well.
[00:20:28] So really across the board, and when I'm talking to customers, the common question is, well, where do I start?
[00:20:34] And because of the breadth we have, not only in the suite of applications, but where the new agentic apps are, it's really where your pain point is.
[00:20:40] So if you have that supply chain challenge that I spoke of, it's there.
[00:20:44] If you have a lot of sort of spend in the service area, you can deploy it there.
[00:20:50] In finance, there's always room for optimization, optimizing your cash position or, you know, what you do in terms of economic impact there.
[00:20:58] So wherever your area of most pain point is, we have the agentic app to help you there.
[00:21:03] So let's shift gears then to the database layer.
[00:21:07] Juan, what are we announcing today in the database to make agentic AI work, specifically with customers' private business data, both securely and, of course, at scale?
[00:21:19] Yeah, so we have about a dozen announcements today, and many of them are agent-related.
[00:21:25] I used to say when we talked to customers, it's all AI all the time.
[00:21:30] Now it's really all agents all the time.
[00:21:32] That's the new form of AI.
[00:21:34] And one thing, when you hear agents, you're going to hear everyone talking about different kinds of agents.
[00:21:40] And so the way I think about it is agents aren't a thing.
[00:21:43] They're really, you really have to specialize around what kind of agents you want.
[00:21:50] So Steve is building agents around his applications that understand the applications and get very effective by knowing the applications.
[00:21:58] What we're doing in the database team is we're building agents for data-centric apps.
[00:22:04] That know about the database, know how the data is structured, know how to combine data to produce results.
[00:22:11] So one of the big things we're announcing is something we call a private agent factory, which is a no-code environment for easily building agents, even if you don't know a whole lot about database.
[00:22:21] So that's a big thing.
[00:22:22] We're also doing something called a unified agentic memory, because agents need to remember things while they're doing things and between different operations.
[00:22:31] So we're making that super efficient and super simple, and lots of other things we're doing with agents also.
[00:22:37] So you're obviously putting AI very, very close to the data.
[00:22:40] How should customers think about AI that's very close to the data versus maybe AI that's off to the side?
[00:22:49] You know, what are things that they should think about?
[00:22:50] Yeah, it's a good question.
[00:22:51] So again, the way to think about this is the data has the ground truth of the enterprise.
[00:23:00] So by combining these two things, putting the AI essentially in or right next to the database, you can combine business data with AI and make the whole thing very effective.
[00:23:22] And that's the key to getting real business benefits is it has to combine both business data and a lot of the AI operations, things like natural language, things like search across documents and all that.
[00:23:34] Yeah, as you say, you know, I think it's been our core mantra to build it in, not bolt it on at every single layer of the stack.
[00:23:41] And, you know, hopefully that makes it far easier for our customers to consume.
[00:23:47] Yeah, that's right.
[00:23:47] We have to architect them together.
[00:23:49] That's the way we look at it.
[00:23:50] Initially, people are kind of bolting it together because that's fast to market.
[00:23:53] But what we're doing is architecting both together to work seamlessly together and to produce better results.
[00:23:58] Great.
[00:24:00] So let's talk a bit about OCI, Nathan.
[00:24:02] What are we doing to help our customers build and run enterprise AI at scale with both governance and security?
[00:24:11] In my opening remarks, you know, I touched on quite a few very heavily regulated industries.
[00:24:16] How do we think about that?
[00:24:17] And how is that built into what we're doing at OCI?
[00:24:20] Well, actually, today we're announcing the OCI enterprise AI platform, which I think is really on point for this.
[00:24:25] So this is the combination of our generative capabilities as well as our agentic capabilities offered in a very easy-to-consume format for customers who are building applications, maybe beyond just the data tier, but thinking about all of OCI, any kind of application where they need to be integrating and infusing AI natively.
[00:24:42] So this is something that really includes both models as well as agents as well as governance.
[00:24:47] So those models so that you can get the intelligence, the agents so you can take action, and then the governance so that you can have the right guardrails and make it happen securely.
[00:24:55] And this is something that we've actually been working behind the scenes on for a while.
[00:24:58] We've got customers already onboarding, customers like TIM in Brazil who are actually using it as part of their solutions for their customer-facing call center applications.
[00:25:08] They're seeing significant reductions in call center times, 30% in some calls, while seeing increasing customer satisfaction, which is kind of the gold standard.
[00:25:17] That's what you want to get out of that.
[00:25:19] So it really is that kind of combination of things.
[00:25:22] That is all then running inside of OCI, right?
[00:25:24] So all of those standard benefits you get out of that high-performance, low-cost, most secure cloud environment.
[00:25:31] As we go through this, what I hope folks can see is that you're building the infrastructure, you've got the database layer, you've got the application layer.
[00:25:42] We're going to talk about the new data platform here in just a minute.
[00:25:45] But the fact that it's all engineered to work together, particularly important when we talk about security and we talk about regulatory compliance, because we're not in the nice-to-have business at Oracle, right?
[00:25:56] Our customers out here are all in mission-critical businesses.
[00:25:59] And as you think about that, we push and pull on OCI with the applications.
[00:26:06] We push and pull on OCI with database use cases.
[00:26:09] How has that helped shape your thinking in the infrastructure, having the rest of Oracle, in a way, as a customer for what you build?
[00:26:17] Well, it certainly means that you've got no room for error, right?
[00:26:20] We have to put a lot of that effort in right from day one.
[00:26:22] And I think you see that reflected across the board, whether that's in the kind of approach that Juan's describing.
[00:26:27] We've got the integrated components we need to be able to set the right governance models inside of the data tier.
[00:26:34] You know, I think when you get into a world where you see customers layering on agents and the ability to go create net new capabilities far exceeds your ability to be aware of or govern that in any other way,
[00:26:44] you need that data layer to really have that governance model built in.
[00:26:48] And it's really advantageous, I think, for OCI that we have this strong relationship directly through that tier inside of Oracle.
[00:26:55] Steve?
[00:26:55] Well, I get very excited, and I think our apps customers get very excited because we inherit, when I hear new features coming up from the database and OCI and new locations and, you know, data support and data security, it's great because it just strengthens our application.
[00:27:09] So we're oftentimes, you guys talk about customers, most of the time, Fusion is your first customer.
[00:27:14] So we deploy just about everything.
[00:27:16] And what our customers see, the Fusion apps customers, is sort of an automatic behind-the-scenes improvement, speed, security, locations, everything.
[00:27:24] It's just fantastic.
[00:27:25] Yeah, that's right.
[00:27:25] We are always the first consumer of all of our technology internally, right?
[00:27:30] Every single layer of our stack consumes another layer, and we always roll it out internally first before anywhere else.
[00:27:36] So it's been something that we hope continues to be beneficial in the age of AI.
[00:27:42] So, TK, the AI data platform is relatively a new announcement for Oracle.
[00:27:49] Can you add some color, maybe some more detail for folks who may not be familiar with AI data platform, how it works, and maybe some of the latest news?
[00:27:59] Yeah, yeah, yeah.
[00:27:59] So we announced the AI data platform late last year at Las Vegas at our annual AI world event.
[00:28:07] And the AI data platform sort of complements everything we just heard from Juan and Steve and what's happening in OCI.
