What separates an impressive agentic AI demonstration from a deployment that produces measurable business value across an entire company?
In this episode, I speak with Frank Theisen, Vice President of IBM Technology across Europe, the Middle East and Africa, about how businesses can move AI agents beyond isolated pilots and into the processes where work actually happens.
Frank believes the conversation has changed considerably. Most large companies are deploying some form of AI, yet many still struggle to demonstrate a significant commercial return. The difference comes from connecting AI with end-to-end business processes rather than creating another assistant that sits outside the systems employees use every day.
IBM has attempted to prove this internally through its "client zero" approach, using its own technology across human resources, IT, procurement, sales and software development before taking those practices to customers. The company reports that AI, automation and hybrid cloud have contributed to $4.5 billion in productivity gains over three years.
Frank explains how IBM's AskHR service handles common employee inquiries and helps managers complete administrative tasks without learning how to operate several separate enterprise applications. IBM reports that AI now resolves 94 percent of common HR requests automatically, while similar work is taking place across IT support and procurement.
The discussion then turns to orchestration. As companies acquire agents from multiple software providers, the problem becomes far larger than creating individual assistants. Businesses need to understand how agents communicate, which systems they can access, what identities they use and who remains accountable for their actions.
Frank expects the number of applications, agents and non-human identities to grow rapidly. Without orchestration and governance, companies risk recreating the same application sprawl they have spent years attempting to reduce, this time with software capable of making decisions and generating additional code.
Data presents another barrier. Publicly trained models rarely contain the proprietary information that gives a company its commercial advantage. That information remains distributed across databases, applications, mainframes and cloud services. Frank argues that enterprises need a governed, federated way to bring AI to their data without repeatedly copying everything into another repository.
We also discuss digital sovereignty across Europe and the Middle East. Frank describes sovereignty as a matter of control across data, operations and technology. Companies need to decide which workloads require isolation, which regulations apply and where dependence on one provider could limit their future choices.
Wimbledon provides a timely example of these principles in practice. IBM Bob helped modernize the tournament's digital platform by mapping and migrating approximately 15,000 articles, videos, photographs and related metadata. IBM says work that would traditionally require four or five specialists over several months was completed by one engineer within four weeks, with the assets themselves extracted in 47 minutes.
Frank closes with three practical priorities. Understand where AI could affect the business, determine how successful use cases can be automated across complete processes, then address security, governance and provider dependence before expanding them.
If your company already has dozens of AI pilots, should the next investment create another agent or coordinate the ones you already have? Listen to the episode and share your thoughts with me.
[00:00:04] - [Speaker 0]
What separates the companies making billions in productivity gains from AI? From those that are famously still stuck in pilot purgatory, just running pilots and impressive demos. Well, as businesses rush to deploy agents, the challenge is no longer getting access to AI. It's connecting those systems to trusted enterprise data and governing exactly what those agents can do, and also making thousands of agents working together without creating even more complexity. Well, my guest today leads IBM's technology business across EMEA, and he joins me to share what IBM has learned from its own AI transformation.
[00:00:49] - [Speaker 0]
And we'll also talk about why orchestration matters more than the number of agents that you deploy. And also, what could the Wimbledon tennis tournament possibly teach every single business about putting AI to work in the real world. Intrigued? I hope you are. We've got a good conversation coming your way today.
[00:01:10] - [Speaker 0]
So enough for me. Let me introduce you to my guest now. So a massive thank you for joining me on the podcast today, Frank. Can you tell everyone listening a little about who you are and what you do?
[00:01:24] - [Speaker 1]
So my name is Frank Tyson, and I'm responsible for the IBM EMEA technology business. What does this mean? So IBM has kind of, three big parts. One is IBM Consulting. That's our consulting business.
[00:01:38] - [Speaker 1]
Then we have technology, which is all about our products both on the hardware, but specifically on the software side alongside this product services. And then we have research and development. We're really taking care about the new things like quantum. And myself, I'm more than thirty years in IT industry. Started off in r and d, ten years in development in Birblingen in Silicon Valley before I then went to the client side of the house, really serving clients in IT.
[00:02:10] - [Speaker 1]
And my red line is data, data and AI. I'm doing for more than twenty five years. And now since beginning of '24, I'm heading the EMEA technology business.
