Google Cloud Summit Special: John Abel, MD, Office of the CTO at Google Cloud, and Alex Rutter, EMEA MD for AI at Google Cloud

0:00In this era, speed doesn't mean at the cost of security, doesn't mean at the cost of scale, doesn't mean at the cost of having poor data. Today, over 75 % of Google Cloud customers utilize one or more of our AI products in their business. So we're starting to see real widespread adoption across the platform as a whole.

0:25this week's episode is a little different the google cloud summit 2026 in london has just taken place bringing together tech leaders from all over the world to discuss the latest developments in ai cloud and data from one of the major players the event featured a number of significant announcements and there was a clear theme running through the event that the conversation has clearly moved on. We're no longer talking about whether organizations should adopt AI, but how they deploy it at scale and turn it into real business value. So with that in mind, I had the opportunity, I was very privileged to speak with two senior executive leaders at Google Cloud to get their perspective on where the market is today and where it's heading next.

1:12First up, the Managing Director of the Office of the CTO at Google Cloud UK in Ireland. Without further ado, this is Mr. John Abel.

1:25How's today gone? Do you know what? Brilliant. Lots of conversations, showing the latest technology and also helping people transition to the new generation of this technology that will hopefully help people a lot in the world going forward. Absolutely. I think it's some of the most transformative technology we've certainly ever seen. But for the listeners who are unfamiliar with yourself, John, can you give us a little bit of introduction of your journey with Google Cloud and your current role? Yeah, so I've been in the tech industry since the 80s. And then seven years ago, I joined Google Cloud.

2:00I'm in the office of the CTO, the managing director of the EMEA team. And we are an engineering group that works into Google Cloud. we do emerging technology and experiments that hopefully will end up in a product. Fantastic. So in terms of, obviously, we know about the importance of some of the announcements today, but from your perspective, can you talk us through the key announcements, the key themes and the things that have been announced so far that are really important and profound? Yeah, for me, first of all, the biggest one and the one that I'll probably home in the most is the stack of technology that is actually operating as one one of the things that we already know is a lot of people uh struggle with this very uh large ecosystem that's been born in a short time period and one of the things that we have as a duty of care is making sure that people get the value they need out of it in the fastest route they can so when you look at like security with agentic defense or you look at the agentic data cloud where we're accessing data, or even the ability of using anti-gravity to provide code, now having one stack where it's integrated and actually provided is super essential.

3:16In this era, speed doesn't mean at the cost of security, doesn't mean at the cost of scale, doesn't mean at the cost of having poor data. So taking that responsible route to making sure that we have the best technology that's integrated, that provides the best value, allowing the users to focus on what they want to be famous for is essential. Yeah, absolutely. I think the seamless transition from doing what they've always done to using these incredible tools to produce better outcomes, that's the main barrier to adoption, isn't it, from an individual standpoint? Correct. And if you think about this era, machine learning has been around for multiple decades.

3:57And we've used it in financial services, in many industries, healthcare example. This generative AI era that's now moved into the agent era is about actually knowledge experts getting their hands on the technology to be incredible knowledge experts. One of the things I'm really fascinated by is that we're in the era where we're going beyond human scale limits. So I can scale myself much easier in this world with agents and this agentic world we're moving into than I could ever do in the past. Yeah, absolutely. So we often hear that sort of AI success starts with data, of course. Yes. In your view, John, are organizations generally further behind on data readiness than they realize?

4:43That's a great question. I think the first thing is what data do I need to answer the question? And what I mean by that is that it's less about do I have data? It's more about do I understand the contextual benefit it gives me. And one of the things that you need to look at is actually, what do I need to do the job in hand? So when I look at agents, for example, I look at the skills, e.g. the things that they need as skills to be able to perform the job. I look at the tasks that they need to perform. When you look at the skills, you then quickly look at APIs and data that's required. Once I understand the data that's required for that, I understand, is the data of the right quality to help my agent achieve it with the smallest amount of human in the loop.