[00:28:13] Because there, it's all about infusing AI right within our core application suite or right within the database where data lives.
[00:28:22] But many customers come to us and say that they have a wide data estate across the organization that consists of Oracle databases and applications.
[00:28:31] It might have third-party databases and applications.
[00:28:33] They might have document stores, and they want to be able to bring all of this data into one place, provide a unified catalog on top of it, not necessarily incur the heavy lifting that's required in moving or copying data, running ETL data pipelines, and so forth.
[00:28:50] They want to create semantic models and ontologies on top of the data because that's what tells these AI models how to effectively reason over the data.
[00:28:57] And then ultimately, they want to be able to build AI agents that can then be infused into application workflows, honor the security models that are already in place in the organization.
[00:29:09] So, we need a general-purpose platform that can help customers operate on top of their entire data estate.
[00:29:15] And that's kind of what the AI data platform tries to do in one integrated platform environment.
[00:29:20] And you mentioned something there that I think is worth maybe double-clicking on, semantic models.
[00:29:26] Yeah.
[00:29:26] You know, along the idea in the vein of Oracle helping our customers prepare their data for AI.
[00:29:32] Yeah.
[00:29:32] Can you talk just a little bit more around semantic models, talk about the different data stores that we can build semantic indexes and semantic models on, and why that's so important for AI?
[00:29:46] Because I think that there's, as we start to talk about the inferencing market, and we start to talk about customers using their own private data, and keeping it private, by the way, and how much business value or government process value that unlocks,
[00:29:59] why is technology like a semantic model so important for AI?
[00:30:04] Yeah, semantic model, the notion of a semantic model, the notion of semantic model is not actually new.
[00:30:08] It's been something, it's been something that's, it's been something that's used in the context of traditional analytics use cases and so forth.
[00:30:16] Because ultimately, semantics on top of the raw data is what tells humans how to reason over the data,
[00:30:24] even if they don't necessarily know how the data is physically organized in a store and so forth.
[00:30:30] And it turns out that AI is no different, but AI models need to know how to interpret your data, how data in one area relates to data in another area.
[00:30:41] Now, when Steve's team builds AI agents for the massive application estate that we have, well, guess what?
[00:30:47] They've got very rich semantics defined on top of all of the data sitting in their ERP systems, in their HR systems, and so forth.
[00:30:55] But customers also need that same ability to provide these semantic models and ontologies over their entire enterprise data estate,
[00:31:03] which might include bespoke custom applications that they've developed in-house, databases that they've collected over the years,
[00:31:10] document stores sitting within their organization.
[00:31:14] So this becomes very important because that's how they tell these AI models to best interpret and reason over their data.
[00:31:25] Yeah, I think it also becomes very important, too, as we speak about the eventual end-user interface,
[00:31:31] which in most cases is conversational, conversational user interface, whether somebody types it in or speaks.
[00:31:37] But having, you know, having a system that creates, you know, semantic ontologies and semantic models is what allows that conversational interface from the end-user to actually translate into how does it interact with the raw data.
[00:31:49] And, you know, I think one of the other things that is really interesting about what you've built at the AI data platform is that data can come from anywhere, right?
[00:31:56] It's not just an Oracle database.
[00:31:57] And sometimes there's a bit of a misnomer that, you know, if it's not an Oracle database or an Oracle application, it's not part of the system.
[00:32:03] But that's not what you've built.
[00:32:04] You've built something completely agnostic.
[00:32:06] Open-ended platform, obviously, it works great with Oracle databases and Oracle applications, but it's really agnostic towards the source of the data.
[00:32:16] And, of course, when you point at data within our own applications, whether it's Fusion or NetSuite or healthcare applications,
[00:32:27] it gives you a lot of immediate value out of the box just because it intimately understands the nature and the structure and the semantics of the data.
[00:32:35] But if it's a bespoke custom application, we provide the tools required for customers to be able to enrich and semantically enrich the data.
[00:32:43] So the AI data platform is now generally available, being rolled out, customers, partners.
[00:32:48] What's some of the feedback?
[00:32:49] What are you hearing?
[00:32:50] Yeah, we're seeing a lot of traction.
[00:32:54] We're seeing a lot of traction on our customers in variety of different industries, engineering and construction, retail, financial services, et cetera.
[00:33:01] Actually, Mike, you and I worked with a customer who happens to be a major construction conglomerate in the Middle East.
[00:33:10] And we worked with them to semantically enrich their data with all of their construction projects and then help them optimize their supply chains, improve profitability and so forth.
[00:33:20] And this was all done with our forward deploy engineers working with the customer in concert, and we were able to get them some AI agents up and running in a matter of a few weeks, actually.
[00:33:31] Yeah, I think that's probably the most impressive part is that the time at which you can deliver value is stunning.
[00:33:39] And it's not just a little bit of value.
[00:33:41] It's really amazing value.
[00:33:44] In that example that we worked together on, you know, in a large-scale construction example, you would think that, well, construction schedules, scope, cost.
[00:33:51] But some of the unintended consequences, positive consequences of building semantic ontologies and layering the AI data platform and Agenic Studio on top of that is to actually get visibility into a supply chain that was suboptimal.
[00:34:05] And it was not actually one of the goals or understood to be one of the goals of the initial platform rollout.
[00:34:10] But we figure out that by aggregating all these projects across the entire country, we're paying very different.
[00:34:15] We have very different component prices, for example, HVAC chillers in one side, one level projects than we do in others with the same vendor, with the same supplier.
[00:34:24] So these are the kind of things that, you know, with the right organization and the right data gravity strategy and the right and the right.
[00:34:31] I keep using the word semantics, but I think it's so important because I know it's not a new concept.
[00:34:35] Yeah.
[00:34:35] What I think is new is the scale at which we can deliver and the scope at which we can deliver this kind of a semantic index and how it becomes.
[00:34:45] And one of the things that we, one of the things we found in that project was you can actually use AI to semantically enrich data as well.
[00:34:53] It's not just, it's just not, it's just not semantic enrichment for the sake of helping the AI models.
[00:35:00] You can actually use AI models to do semantic enrichment as well.
[00:35:03] So it's pretty powerful.
[00:35:05] Okay.
[00:35:05] So that's, that's a little bit of an overview of just the headlines.
[00:35:09] There'll be lots more information throughout the day on, on deeper, on, on, on, deeper detail around the announcements.
[00:35:14] But let's connect now to what I think folks in the audience care about even more trust choice, scale, and doing all that without adding risk or complexity.
[00:35:26] So if there's one layer that has to make those promises quite real, it's the data layer, because that's where access is enforced, policies are applied, audibility lives.
[00:35:40] And if AI isn't trustworthy at the data layer, then, you know, then nothing above it really matters.
[00:35:46] So, Juan, when we say trust in the AI era, what does that actually mean?
[00:35:53] And again, how are we building that?
[00:35:55] I mentioned earlier, very close to the data.
[00:35:57] How are we building that directly into the database so that the agents can work with real enterprise data, both safely and securely?
[00:36:05] And at scale.
[00:36:05] Yeah.
[00:36:06] Yeah.
[00:36:07] So I think this is the key area for AI.