[00:02:23] - [Speaker 0]
Well, thank you so much for joining me. There's so much I wanna talk with you about because we're recording this in the summer of twenty twenty six where things are a little quiet now. The first half of the year, first six months, I went to 16 different, tech events from Egypt to Vegas, and predictably, every single one of them was all about AgenTek AI. And as a result, every organization seems to be talking about this now. But there is a big difference between running a successful pilot, a shiny demo at a tech conference, and creating business value at scale.
[00:02:57] - [Speaker 0]
So I've got to ask you. From everything that you're seeing here, what is it that's separating organizations that are making real progress from those that are still struggling to keep up? What are you what are you noticing there?
[00:03:08] - [Speaker 1]
Yeah. And then you made a good point. So it's got a bit quiet. Yes. But on the AI side, I think you saw over the last months is huge evolutions in all areas.
[00:03:19] - [Speaker 1]
And also how the outside world, the analysts are kind of looking at the at the companies and corporations. So it's now really about putting AI into action into the processes versus doing the demos, doing the pilots, doing the trials. And and that is exactly what we see. So if you look at the analysts, what they are they are seeing. So 90% of the enterprises, they are really deploy AI, but 94% are seeing not that significant value.
[00:03:51] - [Speaker 1]
So why is this? I I think it's exactly about this. Those ones who are bringing AI into their process with a scale for the change, they really benefit the most, and they see also then the benefits in all terms of cultural shift, cost, productivity. And I can give you a few example if you like.
[00:04:15] - [Speaker 0]
Yes. Please. Go for it.
[00:04:17] - [Speaker 1]
So so very good. So for example, I'm working with one of the larger European banks, and what they are already doing is they're providing all their their banking advisers with help of the agentic AI capabilities so that the meetings they have with their clients are automatically prepared. So with all the knowledge, all the expertise, so that creates a lot of productivity and quality then in the client engagements they are doing. Or what they are also doing is we all have the in in financial services, the instant payment regulation now since last year. That means we all, as private persons, must have the capability to do an instant banking via transfer to to another person.
[00:05:04] - [Speaker 1]
So what does that mean for the banks? Their fraud risk is growing significantly because now they cannot do the processes outside. So and their bank decided, okay. Where to do it? So they did do it in the transaction on the mainframe with the AI capabilities that the mainframe now provides to put a real seamless process for doing instant payment and for detection at once.
[00:05:31] - [Speaker 1]
So those are the kinds of thinking which are really making the difference in terms of piloting or really end to end processing of the core business.
[00:05:41] - [Speaker 0]
And as a result of everything you've mentioned now, I think there are many business leaders that are increasingly under pressure or certainly feel under pressure to justify AI investment much sooner than they expected, let's say, two or three years ago when things were a little bit more relaxed and there was the gold rush, AI gold rush and AI everywhere, and things are getting a little bit more sensible now. So where are you seeing the the fastest and most measurable returns today? And and what characteristics do these successful projects have in common? Are there any trends here?
[00:06:13] - [Speaker 1]
Mhmm. So so I I take IBM as an example. So Yeah. Started off in 2023, mid twenty twenty c to say we want to be the client zero in adopting AI and also adopting our own technology, our own products. So we started that, and and and really one of the key things is it started from the top with the CEO and all the kind of functional leaders.
[00:06:39] - [Speaker 1]
So then then we put emphasis on where is the lower hanging fruit to drive this transformation, and this is about all the functions like sales, procurement, IT, HR. So when we started with HR, for example. So HR was traditional until that, and then we changed to to make HR really a completely agentic 24 by seven HR for employees and managers. So what does it mean today for me as a manager? If I want to to move a person to a different manager, so I have an agentic chatbot who understands me and say, want to move Suzanne to Richard August 1 this year.
[00:07:23] - [Speaker 1]
Please do that for me. So what happens underneath? So the the agents are looking into the CRM system. Am I allowed to do this? Do they I have the the the kind of right rights to do this?
[00:07:36] - [Speaker 1]
So is that person existing? So they are going into SAP, into CRM, into Salesforce, bringing all together. And if there's an issue, the chatbot is coming back and asking me further questions until everything is fine, and then the agents are doing the transfer. Completely autonomous. I don't need to know how to deal with an SAP kind of thing or an an HR kind of application we have.
[00:08:03] - [Speaker 1]
So so these are the levels which drives huge productivity. And on the other side, very high customer set also internally. So now 94% of our HR processes are running autonomously with agents as an example. Then we went into IT, did the same for Ask IT. If you have problems with your with your IT internally, we went to procurement, getting to really 90% touchless procurement running through Agenty AI in the process.