5:26Once you go through that, you quickly find out how good or bad your data is. Yeah, no, absolutely. Very well said. So I think we've spent the last couple of years talking about AI models, essentially, and extracting value from models. But the announcements today and recent discourse and product releases from the major technology players, it's all shifted towards agentic AI. So can you just talk us through that development and what that shift means from, you know, first generation AI, if you like, to agentic AI? Yeah, and I'll actually give you an example. Today, I did a keynote where I showed an agentic board running.

6:09So I took a, I created a simulated annual report. I gave it to board members that were agentic agents that ran in autonomy against the persona style that I gave them like COO, CMO, CEO, and I asked them to evaluate the annual report. Now imagine if I gave them the profile of my actual execs or if I gave them the profile of how those people should operate. It's very easy to find information about how a CFO should work. It's very easy to find out how a CEO should work. My agents can quickly process this large array of data, hundreds of pages, and go through scenario, and they can go through the ability to give me guidance of the things I need to worry about.

6:51Well, that's game-changing. Because today, getting board execs together is quite an under-tasking. I can get a first pass. then when my real people meet, I can have them not only use their own intelligence and their own creativity to look at problems, I can enrich it with the agent's view. So the agent becomes like a team member of the board. Yeah, that's incredible, isn't it? I think we are increasingly starting to perceive agents in that way, but I think it's got a long way to go. So in terms of pitfalls from an organizational and a governance standpoint, John, in terms of adopting AI agents to deliver real-time tasks across the business.

7:33What kind of challenges are you seeing for companies from a governance standpoint? First of all, there's always the discussion about security data. They're not going to go, and that's something that we've been really homing in on. And obviously, with the recent acquisition of WIS, you would have seen that we've really reinforced our security portfolio. I think the bit I'm most interested in is not the technical aspects, it's the human aspects. And what I mean by that is that I typically see three aspects of human character that are important and you should never forget. One is the culture of the company because you want to work with the culture and add to it, not replace it.

8:11Two is biased. People have their own journey with technology. Everybody's got a different skill level. If you asked your listeners today, how did they find your wonderful podcast? Everybody would find it differently. We didn't all go through the same route. And then finally is the beliefs. The belief is one of the most strongest elements of any company or organization. Our belief, we do this for this job. So what I say to people when you're dealing with agents, first of all, the knowledge expert is the most important person because they know the exam question they need to have answered and they know if the answer is correct.

8:45second is invest in the right simulation and also the right evaluation you want to evaluate your agents and simulate your agents but if you do this you can work in a different way and never make it a downwards mandate make it organic strategies from the corporate or the leadership need to be we're investing in this for your future. Then empower people. The stack that we presented, and we're now producing for our customers, allows the right safety rails to be put in place to allow executive leaders and the directional strategies of the company to empower the people that need the tools, because they're going to be the biggest benefit for this.

9:31Absolutely. Would you be able to unpack the stack a little bit in a bit more detail, John, so the listeners are familiar with exactly what you're talking about? Yeah, so there's many elements. I'll call out the most important ones because they're there. I'll start with the bottom. The agentic data cloud is a fundamental building block. We need data for AI. Then you have agentic defense, which allows you to have areas. Example would be model armor. So I know that my models are working in their correct way. Then you work up the stack. And one of the biggest areas of announcement, and as a deep technologist, I was like, wow, this is impressive, is the split in our own TPU processes.

10:11We've got our AI for inference, the execution of AI, and AI for model, the training of AI, because they're different tasks and having that split aspect. Probably for me, the one that I really like is Agent Garden, because it gives me agents that I can already use out of the box. And then having that single surface, Gemini Enterprise, that allows me to interact with my agents, allowing me to interact so today for example i have my gemini age and my gemini enterprise has summarized all my meetings today make sure i have all the references to my briefings and now i'm actually ready to go and do the job i need to do and the other thing that's really important is i'm currently missing emails because i'm talking to your wonderful listeners which is at the moment is my what i want to be famous for is help them on their journey and that means that I need to have people know that I'm out of office.