[00:36:12] AI can already build amazing things, literally in minutes.
[00:36:18] It can write a 10,000 line program for you in five minutes.
[00:36:23] It can generate very complex SQL in seconds.
[00:36:27] Then what really it boils down to is how much do you trust that?
[00:36:31] Can you take up something that was generated 10,000 lines of Java, let's say, and deploy it into production?
[00:36:38] That's going to be the limiting factor.
[00:36:40] And what I see over time is we're going to spend less and less time on worrying about how to generate things and more and more time on how much do we trust it?
[00:36:48] How do we validate it?
[00:36:49] And for that, we have to kind of change the way we've been thinking about things.
[00:36:53] It's we can't just do what we've been doing in the past.
[00:36:56] And one of the things that I've been telling people is we have to change the way that privacy and security works.
[00:37:04] Up to now, end-user privacy, end-user security has been enforced at the application level.
[00:37:10] The application level is the only one that understands the end user.
[00:37:14] So they understand, are you a bank user?
[00:37:16] What data can you see?
[00:37:17] What data can't you see?
[00:37:17] Are you a healthcare user?
[00:37:19] What data can you see?
[00:37:20] What data can't you see?
[00:37:21] At the database level, we have no idea about end-users.
[00:37:24] Now, to get the full advantage of AI, we're going to use it at the application level.
[00:37:29] But we also need to use it at the data layer because the data layer gives full access, much richer access.
[00:37:35] And so we have to move this end-user security concepts, privacy concepts, directly into the database to make it AI.
[00:37:44] So you can point AI at it.
[00:37:46] And it works on behalf of an end-user.
[00:37:50] And the database will make it so that it only sees the data that that end-user is entitled to see.
[00:37:56] So, for example, a bank, it will only see my account information.
[00:38:00] I might have several accounts.
[00:38:02] Or medical data, it will only see my medical data.
[00:38:04] You can't see anybody else.
[00:38:05] So it cannot leak anybody else's data because, of course, that is a fatal problem.
[00:38:10] The minute AI leaks private data, enterprises are in huge trouble.
[00:38:14] So that is going to be a key change that has to happen.
[00:38:18] And we're now releasing ability to push that end-user very complex privacy information into the database.
[00:38:26] And then when you're building apps, it's the same thing.
[00:38:28] When you generate these SQL statements, when you generate these apps, how can you trust them?
[00:38:33] So we have to build as much trust as possible.
[00:38:35] So it can basically validate what the AI is doing and say, hey, this is a valid thing to do.
[00:38:42] This is an invalid thing to do.
[00:38:44] That's got to be built as low as possible so it can't be bypassed.
[00:38:49] At the database is the only level that can't be bypassed.
[00:38:53] So the data has to be enforced.
[00:38:53] So the data has to be enforced.
[00:38:54] All that trust has to be enforced as low as possible.
[00:38:57] Yeah, I think it's incredibly important when we talk about inferencing as well.
[00:39:00] We talk about helping our customers take all of their private data and leverage platforms like the AI data platform.
[00:39:09] We want to make sure that we don't have inadvertent situations where data from a bespoke system is merged together in the AI data platform in a semantic index or semantic ontology.
[00:39:18] And we don't accidentally have a situation where somebody now has access to data that was only enforced by some application that was 20 years old and nobody knows how to work anymore.
[00:39:29] So this is really an important architecture step for all this.
[00:39:34] Well, I think we see it up the stack.
[00:39:36] The reason why in the applications level, we talked about speed to implement during TK's part.
[00:39:41] The reason why our agentic apps and our agents are fast to implement is they sit directly on top of the data.
[00:39:46] So we use Mulan's data protection.
[00:39:48] And they have the context of the business.
[00:39:50] So it knows the legal entity setup.
[00:39:52] It knows your workflows.
[00:39:53] It's kind of a different form of semantic layer.
[00:39:54] And it knows all the regulatory compliance things that you have to follow.
[00:39:59] It knows tax treatments and filings you have to make, etc.
[00:40:01] So the agents are being aware of that.
[00:40:03] It's more than a theoretical.
[00:40:05] Could I use AI to build something on top of the applications?
[00:40:08] Sure, theoretically.
[00:40:09] But the complicated part is having that semantic data as well as the business knowledge to be able to execute.
[00:40:16] So let's make it even a little bit more complicated.
[00:40:18] Nathan, we have, as you know, an industry-leading multi-cloud strategy where customers have choice.
[00:40:25] Customers have choice.
[00:40:25] They want to consume Oracle technologies from other cloud.
[00:40:29] Building upon what Juan said, what Steve said, how do we make that work?
[00:40:34] Yeah.
[00:40:35] No matter the deployment, you know, where the deployment is, where the physical deployment is of the Oracle Cloud?
[00:40:39] Yeah.
[00:40:40] Well, just to reiterate, obviously, we now offer our Oracle AI database offerings inside of our cloud partners, Google, Azure, and AWS.
[00:40:49] And that means that customers can take these same capabilities directly running on Exeter hardware inside of those environments.
[00:40:56] And we hear from customers all the time that they want to use their AI pipelines in those environments.
[00:41:01] They want to be using Vertex.
[00:41:02] They want to be using Copilot.
[00:41:04] They want to be using SageMaker and Bedrock.
[00:41:06] And they want to have their data local to that, you know, low latency, high performance accessibility to it.
[00:41:12] But they still need to take all the governance with them, right?
[00:41:16] And that's why I think Juan's exactly right.
[00:41:18] You need to be building this into the data tier so that wherever you take that, you know, we've been doing a lot to make that portability simpler, not just at a technical level, but at a financial and commercial level.
[00:41:29] So we've got multi-cloud universal credits, simplifies the adoption.
[00:41:33] But when you get there, you needed to actually be able to model all of your existing security capabilities.
[00:41:38] So simplifying the access, but then giving the controls as well.
[00:41:41] So speaking of governments, we, as you know, we have many customers with either very strict data residency requirements and or are the custodians of sensitive workloads or very, in some cases, very sensitive workloads.
[00:41:53] How do we bring the same, you know, level of intelligence to these data sources with those restrictions in mind?
[00:42:01] And how do you get the same performance security and controls when keeping that in mind?
[00:42:07] This is a huge differentiation for OCI, I think, versus any other hyperscale cloud vendor, which is that while we focused on building the largest AI superclusters in the world, we've also focused on how can we make that cloud as small as possible.
[00:42:19] So when we want to land a private region, we can do that very easily as well.
[00:42:24] We call that distributed region.
[00:42:26] And that is a huge part of our sovereignty play.
[00:42:28] So we've got customers around the world who are landing OCI regions effectively inside of their own data centers.
[00:42:34] That means they get all of the same capabilities in terms of governance, all the data layer right inside of their own data centers.
[00:42:41] I'll also say, of course, here in the EU, we have launched Madrid and Frankfurt, as well as regions in the UK.
[00:42:48] So there's certainly a lot of other form factors we can go to, but we can go as small as we need to to meet those customer needs.
[00:42:54] Yeah, I think what's quite interesting is that there's one and only one version of OCI, right?
[00:42:59] The form factor, the physical location, whether or not it's consumed in a multi-cloud environment, it doesn't matter.