[00:08:37] - [Speaker 1]
And now we are going into sales, helping the sales, the sellers really to to have more insights into what they have to do day by day. And with that, we were until end of twenty twenty five, and we do thorough measurements around this, we saved 4.5 billions in really productivity gains at IBM, which can be reinvested then for research and development for all the things that are really important going forward.
[00:09:07] - [Speaker 0]
Wow. That is a big stat there. That if anybody listening on and wanting to find out more about ROI, I think you certainly answered that question. And an area that I think you're also passionate about is orchestration rather than simply deploying more AI agents, which is incredibly refreshing to hear. So for people listening, can you explain what good orchestration looks like in practice and and why it has become such an important part of enterprise AI now?
[00:09:34] - [Speaker 1]
Yeah. Let me start with the latter. So why why why test this important? So so the analysts say that by 2028, we have 1,000,000,000 AI apps. So you can imagine what that means.
[00:09:47] - [Speaker 1]
So if you look back how long it took Apple to do and create this environment with the apps in their environment, it was millions in years. You know? Now we talk billion only two years away. And and that means we have 2,000,000,000 of agents by 2029. So and and there is a lot of code which will be generated by the agents itself.
[00:10:13] - [Speaker 1]
And on the other side, 40%, we expect the code have security vulnerabilities and think about mules and everything. So so that's the importance on on why is it so important to have an AI orchestration layer who can deal with that as a company. Another example. So currently, if you do identity management, it's mostly that humans are kind of interacting. In future, you will have 60 x nonhuman identities.
[00:10:41] - [Speaker 1]
So you have to make sure you have the right governance, you have the right integration, and you have the orchestration around all those agents. Because every application, every vendor is coming up with their own agents, so how you make them seamlessly talk to each other so that you can scale and not die in complexity. So that is a that is a key thing. And, IBM has we have our own environment with what's next orchestrate, which is exactly on that space. But I give you again another example, for for, everybody who listened to the post podcast, which is software development life cycle, which plays a lot into that agentic AI.
[00:11:22] - [Speaker 1]
So we started mid last year with our own new development companion. So we named it Bob, and Bob is now also available since a few weeks as a real product. And what is Bob doing? Bob is helping the developers in all series of the development like cycle, understanding code, transforming code, do blending, do architecture stuff, working together in groups. So so it's kind of a real like an agent who is your kind of teammate, but he's super intelligent.
[00:11:57] - [Speaker 1]
So and now you can can think about the companies who are also having their application developers and try to improve productivity and have all the orchestration of those agent included. Because if you look at vulnerabilities, so Bob will look at that your code is safe from everything what is there. So those are those is an example. So why is orchestration important and how we do it?
[00:12:23] - [Speaker 0]
Love it. And we will have people listening inside organizations where data quality, governance, and disconnected systems continue to slow down their AI initiatives. So when you first begin working with an an organization, I'm curious. What are the the first warning signs that tell you that they might need to strengthen those foundations be before simply adding more AI on top?
[00:12:47] - [Speaker 1]
Yes. It's a very good point. So so we we we usually talk about the great capabilities of all those new models in every flavors. But, yes, only 1% of the enterprise data is in those GenAI models. So that means 99% of the core value of a corporation is not in the model.
[00:13:08] - [Speaker 1]
So so how do you get this data, which is usually very fragmented? It's it's kind of not potentially ready for AI into shape that you can really use it in in the AI area. So it starts with the simple task, make your data ready for AI. And there you have to ask questions like, so how can I do that? Have do I have to duplicate all this data many, many times, for sure, with the rising storage costs?
[00:13:39] - [Speaker 1]
So that is not such a good option. But the other thing is how about speed and real time data? So the analysts are saying in 2028, 60% of the data which flows into the agents will be real time. So how do you create a kind of a federated real time data environment which really make sure that the AI comes to the data and you do not duplicate everything and to have a real time momentum to bring the data from all of your different processes in a governed and clear way to the process you want to drive with your agents. So that is a a key thing.
[00:14:19] - [Speaker 1]
So look at the data, your corporate data, how do you apply to the AI, make sure you have an an integration and, real time runtime environment and taking care about all the governance and security.
[00:14:34] - [Speaker 0]
Are you still dealing with agentic AI semantic confusion? Well, Denodo keeps your business harmonized. And by positioning your data platform into a universal semantic layer, everybody can be on the same page. So get back to scaling your business and trusting your AI agents to create proper business decisions, and you can do that by visiting denodo.com to see how. But now, back to today's guest.