11:06And the out of offices that I had in the past, I'm out of the office for today. It's not acceptable. Some of these can just be directed to my team colleagues and let them work on it when I'm not there in the office. So my agent's helping me. So these are the ones we're in this world, and that's why the stack is so important. Is that, do you have to have a G, do you have to have G Suite to have access to this stack? Do you know what I mean? No, you can have different experiences. services we've even got gemini enterprise integrate with office 365 so you can do different surfaces but also you can have a developer experience you could have a no code experience you know one of the tools that i've been showcasing a lot today is anti-gravity i love vibing in creating solutions and i do that with anti-gravity so if you're a developer there's a developer experience if you're just a business user there's a business user experience yeah absolutely that's that's amazing.

12:00Thanks for elaborating on that, John. It sounds really exciting. And there's all sorts of questions I have coming to mind, but I know we're a bit pushed on time. But there is one thing I wanted to ask you about. So businesses are evolving quickly as a result of this technology and comparable technology. What skills do you think will become more valuable as organizations increasingly adopt AI and AI agents? What skills do the individuals in those companies need to work gone to keep up with this pace of change. Every year I give a Christmas lecture to students. And one of the things I say is there's three skills that a human has to have in this era.

12:40One, agility. Two, flexibility. Three, creativity. And the reason I say that is when I first started in my career, my skill as an engineer was my forte to make people's dreams come true. now my skill is being a creator of my own ideas and allowing my ability to have ai authored code make my dreams come true the create the key here is the dream that i had the creative moment is my key asset so what i say to people is in your career you're going to have many pivots you know if i go back to when i started in the 80 many of those roles today don't exist this has been the same for human over time you know there was no concept of you know web designer there was no concept of influencer there was no concept of digital marketeer well they've all been born in the cloud era and there's going to be new roles born in the cloud era but if you've got the ability to pivot and be agile and be inquisitive and be a learning mindset and be a creator you can pivot to many industries of your choice yeah absolutely no very well said that's a great answer.

13:50So in terms of the increasingly AI, competitive AI marketplace, okay, you've got a lot of competition popping up everywhere. Google has been investing in AI for many, many years. What do you believe differentiates Google from the competition? The first of all is it's a healthy market of competitors, you know, which I think is really rich for customers sure i suppose one that i take away there's three things that i take away and i don't call it unique by the way i call it a value to the car everything is about what value and outcomes can we give a customer well first of all we have a frontier model from our wonderful team in research and deep mind and for us we're lucky they're based in the uk uh where i live so we have a good understanding why is this important we understand safety training the ethical usage The second bit is the amount of open source that we create.

14:47If you look at the amazing technologies like Kubernetes that allow clustering, we've got the Gemma open models means that we were OK in open sourcing our technology. And then finally, I just I love being at the company that makes me driven to learn. And it's not a technology. It's a mindset. when I joined Google I just found it that I got the urge to learn again and I'm not saying my previous employer wasn't amazing they were amazing I loved being at Oracle for 25 years but it really made me hungry hungry to learn and that learning keeps me excited every day yeah that's a great answer as well I think you're the fourth executive we've had on from Google and they've all said the same thing and I know from in terms of this uh in coming up with your own project and being encouraged to innovate is where Google Maps, Google Wallet, and all of these incredible products were born from that culture, weren't they?

15:45So that's really interesting, John. So let's finish on this one then. So we're an IT consultancy firm, and I speak to CTOs quite a lot, and what I'm seeing is that a lot of companies are stuck in pilot mode. So my question to you is, what separates organizations that are generating real business value from AI and those who are still stuck in limbo and pilot mode? What are the ones that are creating the value? What are they doing right? Following five steps. First step is they have an autonomous agent that can run in its own context with a well-grounded problem to solve. Two, they use a framework that they're not building and they're just consuming in our case that's the stack of ai technology three they understand their cognitive architectures they're trying to build is it an act and use is it problem solving four they have clarity in the evaluation and also how they do grounding and they understand the qualities they need and five they've understood when human in the loop is required absolutely brilliant and on that note john what a what a perfect way to end brilliant it's lovely talking to you, Garel.

16:56Likewise. John, thank you so much. This episode was brought to you by Be Digital. Be Digital support leadership teams to optimize cost and get more out of technology investments. Be Digital and the team have unrivaled expertise with technology license management and data remediation and are therefore perfectly positioned to help prepare organizations for AI technology capability. And on the last point, BDigital have developed a cutting edge AI readiness assessment, which provides tech leaders with a platform they need to make well-informed decisions about AI adoption strategy. Go to BDigitalUK.com to find out more and get in touch.