[00:43:05] There's one version, and that is, we think, it's a customer's advantage because we're able to deliver all of the services, 100% of the services, no matter how big or how small the data center is, no matter how big or how small the consumption model is, as well as all the same privacy, security, cyber defenses.
[00:43:20] Very much so.
[00:43:21] You know, this is an area where we really feel like a lot of the other cloud vendors have been let off the hook.
[00:43:25] There's some sort of idea that, hey, what OCI doing is super special or different.
[00:43:29] We kind of think that's backwards.
[00:43:31] You know, every cloud really should, but at fault, be able to scale down.
[00:43:34] And we've got all 150 OCI web services available in all of those form factors.
[00:43:38] The pricing is the same in all those environments.
[00:43:41] It's still metered billing the way that we would have in any public region.
[00:43:44] It's just the same everywhere.
[00:43:46] And that's really what you want so that it's easy to run your applications and so that you can make sure that it's going to be exactly the same everywhere.
[00:43:53] So, Juan, I want to go back to AI close to the data a second because I think it's important.
[00:43:58] I spoke in the opening comments about moving from incremental gains, you know, gains gained from small bolt-on type AI solutions to a step function, step function automation.
[00:44:10] This concept, again, of AI very close to the data, what does it change in practice for business leaders?
[00:44:16] And what does it change in order to enable faster answers, lower risks, simple compliance?
[00:44:22] And how does it help bend this curve of some folks saying, I'm not sure how I'm going to get value from AI?
[00:44:27] Why is what we're doing here so important to enable all that?
[00:44:32] Yeah, so, I mean, again, the business data, mostly in database.
[00:44:40] The AI has to work with that data.
[00:44:43] You can't move terabytes of data to some external entity when you're trying to answer a question, right?
[00:44:50] So the AI has to be very close to the database and ideally integrated with the database.
[00:44:55] We've implemented what's called AI vectors, which is the new data type for AI directly in the Oracle database.
[00:45:02] We can search those things.
[00:45:03] We can combine them with business data.
[00:45:05] And what it means is, it's pretty simple, is you get better answers.
[00:45:09] Number one, better answers from AI.
[00:45:12] Number two, safe.
[00:45:14] It's much safer.
[00:45:15] And number three, it's much faster.
[00:45:17] So those are the three key benefits that you're going to get by architecting and doing AI together with the business.
[00:45:23] Yeah, I think it's also worth saying, you know, maybe it's obvious, but, you know, the AI in the wild, if you will,
[00:45:31] the large language models don't know anything about this data, right?
[00:45:34] They haven't read this data.
[00:45:35] They haven't been trained on this data.
[00:45:36] So the fact that you can now unlock it, but also still keep it secure and safe.
[00:45:40] It's not like you're, when we say we're putting AI next to the data, we're not giving that data or donating that data to some model here in the wild.
[00:45:51] I mean, what you're doing is actually, you know, exposing it from, from a, from a useful context standpoint.
[00:45:55] And we think, as we said, adding step function like abilities.
[00:46:00] So Nathan, then let's talk a little bit more about, about security because, you know, it, it, it is top of mind.
[00:46:09] How are we delivering, you know, all this performance, all this security as a service in the world of AI, when things are moving so quickly, you know, we're seeing, we're starting to see, you know, there was an ocean in certification, security compliance certifications at any level of the stack.
[00:46:29] That they were, there were sort of static processes and that you certified yourself and then that version of the software was locked down, should not, and this is in the old days, right?
[00:46:37] Like a couple of months ago, that version of the, that version of the stack was locked down, shall never change until the next certification comes out.
[00:46:45] We actually found that to be, you know, I think organizations and governments out there were saying that's probably actually not a good idea because, you know, those trying to do damage have access to, to lots of new technology as well.
[00:46:56] So we need to keep pace with all that, but we also need to maintain all of our security guarantees for customers, all of our certifications that are required.
[00:47:03] How are we doing that at scale in the age of AI?
[00:47:05] Yeah, I think that there's kind of a revolution that needs to get driven across the industry.
[00:47:10] And I think that one of the ways that we're looking at this is there's a whole different set of threat vectors when you've got AI enabled threat actors, right?
[00:47:18] So the kind of brittle security models, I think that we've had in the past are going to get exposed, right?
[00:47:24] So I think one of the places this really comes out that's obvious where OCI is investing is in Zipper, our zero trust packet routing solutions.
[00:47:31] So traditionally, if you look at cloud vendors, they have tended to have network ACL based solutions where you're doing network based security permissions.
[00:47:40] It tends to be a bit of an unknown environment for a lot of customers.
[00:47:44] And it turns out that the threat actors are able to go and figure that out and take action against it.
[00:47:48] By moving to zero trust packet routing, we're actually resetting to think about service to service allowable actions.
[00:47:55] And so it means that some of those common kind of cloud exploits you've heard about in the past, like open as three buckets and things start to just go away.
[00:48:03] You begin to eliminate whole categories of risk.
[00:48:05] Now, certainly there's the ability to go offer and we do AI enabled detection and other mechanisms on top of that.
[00:48:11] But I think some of those fundamental primitives being, you know, kind of changed around to think about, you know, I no longer can rely on any kind of obscurity for the types of brittle approaches that I've taken in the past.
[00:48:22] So, let's talk a little bit more about zero trust packet and what it means in, you know, in practical terms.
[00:48:32] Does that mean that we sort of assume all traffic is bad traffic?
[00:48:36] Like, what does it mean in, you know, non-tech terms?
[00:48:41] It means more that you're defining that you've got two services that you would like to have rights to be able to communicate with each other in a way that you've defined, you know, as a developer,
[00:48:51] as an application tier rather than assuming somebody above you in the networking tier has made the right choice about who's got access between two points.
[00:48:59] That idea, I think, really made sense at a point in time.
[00:49:03] But I think these days it just doesn't make sense anymore.
[00:49:05] We've recognized we have to be doing this more intentionally.
[00:49:08] And when I say that I assume that we assume all traffic is bad traffic, I don't mean to be negative on that.
[00:49:12] What I mean to say is that we trust and re-verify at the packet level, right?
[00:49:16] Not just the network level, not just the application level.
[00:49:18] That's exactly right.
[00:49:19] That's exactly right.
[00:49:20] Yeah.
[00:49:21] And we've now done this, you know, with technologies like Acceleron where that is now a native part of the solution for OCI when we talk about network packet routing, which means that there's very little performance impact of any of those benefits that you get from that security capability.
[00:49:34] Well, that's great. Well, thank you. I mean, thank you all. A very helpful overview. And again, lots more to come here in the coming days.
[00:49:41] So as we as we as we move towards towards a wrap in the spirit of doing things in in seconds that used to take hours or minutes that used to take days and weeks, I'm just going to go line by line here.
[00:49:57] Tell us what AI will look like in your product area a decade from now, maybe an unfair question and how fast things are changing.
[00:50:05] But Juan, at the data layer, what is what does AI look like for customers day to day?
[00:50:12] And what what's the transformation that, you know, seconds to hours, months today, you know, now and a decade from now, what does that look?
[00:50:19] Yeah, I mean, I'm not sure 10 years, I don't even know what's happening. I think in the next year or two.