[00:15:04] - [Speaker 0]
And something else that I'm hearing a lot on the show floor at multiple tech conferences around the world is this ongoing debate around digital sovereignty, and that's happening more and more particularly across Europe and The Middle East right now. And I'm curious from what you're seeing and hearing there. Are there any misunderstandings that you encounter most often? And and how can organizations better regain control over their data and operations without limiting innovation?
[00:15:31] - [Speaker 1]
Yeah. You hear a lot about sovereignty washing. So everything has to be sovereign. So so what is it really about? It's not about location.
[00:15:42] - [Speaker 1]
So I think sovereignty is about control. So and and and how you drive the right sovereignty from a data sovereignty, from an operational sovereignty, and from a technology sovereignty. And and there are different costs aligned to those different kind of sovereignty pieces, and you might not need for every single complete sovereign environment. But I think there are some core principles which help the clients. So first of all, so you you should have if you if you talk really about highly secure data, you want not to keep on a public cloud, then you have to have an air gapped environment for that.
[00:16:21] - [Speaker 1]
So how how do you drive really air gap the sovereignty? But it should be aligned to transparency and optionality at the end. So you do not want to log in because if you log in into into one angle, then you might have either high costs or really complexity to get out. So I think look at the sovereignty from those three motions. So it's data.
[00:16:47] - [Speaker 1]
It's how you drive the operational environment and how you drive the technology. So I'll give you an example from us. So we announced, I think, in Boston, IBM sovereign core. So that is really an open solution based framework which addresses all the capabilities. So first of all, it sits on top of Red Hat OpenShift.
[00:17:09] - [Speaker 1]
Why is that important, to name to name Red Hat here? Because, it provides you this open source based capability, and that's a key differentiator because whatever happens, you are still in charge onto that. So the next thing is you want to have multitenancy capabilities. You want to have security capabilities, but you also want to have compliance. If you look at Europe but also The Middle East, you have a lot of regulations.
[00:17:38] - [Speaker 1]
So we have built in those regulations like EU AI, GDPR, Dora, those kind of things so that you can do your audit trails directly out of the sovereign capabilities. And then you have your your your kind of sovereign data services as a catalog, and you can think of even provide AI services curated to the tenants. And it's like a GPU as a service. It's only a use case, a data service, or AI service, which will then be be transferred. So those principles, I think, about sovereignty are important to understand to decrease dependency.
[00:18:21] - [Speaker 1]
I think that is the key thing for for the clients to see where we are defend dependent on and what do we want to control and, how can we achieve this.
[00:18:33] - [Speaker 0]
And if we zoom out for a moment, I think IBM's own transformation has generated billions of dollars in productivity gains through AI and automation, etcetera. And for for people listening here, if if you look back at that journey, what lessons surprised you the most? And which of those lessons might help listeners from organizations in any industry just apply immediately? Any quick wins there? Anything that you'd share?
[00:18:58] - [Speaker 1]
Yeah. So at the end, the first key thing is to drive the behavioral and cultural change within an organization and the workforce really there. And and that has to be driven both top down. So on our side, really, our CEO, Arvind Krishna, was all about this implementing this, but also leverage this bottom up. So you have to do the right enablement.
[00:19:23] - [Speaker 1]
And if you do some, for example, process transformation, you create catalysts, who are then scaling across the organization. So we did an HR after we have done our huge ask HR transformation. We have now new HR consultants out of this team doing that for other clients. And, that is the kind of thing. So we create also new roles and new capabilities and new expertise so that the transformation becomes really something which is seen by the whole workforce as super important and also a kind of a career development task.
[00:20:01] - [Speaker 1]
And then you align it with, for example, gamification. We do this what's next challenge every year. Now we are just starting with Bob so that everybody gets hands on on these agents so that you get the team around this and the new capabilities.
[00:20:17] - [Speaker 0]
And as we record this episode today, not only is there a World Cup going on, but there's also the Wimbledon tennis championship entering its second week. And one of the reasons I wanted to raise that today is I think the Wimbledon tennis there, it provides a fascinating real world example of AI that is operating under intense public scrutiny because there are millions of fans expecting fast, most importantly, accurate information throughout the entire tournament. So, when we look at that like this sporting event, what can enterprise leaders learn from the way IBM is applying AI and and data there that that could adopt within their own organizations too.