17:48I love that one. That was a great conversation. Thanks to John Able. He didn't have much time. But needless to say, we covered a lot of ground in just 12 or so minutes. And what particularly stood out was his emphasis, I think, above everything else on the foundations required for successful AI adoption, especially for sort of big, complex enterprise organizations. whilst much of the attention at the event and at the moment is focused on agentic AI, AI agents, and the latest technology around that, the organizations seeing the greatest success quite clearly are the ones getting the basics right at a foundational level with their data and with their IT asset management.

18:31John spoke about the importance of the five pillars, as he called it, which I thought was great. Just to break that down, that's data, governance, security, technology, and of course, people having a structured approach to moving from experimentation to real business outcomes. I thought that was great. Of course, understanding this framework is one thing, but implementing it is another. So next, I sat down with the EMEA Managing Director for AI at Google Cloud, who needless to say, is right at the center of Google's push around agentic AI and enterprise scale deployment. So this one was a real treat, a slightly longer one than last time, but again, packed full of value and insight.

19:11This is Mr. Alex Rutter.

19:16We've had a fabulous day here. We are packed the rafters in terms of clients and partners that are here with us. The level of insight we've got from people has been really fascinating, kind of about many things AI related, as you'd expect. But also just, I think, the sharing of learnings, I think, is my favorite part of these days, which is customers informing other customers how they're using technology in a way that's relevant to their business. And just seeing those creative sparks come to life is just a great way to spend a day, really. Yeah, absolutely. For those listeners who may not be familiar with your role and what you do at Google Cloud, could you give us a little introduction?

19:48Sure. So Alex Rutter, I'm the managing director for Google's AI organization. That is our cloud AI business. And I'm responsible for that operational territory across the UK, Europe, Middle East, Africa, and the Gulf. So quite a significant part of the globe from an operations perspective. and I look after our go-to-market AI specialists, so people that are very deeply focused in AI, both as applications of business and this utilization. We have a deep customer engineering function that looks at kind of technical capability for pre-sales, and we have our four deployed engineering teams as well that all ultimately roll up into what is our bad AI business globally.

20:26Wow, you've got a lot to keep you busy then. Definitely. So a lot going on, massive announcements today. What are the biggest takeaways from the summit this year, Alex, that you want people to remember when they leave at the end of the day? End of the day. Well, I think there's a couple of us. I think the first of all, the first one primarily has to be from a technology footprint is that we have made an announcement today that Gemini 3.5 Flash will be localized to the UK for both processing, for data residency and for machine learning, which is a really strong announcement for the UK economy as a whole, because we'll have that model running locally here, which is great for sovereignty and for those organizations.

21:09We've got a great increased levels of data sensitivity. That's a really strong statement for us. The other bits that are really important that we found is, you know, we have got a significant investment going into UK infrastructure, five billion pounds going into building a brand new infrastructure data center of Waltham Cross, which will bring next generation services to date. across the UK, again, serving the UK market from inside the UK territory. They were the two that were massive takeaways. I think we are so close in proximity to Google I.O. There's probably another couple of ones. If I think back to my own keynote this morning, the other couple of things that people want to leave here thinking about.

21:47The first one is Gemini Spark, our first personal assistant agent that runs 24 by 7. It is optimized for long-running tasks, directly integrated into the Gemini interface. um i've been using it for the last week and i'm super impressed by it the other one which is gemini omni so you'll be familiar with nano banana the brilliant image editing capability that came out last year gemini omni is that but for video so the ability to create anything from anything so we've seen some amazing initial creative from uh influencers and tech influencers around the world that are playing with in a pre-ga fashion so i'm really excited to see what people do with that when that finally goes live on the market.

22:31When does that go live, Alex? Do you know? So we haven't got a definitive date for it at this point in time graph. We're working through a GA probably sometime in Q3, but we're not there on that yet. But we can follow up with further details when we know more closely to the time. Absolutely. Well, one of the dominant themes from the summit was agendic AI, of course. And the industry is obviously shifting beyond AI assistants and chatbots and large language models towards Agendic AI and, you know, sort of AI basically taking action, executing workflows, operating across business processes. Could you talk us through what the technology and the product releases and how this impacts how your customers can basically benefit from Agendic AI?