[00:50:24] I would say the same thing. That's a big enough change. So I didn't write the question.
[00:50:28] Yeah. But how about next week, next month, next year?
[00:50:33] Yeah, that's an easier question. So I think the biggest transformation that's going to happen is today data is kind of owned and controlled by specialists, by knowledgeable specialists that know special languages like SQL.
[00:50:51] And what's happening is business users are going to be able to directly interact with data and you can already does this. It's not it's not a future.
[00:51:00] The real the real breakthrough is going to be you're going to be able to talk business users talking to the data, asking questions, getting answered and being able to trust the result.
[00:51:09] And that second part is the really, really hard part, because talking to the data, you can pretty much do today, but trusting the results is the hard part.
[00:51:17] And same thing when you need some kind of a service or an application, the business users will be able to do that again in natural language, ask for the thing that they want.
[00:51:27] And then again, the key part of that is trust that that does exactly what they asked for and doesn't, you know, spill private data, doesn't corrupt the data.
[00:51:37] All that has to be baked in or else you just can't use it. So that's going to be the big thing today.
[00:51:43] Today, it takes days when a business user asks a question that's not in an existing report.
[00:51:48] You got to go to somebody who's got a queue of things to do, who gets around to it eventually and gives you the data that's going to happen in seconds.
[00:51:54] And same thing when building an app, it'll happen in seconds. And that's going to be a gigantic change.
[00:51:59] It's going to unlock incredible productivity.
[00:52:02] So built in, not bolted on AI, Nathan, how does that help the CIO's worry list tend to get smaller?
[00:52:10] Yeah. So I think when you play that tape forward, you know, X number of years, you're going to see some of the kind of AI paranoia, I think, that they've had to deal with beginning to get resolved.
[00:52:21] Right. So the security paranoia, I think, around the fear of data access, I think, as Juan is describing, I think begins to get a lot more manageable.
[00:52:29] I think you're going to have a lot of customers who've gone through that journey and they've learned effectively how to manage that process.
[00:52:35] I think when you see solutions that we're providing, that's certainly, you know, part of that.
[00:52:39] I also think you start to get away from some of the FinOps focus.
[00:52:42] I think we start to see some shifts to seeing more of our productivity and value and less about just token cost and some of the other kind of areas.
[00:52:49] I think we've gotten high centered in the immediate term.
[00:52:51] It's going to be a, you know, a pretty radical shift, I think, over the next 10 years, though.
[00:52:56] Steve, uncovering the new art of the possible, you know, what's something, an example of something that we're not maybe not even considering today that you think can become fully automated?
[00:53:09] Yeah, well, I'm going to go 10 months or closer to 10 months and 10 years.
[00:53:13] You know, I think a couple of things.
[00:53:14] So first off, with the agentic apps to repeat, I think people are going to have to get their heads around, this is more than just process automation.
[00:53:20] It's more than just, it's a reasoning process automation.
[00:53:24] So you can now ask the system questions.
[00:53:27] The system has access to your data, has access to your rules, can apply some reasoning logic and give you recommendations.
[00:53:33] I think that's very, very different.
[00:53:35] The second aspect, if you imagine a world where everything is automated, all your transactions, well, we've done a decent job of automating manual transactions.
[00:53:44] Now imagine everything's automated, but it's, again, it's automated with a reasoning automation.
[00:53:49] This is not, you know, an RPA type of solution, you know, beefed up.
[00:53:52] There's a reasoning to it.
[00:53:53] So all of a sudden, you can make, you know, micro decisions on every single transaction.
[00:53:58] You want, you know, yes, we can automate all your invoice processing.
[00:54:01] But in addition to that, you can make real-time decisions on every invoice based on the interest rate fluctuation, based on currency fluctuation, based on if a supplier's giving you early payment discounts, based on if you get a rebate from your virtual card.
[00:54:13] You could never do that today.
[00:54:14] One, because it's manual.
[00:54:15] Then two, to the extent they're automated, the automation is very push button.
[00:54:19] So it's, this is intelligent automation with some reasoning behind it.
[00:54:23] It's going to totally open up the ability.
[00:54:25] Yes, to make you more efficient, but also have this system advise you and then to have you make micro optimizations or real-time optimizations, which again, in today's environment, if you're talking about supply chain or cash or anything managing your business, it's going to be a significant differentiator for our customers.
[00:54:42] Thank you.
[00:54:43] And finally, TK, what kinds of work that we're doing today really just moves to the background?
[00:54:49] What becomes trivially easy, what becomes safer by default, and what ultimately becomes possible that isn't possible?
[00:54:57] Yeah, yeah.
[00:54:58] I think, yeah, I think like, like, like, like Steve, like Steve and Juan said, I think it's a matter of months.
[00:55:03] I think it's a matter of months, maybe a year or two where an enterprise, where an enterprise can, an enterprise can have a central, it can have a central catalog of all of, a central catalog of all of their data assets, their MCP servers, their tools, not just, not just the agents, not just the agents that are, you know?
[00:55:17] Available within their applications, like what Steve was talking about, but agents that maybe are embedded within their databases or bespoke agents that they've created.
[00:55:29] Also, all of their data, all of their document stores, SharePoint sites, Confluence sites, and so forth, all of that information is available through a central catalog.
[00:55:38] And business users can then start to put AI to work and automate their day-to-day lives.
[00:55:44] For example, you can, you can have conversation interfaces over the data in your organization, which can then go reach out to a collection of different agents that can collaborate with each other.
[00:55:55] And essentially give you an experience where, and you can ask questions like, hey, what are my, what are my top, who are my top customers in a specific region?
[00:56:04] What are their, which are the ones with high satisfaction scores?
[00:56:08] What sort of new products can I be pitching to them?
[00:56:11] What references should I be using?
[00:56:13] Or maybe, or maybe you're not even, or maybe you're not even conversing with the agents.
[00:56:17] You just put these agents to work for you while you're off doing something else.
[00:56:22] Maybe you're at, maybe you're in bed at night and you've given an instruction to your personal agent to say, hey, when, if that payroll processing job for whatever reason fails in the middle of the night, go do some basic analysis and send me a report or text me.
[00:56:37] If, if something bad happens, things of that nature, I think will become more and more possible as we start to bring all of these enterprise wide data assets, MCP servers, agents and so forth, and provide those kinds of conversation interfaces.
[00:56:56] And there's a lot of work happening in the industry around these kinds of experiences.
[00:57:01] We just launched the agent hub experience in the area to platform that is also our attempt at cracking this problem of dramatically improving businesses productivity in an enterprise today.
[00:57:15] Well, excellent.
[00:57:16] Well, thank you all very much for being here.
[00:57:17] Thanks for sharing a bit of a preview as to what we're up for.
[00:57:20] And again, lots more to say here throughout the events in every one of these, these areas.
[00:57:26] Thank you.
[00:57:26] Thank you.
[00:57:27] All right.
[00:57:41] So, of course, the real test of any technology is what customers do with it.
[00:57:47] So we've saved the very best segment for last.
[00:57:51] The final segment is where we will truly hear to hear from our customers leading the ways in their industries.