[00:20:58] - [Speaker 1]
Yes. So first of all, it's for sure a fantastic event, and and IBM is partnering with Wimbledon for more than thirty six years now, on on world class experiences on the championships, and and and getting the fans really, very active into into what's happening. So I give you again an example here, how we how we evolve this going forward. So so now you get a lot of digital assets, articles, videos, photographs, everything. And and how to how do you get this into this system where the fans can have access, where where we get the value out of it?
[00:21:36] - [Speaker 1]
So up to now, this was a more traditional process and and would really take months just to get this done. So with Bob, we were really able with one single engineer in four weeks to get 15,000 assets really extracted in only minutes to to bring that into the system and now have a much, much stronger system and capabilities. By the way, we are doing the same, you know, for Ferrari as a big partner, and there we have also 400 millions of TFOZI. You want to become better part of, you know, being there. And if you can imagine 400,000,000 TFOSY, how many can go to a track?
[00:22:20] - [Speaker 1]
Very, very few. So you have to engage all of them, And and I think we are very strong in those capabilities to increase kind of fan satisfaction to bring really real time capabilities to all of them.
[00:22:35] - [Speaker 0]
And I always try and give everybody listening a few valuable takeaways. So finally, for the people listening today who maybe want to move just beyond AI experimentation, are there any practical actions that you would recommend that they take over the next six to twelve months to to better put themselves in a stronger position for long term success? I appreciate it's a big ask, but anything that you would pass on to those people listening to get them up and running?
[00:23:01] - [Speaker 1]
Yeah. Let's go over these three things.
[00:23:03] - [Speaker 0]
Okay.
[00:23:03] - [Speaker 1]
Well, first, understand what's happening in the business, what are the risks, and what AI would and could mean for them. Second is if you have kind of made up your mind there, how would you automate it and then scale across the processes and the enterprise? And the third one is secure. Understand the risks around the deployment, how is it governed, and how you create really the optionality or you keep the optionality where to put it in place so that you are not getting dependent.
[00:23:39] - [Speaker 0]
Fantastic advice. And for anybody listening that would like to dig a little bit deeper on anything we talked about today and find out more about how IBM are working, how they might be able to help or connect with you or your team. Where would you like me to point everyone?
[00:23:53] - [Speaker 1]
Yeah. So first is easy. Visit ibm.com. You will find everything there. If you want to listen more, I think, my LinkedIn account, you can look for.
[00:24:03] - [Speaker 1]
And, there are a lot more about, very highly expert IBMers, and for sure in YouTube, we are also available.
[00:24:11] - [Speaker 0]
Well, we covered so much today from how businesses can win with AgenTik AI. Yes. There's a lot of excitement about the promise of AI agents to transform businesses, but for many people listening, there are many challenges around unready data, sprawl of apps, agents, and assistance from so many different providers that could potentially create new risk. But hearing your story today, how you've overcome this, the kind of ROI that you've unlocked here, and the difference you're making with people alongside AI is incredible. So I will add links to everything you mentioned, including your LinkedIn.
[00:24:46] - [Speaker 0]
But, more than anything, just thank you for taking the time to stick down with me today and sharing your story. Appreciate your time.
[00:24:52] - [Speaker 1]
Thank you very much, Neil. It was a pleasure to talk to you.
[00:24:55] - [Speaker 0]
I think today's conversation perfectly showed just how quickly the AI debate is moving from experimentation to execution. And Frank shared how IBM generated $4,500,000,000 in productivity gains and how 94% of its own HR processes now run autonomously with agents. And while the next challenge will be governing billions of agents, applications, and nonhuman identities, and doing so without losing control. And also from digital sovereignty and enterprise data to software development at Wimbledon. I think the message was remarkably consistent today.
[00:25:38] - [Speaker 0]
Start with the business problem, automate what makes sense, scale what works, and secure everything along the way. Sounds simple. Right? But love to hear your thoughts. Is your company building AI in the way that work actually gets done, or are there still too many projects struggling to move beyond pilot stage?
[00:25:59] - [Speaker 0]
Let me know. If you wanna share your story, let, contact me over at tech talks network dot com. You can connect with me on socials or just have a listen to one of the 4,000 other interviews we've got over there. Whatever it is, please let me know. Send me a message.
[00:26:14] - [Speaker 0]
I'll get back to you, and I'll also be back in your podcast feeds tomorrow morning. Thanks for listening as always. Bye for now.