23:18Of course. So I think the first thing we need to do is need to just break that down into two levels. So when you think about Google, and for those people that have listened to Chimuty before, and they know a lot about the Google stack, we are the only hyperscaler within the organization, just forget hyperscaler, that effectively enables a fully integrated stack from an AI delivery perspective. That is from the AI hypercomputer at the very foundational level, where we offer our tensor processing units and our access to world-class GPU accelerators. We then have access to our own research and models with Google DeepMind here in London, which is great, especially a big part of the London summit.

23:53We then have our agentic data cloud, which I'd like to start thinking as kind of the context layer for agents. We then got our cyber layer, which effectively looks at the new acquisition by Wiz that brings into it multi-cloud stability and multi-cloud sensory behavior for cyber. We then have our platforms that sit on top. And then we have our agents or our agentic solutions that eventually accelerate business values. That's kind of the stack as we see it. When it comes to delivering value, it comes in two key areas. The first one is around the platform. So Gemini Enterprise Agent Platform. And this is the governance layer.

24:27This is across an organization that enables you to build, govern, scale, and optimize across the business. And we made a few announcements today, or we reiterated a few announcements, I should say, that came from I.O. that are really important for people to think about when they think about deploying AI agents across the business. So when we think about build, we're fundamentally rewriting the way that software is created today. We now use agents and agency capability to govern, manage, deploy, and iterate on software. And as part of our announcements at I.O., which are directly integrated into this platform, we announced Antigravity 2.0.

25:05So the first fully agentic IDE environment that customers can build, code, create in. but crucially we also rationalized our harnesses down to one single harness across all three surfaces whether that is ai studio whether it is gemini anti-gravity 2.0 from the desktop environment or through the cli environment all in one place to provide a a easy to create natural progression for low code agents natural languages all the way to full code all in the same place the second area is around what i would call agent orchestration the ability for agents or teams of agents to be able to be seamlessly able to pass tasks and workflows between agents inside the dynamic process called agent-agent orchestration.

25:52Probably the most important ones we've had the most conversation today is around the governance of AI agents. And two particular things in importance. One is agent identity. So think of that like a cryptographic ID or a MAC address if you're old school enough to think about something unique for each agent that effectively allows you to trace, track, understand what that agent is doing, why it's doing it, and where it's going, just to make sure that you've got that level of governance inside the organization. Secondarily is the agent gateway. Not all agents will run inside GCP. Not agents will run inside every cloud.

26:25So being able to manage, govern, and optimize third-party services inside one place is crucial. And then finally, at the optimized layer, in order to optimize on a platform, you need to have the greatest amounts of observability. And we have built an observability entity with inside the platform that effectively has hotel compliant telemetry to look at all the aspects of agent behavior across the business, whether they are inside Google or outside the Google Cloud environment. So that's the platform. That platform serves a variety of models, notably first party services for Gemini and for Gemma, our open source model, as well as the Anthropic models, all from with inside GCP.

27:03And so we're seeing that happen quite extensively. When you then look at that on top yeah you have the gemini enterprise app and the app is the front door to ai inside an organization it provides a holistic environment where we are getting uh a platform for people to connect to first-party data services like google bigquery or google or looker or our new data lake house it connects to third-party applications like slack jira confluence and probably most notably the Microsoft 365 suite, kind of putting Google AI on top of that legacy productivity suite. And then the ability to build agents on top of that in a safe, governed way, crucially with quotas, fine-grained cost controls.

27:47Really, that's kind of the way that we're helping businesses bring agentic capability to the organizations they're working for. So one thing you mentioned there, which I think is probably one of the main barriers for large enterprise organizations, certainly is obviously governance and getting the right governance in place. So as AI agents are given greater autonomy and access to enterprise systems, where should organizations draw the line between automation and human oversight? Can you just give us your thoughts on how organizations can govern this so they can roll it out quicker? So I think the first thing is we need to have, you know, every organization, this is, first of all, we should be good.