[00:57:59] We're going to talk with two very big brands doing big things and hopefully will give us all some very big takeaways to think about in our own businesses.
[00:58:09] So, first, please help me welcome Erica from NHS Shared Business Services.
[00:58:16] Please welcome to the stage Erica Bannerman.
[00:58:20] Thanks very much.
[00:58:25] Thank you very much.
[00:58:26] Thank you.
[00:58:27] Great to be here.
[00:58:28] All right.
[00:58:33] Well, thanks again for being here.
[00:58:35] Okay.
[00:58:36] Oops.
[00:58:37] It's an absolute pleasure.
[00:58:38] No, I've just completely lost the spot here.
[00:58:40] Hang on one second.
[00:58:41] There we are.
[00:58:42] All right.
[00:58:43] Well, thank you again for being here.
[00:58:45] So, as we've heard already in the early moments of the event, AI is reshaping expectations around efficiency and accountability.
[00:58:55] And I think that's true also in public services.
[00:58:58] How is that influencing the way you think about finance and operations across the whole of NHS?
[00:59:04] Yeah.
[00:59:05] Thank you.
[00:59:06] Well, first of all, I'd like to say that the relationship and partnership with Oracle is really important to answer that question.
[00:59:13] But let me start with the NHS.
[00:59:17] I think some of the audience might know the NHS.
[00:59:20] It's very complex.
[00:59:24] It's systems within systems.
[00:59:26] And it has multiple hundreds of organisations, sovereign entities that are dependent on each other as well.
[00:59:35] So that's the context that we need to answer the question.
[00:59:39] And so it also has a lot of demands and the demands that the national health system has at the moment, which are very well recorded, are around cost, patient expectations.
[00:59:54] And we have a very long legacy with technology infrastructure.
[01:00:00] So we've got to fix all of those things.
[01:00:03] And so we've been looking at it within NHS share business services for over 20 years and been working with Oracle for two decades.
[01:00:11] But we're really excited about what AI can do.
[01:00:14] It's rethinking and reshaping what's possible for the first time.
[01:00:18] And you need, and I heard with the other panels, you've got to start with the foundation.
[01:00:23] So last October, we went live with our finance platform.
[01:00:29] We had 48 NHS organisations who actually joined us with that.
[01:00:35] And our teams together and my team, and I can see them at the front here in the audience, because this is really important to us.
[01:00:42] They have a lot of domain experience.
[01:00:44] We really know how the NHS works.
[01:00:46] And bringing that together with Oracle's technical capability, we had about 30,000 hours of workshops.
[01:00:53] And from there, we started to see the real potential.
[01:00:57] And so we now feel that the next year, three, five years, are really going to revolutionise how we do things together.
[01:01:06] Well, it's quite impressive.
[01:01:07] You mentioned big and complex.
[01:01:09] Your financial processes today handle up to 250 billion pounds annually.
[01:01:16] What does operating at that scale, at that level, require from your modern technology architecture?
[01:01:23] Yeah.
[01:01:24] So I heard you say earlier that you're dealing with critical national infrastructure and you can't get it wrong.
[01:01:30] Mike, we can't get it wrong.
[01:01:32] So that's the first thing that's really important when you're dealing with.
[01:01:36] So we manage around 355 billion transactions of cash on behalf of the health service, which is more than the budget.
[01:01:45] And it's really important to know where that is going, because it's going right to the front line.
[01:01:50] So it's really important that we think through that.
[01:01:53] So there are a few things we need to do.
[01:01:55] The first thing is we need to reimagine how we manage the processes.
[01:02:00] So we need to look at end-to-end orchestration.
[01:02:04] It's not good enough anymore to take a task and automate it.
[01:02:09] It doesn't give you the gains you need.
[01:02:12] So we needed to think about it from that point of view.
[01:02:15] Secondly, we needed to standardise.
[01:02:18] Standardisation is really important when you're dealing with critical national infrastructure.
[01:02:22] They're supporting the whole health service.
[01:02:24] So that was really important.
[01:02:26] You need strong governance.
[01:02:28] And we're having to rethink a lot of that for AI around ethical governance.
[01:02:33] So we needed to ensure we had the right processes and practices so that that would work.
[01:02:38] And with those combination of things, the last that's critical is the data.
[01:02:45] And what we got very excited about when we think about Argentic AI and all of the different virtual agent capabilities that you have that I know the audience have heard this morning.
[01:02:57] We have a lot of unstructured data within the NHS.
[01:03:01] So what's super exciting now is what we're going to be able to do with that, which will make a real difference to the front line.
[01:03:08] And so that's really important.
[01:03:10] And those are the four elements that we thought about.
[01:03:13] So you spoke about standardising, which is sometimes easier said than done, but you've done it.
[01:03:18] You've moved to a single cloud-based financial platform.
[01:03:23] What are some of the tangible improvements that you're seeing from that decision in terms of cost control, oversight, resilience, or anything else that may be a benefit of standardisation?
[01:03:33] So on the 1st of October, 2025, we went live with the new platform and we needed to process 19 billion pounds on the first day.
[01:03:47] So I think that's quite tangible.
[01:03:51] We needed to be sure that we had the security, the governance, the controls so that we could actually ensure that the money was getting to the front line.
[01:04:00] And that was to pay NHS employees, suppliers.
[01:04:05] It was critical that it worked.
[01:04:07] So that's one of the things we're really proud about because it made a really big difference.
[01:04:14] Our teams, and a number of us didn't have much sleep that day, were in the office 24/7 operation.
[01:04:23] And we were working with our front line teams and saying, "How does it feel?"
[01:04:29] And it's intuitive.
[01:04:32] It didn't need a lot of training.
[01:04:33] And they were feeling, they were saying, "Some of our tasks we're doing in half the time."
[01:04:38] So there are some remarkable productivity and efficiency gains.
[01:04:42] But for me, what's more important than that is we're freeing up time for our teams to actually add further value, to be thinking about not what the retrospective report was telling us around our financial data, but what we can do to make an impact in the future.
[01:05:00] So looking around corners and thinking about really what the insight and the data is telling us.
[01:05:06] So that's where I think we're going to see some really significant benefits moving forward.
[01:05:11] When we implemented the range of Oracle applications, we also saw some cost effective efficiencies.
[01:05:22] And I was talking to some of the panel backstage and we were talking about what those efficiencies were.
[01:05:26] And apparently half a million pounds isn't enough.
[01:05:30] We've got bigger ambitions.
[01:05:32] If the whole system adopted our solution, we would save three billion pounds annually for the national health system.
[01:05:42] That's a lot of nurses.
[01:05:43] That's a lot of hip operations.
[01:05:45] You know, there's a lot to go after there.
[01:05:48] And then the final piece is that sometimes our experience, and we've had a lot of lessons learned.
[01:05:54] And there's been a lot of times where we've had to fail fast, Mike.
[01:05:59] So, you know, that's important to know.
[01:06:01] But we made a commitment that we would build this solution together and it delivered 30% cashable savings in its first year.
[01:06:12] So I think that's, you know, worth noting.
[01:06:15] Well, you've achieved, you know, a stunning amount here in relatively short order,
[01:06:22] even though it's been a decades long partnership, as you said.
[01:06:25] You know, the new technology that you've rolled out has been done so with, you know, with great expeditious priority.
[01:06:33] What's your next priority?
[01:06:35] You know, as you look ahead, what's next on the list for things?
[01:06:39] And does AI play a role in any of what's next on the immediate basis?
[01:06:44] The next party is a good way to describe it because actually we're all learning.
[01:06:50] We're all developing.
[01:06:51] We're all very excited about what we know.
[01:06:54] We're all very clear.
[01:06:55] There's a lot we don't know.
[01:06:56] You know, I'm not a technologist.
[01:06:59] I don't know what's under the bonnet of this that my team here do.
[01:07:04] But what I do know is that we need to deliver more impact back into the health system.
[01:07:10] And so we need to double down on the potential that AI provides us.
[01:07:16] We need to know how to use that.
[01:07:18] And we need to be thinking about how every one of us is actually using it in everything we're doing.
[01:07:25] You know, and I imagine that we're going to be sitting side by side with our virtual agents and we're going to create an environment where the things that only you and I can do, that we have the time to do that.
[01:07:39] So what we're looking at at the moment, when we chose to partner together, what was really exciting was that you've embedded AI into the platforms and we can take advantage of all of those applications.
[01:07:54] And now what we can do is work together to actually create those efficiencies.
[01:08:00] And we've got so much demand.
[01:08:02] So the challenge I leave you with is we need to work through how we manage that demand, how we're going to get after that data challenge so that all of the system can adopt what we know is going to make a very big difference.
[01:08:19] Because this is a one in a lifetime generational change for the national health system.
[01:08:25] So we need to do that together collaboratively.
[01:08:27] Well, well, again, what you what you've achieved here is is is quite stunning.
[01:08:33] Thank you very much for the partnership.
[01:08:35] As as you know, health care has been something that's been, you know, near and dear to the mission at Oracle.
[01:08:42] It's a business imperative, but we also view it as a moral imperative.
[01:08:46] I mean, we owe it to patients and those who care for patients throughout the world, which in some ways encompasses every single human being on the planet to make sure that we're delivering the best technology that makes it possible.
[01:08:58] And I think that what you've done here is just a shining example as I travel throughout the world and meet with governments and health organizations.
[01:09:07] You know, NHS is always at the top of the list of of exemplars in the world.
[01:09:12] So thank you very much for making all that possible.
[01:09:15] It's truly humbling to be to be your partner.
[01:09:18] I appreciate it.
[01:09:19] Thank you.
[01:09:20] Thank you.
[01:09:21] Thank you so much.
[01:09:31] OK, so we're going to bring up now our final guest today.
[01:09:35] Please help me welcome Val Cemi from Zoom.
[01:09:40] Please welcome to the stage Val Cemi Sankalingam.
[01:09:45] Hey, Mike.
[01:09:50] Good to see you.
[01:09:51] Thank you.
[01:09:52] Please.
[01:09:54] Well, welcome.
[01:09:55] Thank you.
[01:09:56] Thank you as well for being here.
[01:10:01] Zoom has become...
[01:10:02] It's great to be in London and great to be at the event.
[01:10:04] Thank you.
[01:10:05] Oh, thank you.
[01:10:06] Zoom has become just really synonymous with meetings.
[01:10:11] How should, you know, how should business leaders think about what Zoom is becoming over the next few years?
[01:10:18] You shared some very interesting things with me earlier that I think...
[01:10:22] Yeah.
[01:10:23] Because I think it's an honor to be considered synonymous to meetings, but Zoom has built a lot more products than meetings.
[01:10:30] So we have...
[01:10:31] Zoom has Zoom Phone, which is an enterprise PDX, and we already have more than 10 million seats on that.
[01:10:39] And Zoom has a contact center, which is a full-blown contact center.
[01:10:43] And I think it's one of the fastest growing business with being in the Gartner within the first three years.
[01:10:51] And also, there is other collaboration tools like Zoom Chat, Zoom Clips, and Zoom Whiteboard.
[01:10:59] In addition, we have also built Zoom AI Companion.
[01:11:03] We are already in version 3.0.
[01:11:05] AI Companion started with providing meeting summaries, but it slowly evolved into not just providing meeting summaries, providing actions, follow-up actions.
[01:11:15] And now it is at a stage where it can actually take actions for you.
[01:11:18] So the way I would look at Zoom is you should consider it as like a platform of action which helps both your employees get work done and your customer service gets work done.
[01:11:30] You know, along those lines, we spoke earlier about just the changing nature of work and how AI will influence the changing nature of work.
[01:11:37] Obviously, Zoom was, you know, a pioneer, particularly during the pandemic and just changing the way people work.
[01:11:43] Now we fast forward, how are you, how is Zoom thinking about the way AI changes work and what specifically are you building into the product set to enable that?
[01:11:56] Yeah, one thing we see is reduce the friction between your communication and your action.
[01:12:03] That's the key that we are looking at.
[01:12:04] Because if you look at it, you have phone calls, you have maybe Zoom meetings, all those data is completely different from your system of record.
[01:12:13] And the only way the data can actually move between is somebody manually going and entering the data.
[01:12:17] For example, a salesperson has a call and it doesn't get updated to the CRM unless the person does that.
[01:12:24] We feel those are the efficiencies that we can actually build into Zoom.
[01:12:29] And if you look at our AI companion and customer AI companion, they pretty much integrate.
[01:12:34] They integrate all of the Oracle fusion suites.
[01:12:37] And so it's a two way integration.
[01:12:40] So if you have a conversation, it's actually good to get all the data from the systems.
[01:12:45] And at the same time, whatever the intelligence that come out of the conversation actually should go back into the system.
[01:12:51] So that's something that we are looking at.
[01:12:53] Yeah.
[01:12:54] So how do you expect this Oracle fusion applications integration?
[01:12:57] You spoke about critical functions, sales, service, hiring.
[01:13:01] How do you think in the future we'll even do deeper connections between what happens in Zoom to, as you said, the system of record like Oracle fusion application so that it's completely seamless for our mutual customers?
[01:13:15] I would say, you know, I work for Zoom.
[01:13:19] I know I like meetings, but I spend most of my time in meetings.
[01:13:23] I don't get any time to do the work in the meetings at all.
[01:13:25] So what if actually while you're in meeting, the AI agent can actually do the work for you?
[01:13:32] It can actually create the actions of the next steps and you can provide all the integrations to say that, okay, if it is a sales call at the end of it, update all the data into like Oracle fusion CRM.
[01:13:45] So all the intelligence of that particular call actually goes there.
[01:13:49] And then if a sales person is actually trying to prepare for a meeting, they can actually, Zoom AI can actually pull the data from the Oracle fusion CRM and provide all the context from the CRM, maybe from the customer support, so they can pretty much get all the intelligence and say, hey, this is the summary of all the meetings.
[01:14:06] And that context also helps the meeting to be more intelligent and provide the necessary information for the real time.
[01:14:15] But at the same time, whatever happens in the real time actually goes back into the system of record as well.
[01:14:20] And today with the integration, you can either do within Zoom, you can actually update all these Oracle fusion or within Oracle fusion, actually you can look at all the data in zoom.
[01:14:30] So that's a level of integration that we already have.
[01:14:32] Well, that's great.
[01:14:33] Yeah, we certainly, as we said, we're rolling in quite a bit of this out internally at Oracle as well and really excited about, about what you're building together with the fusion applications.
[01:14:42] Let's go back a little bit in time.
[01:14:48] I spoke about mission critical, can't fail, can't fail, you know, elements of the stack.
[01:14:53] And I certainly put Zoom in that category.
[01:14:55] I mean, you know, people don't like if, if the system's down, right, particularly since it's so fundamental to, to the way we work today.
[01:15:01] What was, you know, seems like the early days wasn't that long ago, the pandemic was not that long ago, but, but it does seem like a while ago now.
[01:15:09] What was non-negotiable for you as you selected a cloud partner and, and how did that shape your decision to run so many of your key workloads on Oracle OCI?
[01:15:21] Yeah.
[01:15:22] So when you run a service, I think the most important thing is keeping the service up and running.
[01:15:26] Yeah.
[01:15:27] That's the baseline.
[01:15:28] If you don't have that, whatever cool feature you have, nobody cares.
[01:15:32] So that's number one.
[01:15:34] And that's basically part of Zoom's DNA.
[01:15:36] We want to make sure it's easy to use since always available.
[01:15:40] And for that, I mean, we, we've been partnering with the OCI for a long time since the pandemic.
[01:15:46] We've been partnering even before.
[01:15:48] And I think the, the resiliency that OCI provides will be the foundation for like what we do at Zoom.
[01:15:55] And if you look at, so one is you look at a business, the bottom line is your availability service always up and very easy to use.
[01:16:05] I would say the next one is like speed of innovation.
[01:16:09] Like how do you innovate faster?
[01:16:12] We work very closely with the OCI on, we use like a federated model for all our AI and we use the OCI for that as well.
[01:16:21] And it's not, when you say like go to market, it's when you say the innovation speed, it's not just the product.
[01:16:28] When you're running a service, it's also being available in regions where we want to do business.
[01:16:33] And we've worked extensively with OCI to get into certain regions.
[01:16:37] There's a lot of data residency requirements in the world today.
[01:16:40] So we actually work with all of that.
[01:16:42] And then the last part is you want to have a high gross margin as well.
[01:16:46] And we work very closely with the OCI to keep our cost low with a federated AI and OCI is part of the federated AI and it keeps the cost low.
[01:16:55] These, I would say, are the three key areas like running a business like Zoom.
[01:17:00] And I think we work with Oracle and OCI and the go to market also is another part where we use like a fusion as well, because one is like you basically have the product ready and you have the product ready in region.
[01:17:12] But you also need to have the pricing models, the back end to recognize the revenue and all that.
[01:17:17] We need agility and all of that.
[01:17:18] So we work very closely with Oracle in all of them.
[01:17:21] Well, we certainly appreciate the trust in OCI.
[01:17:23] You were one of the original anchor tenants and partners in OCI at scale, at global scale.
[01:17:31] And frankly, you pushed us to make OCI even better.
[01:17:37] So we certainly appreciate it.
[01:17:38] It's a mutual benefit.
[01:17:39] It's been a mutual great partnership.
[01:17:41] Speaking of the partnership, final question.
[01:17:44] I won't ask about the decade.
[01:17:45] Let's just keep it to the next 12 months, which even is a long time in the age of AI.
[01:17:51] What are you most excited about in terms of the Zoom Oracle integration and what should we expect to see?
[01:17:57] I would say, I mean, we continue to work closely with the OCI, using OCI for all the different areas that we talked about and even the Oracle's products.
[01:18:07] But I would say if you look at the way the world is changing, we talked about this whole conversation to action.
[01:18:14] So the conversation systems and the systems where you're actually doing the business, like we need to figure out like how to connect all of those.
[01:18:20] And that's where we are actually our AI companion and custom AI companion are working closely with that.
[01:18:26] And we are tightly integrating with Oracle as well.
[01:18:29] The way I would look at it is I think all of you need to look at keeping those two data separately is a lot of inefficiencies.
[01:18:36] And I think how do you make the business even better?
[01:18:39] So one example I would give is, I mean, we are living this ourselves as well as part of rolling out all the products internally.
[01:18:47] One of the things is, for example, our support.
[01:18:51] I think if you look at any support organization, the primary goal is, hey, what type of CSAT do I get?
[01:18:58] And then what are the costs per ticket, et cetera.
[01:19:01] But rolling out something like our contact center with our virtual agent, what we've been able to do is maybe we are looking at the symptom, not the real problem.
[01:19:11] The real problem is why do we have tickets?
[01:19:14] Okay, instead of saying like, hey, I want to raise the cost of the ticket.
[01:19:17] Now we have been able to go inside and then say that why does a customer actually open a ticket.
[01:19:24] And one of the targets that I've given my team is reduce ticket by 40%.
[01:19:28] That means it's not just the customer support team.
[01:19:30] It goes all the way back to the product.
[01:19:32] And tying all these together is actually key in terms of including the conversation and all the business systems.
[01:19:39] Today, the conversation separate and each business system is very siloed.
[01:19:42] I think that our AI companion, customer AI companion, all of these can be brought together and you can actually solve some of the real root causes of the problems and not necessarily the symptoms.
[01:19:54] And I would say, I would request everybody, I'm assuming most of your Zoom customers enable Zoom AI companion.
[01:20:00] I think that's part of, and also we launched something called AI notes.
[01:20:04] That's something for the individual productivity.
[01:20:06] So if there is an attendee that's attending multiple meetings, it actually gives them a lot more intelligence, not just Zoom meetings.
[01:20:12] It supports Teams and that's all part of your subscription as well.
[01:20:16] So thanks for the partnership.
[01:20:19] Well, thank you.
[01:20:20] Thank you for helping us get so much done together and giving us even more exciting things to look forward to.
[01:20:26] Thank you.
[01:20:27] Thank you very much.
[01:20:28] Thanks a lot.
[01:20:29] Okay, so that's just a bit about how NHS and Zoom are optimizing performance.
[01:20:42] And the imperative for all of our customers in every industry is harnessing all of their data, no matter where it sits, so that AI could use it as the fuel that propels us into the future right now.
[01:20:56] And as I said earlier, the data is where it starts.
[01:20:59] It's at the heart of our origin story at Oracle.
[01:21:01] And we will continue to innovate in our database layer, which is trusted by the biggest banks, governments, healthcare organizations, and so on.
[01:21:10] But now I want to make sure that I say a special thank you to all of you.
[01:21:14] Since Oracle's founding, you have trusted us with your most mission critical challenges.
[01:21:20] And your progress with our technology continually pushes us further to the point where we're all leading this AI revolution together.
[01:21:29] These powerful tools are in your hands.
[01:21:32] We're encouraged by what you're already doing with them.
[01:21:35] And we'll be right alongside you as you use them and leverage them to shape the next chapter of your business, both here and around the world.
[01:21:43] Thank you very much for investing your time with us.
[01:21:46] Thank you for the partnership.
[01:21:47] Enjoy the rest of the event.
[01:21:48] Thank you for the rest of the event.
[01:22:18] Thank you.
[01:22:48] Thank you.