Turning AI Investment Into Measurable Business Value With HP

[00:00:04] What happens when the excitement around AI meets the monthly bill, the device refresh plan and the reality of everyday work? Well, today I'm going to be joined by Neil Sawyer and he manages HP's Europe, Middle East and Africa business. And today we're going to discuss how companies can turn AI investment into measurable business value.

[00:00:30] Because together we're going to look at why infrastructure capacity alone cannot deliver an outcome. And how leaders can identify the workflows where AI genuinely removes friction. And why productivity should mean better work rather than simply filling every minute with another task. And my guests will also explain why AI costs belong in the boardroom now. How cloud and on-device processing can work together.

[00:00:58] And why employee personas should influence device decisions. That's right, we've got a lot to talk about. There are two Niels in today's conversation. But thankfully, only one cloud bill. So, enough from me. Let's get started and let me introduce you to my guest. So, thank you for joining me on the show today. Can you tell everyone listening a little about who you are and what you do? Yeah, hi Neil.

[00:01:27] My name's also Neil, Neil Sawyer. And I work for Hewlett-Packard. I manage the Europe, Middle East and African region for our business that represents around about a third of the company's turnover. We work with hundreds of thousands of large corporates and many millions of consumers in every country around the world. Well, it's a pleasure to have you join me today.

[00:01:54] One of the questions I've got to ask, non-tech related, as a fellow Neil. Throughout your life, has anyone ever called you Ian? Does this just happen to me or has it happened to you too? Well, Ian would be a polite alternative name for myself. But I did do a bit of research on your podcast. And I did actually ask AI. I said, how many Niels have appeared on Neil's podcast? So, I think I might be the fourth or the fifth, according to ChatGPT.

[00:02:22] I don't know how many Ians that thought they were Neil as well. But that's a whole other rabbit hole to go down. But obviously, today we're here to talk about tech. And I know you spend a lot of time with organisations across Europe. So, where are you seeing the biggest gap between AI investment and the business value that companies expect to get from it? Because we've heard a lot of talk around this in our news feed, a lot of noise. There was the MIT study a few years ago, which felt slightly exaggerated that 95% were not getting ROI.

[00:02:51] But it certainly got the conversation going. But what are you seeing here? Yeah. And what a very exciting time we are experiencing both in home and, of course, in the workplace. AI is a great opportunity to improve productivity and deliver more equality in terms of how business is done. And lots of new opportunities as well that will come with that.

[00:03:14] And, look, my observations are that there's a lot of focus around investment in AI infrastructure, the building out of data centres, neocloud environments, and, of course, making sure that there is capacity in the system to cope with the scale of queries that are coming through every single day. But this is also about orchestration, about managing cost,

[00:03:41] managing the environmental considerations that come with building that level of infrastructure. And for HP, it's about making sure that we deliver value for money to our customers. And ensuring that with the technology that we're producing, you don't always have to take everything to the data centre. You have the opportunity to orchestrate that AI query from your device.

[00:04:08] And the point that I'm getting to is making sure that it can be applied in use cases. And we take the healthcare industry or education, research, science, as well as financial services, ensuring that the AI infrastructure can be applied in an industry-specific way that's going to deliver better, more affordable and efficient outcomes for those using

[00:04:34] and those benefiting from the productivity or output from the AI queries. Yeah. And I think here in 2026, most businesses, or at least many businesses, have started to move beyond pilots. But some have found that scaling AI is proving harder than just experimenting with it. So what is it that tends to change when AI moves from that small group of users into everyday workflows right across an organisation?

[00:05:01] Have you seen anything here that could be pointing to the problem? I think businesses are conscious of the fact that with AI scaling in an informal way in their business, how do you manage that? And there is some regulatory compliance and there's some legal considerations around things like intellectual property.

[00:05:25] And any company that is producing anything or documenting or providing a service to a group of people needs to make sure that what they're producing, either through human input or AI input, has the legal and regulatory compliance associated to it. The other aspect is, of course, making sure that costs are controlled within a business.

[00:05:53] You absolutely want to ensure that your business remains viable whilst at the same time investing in AI. And so my point earlier around orchestration is about ensuring that businesses consider HP's proposition, which is where you can produce a lot of the AI that you would traditionally pay for in a data centre

[00:06:17] and actually utilise the AI technology on the devices that will avoid the cost of tokens and token swell and token inflation that I think we will see over the coming months and next couple of years. So this is an exciting opportunity and we shouldn't allow that to be missed. And I think any company that isn't investing in AI and empowering their people to utilise it

[00:06:45] will slow down in terms of their own progress, but making sure that it's managed and orchestrated in a way that's viable and affordable and legally considerate of what's required are all important features for any business, owner, CEO or financial or IT-centred employee. And some will say there's a danger that productivity becomes shorthand

[00:07:13] for simply ask employees to do more work faster. And on that side of things, how should leaders think about using AI to remove friction? Yes. And create better work rather than just increasing output. And one of the reasons I ask that I'm a business of one and I've gone from, I've got all the friction out. I work so much quicker, but am I going to the cinema on a Wednesday afternoon with all this free time? I have no, I'm increasing my output and doing more. So I'm very aware that there's a balance needs to be achieved here.

[00:07:42] Well, in a flexible working world, Neil, if you go to the cinema on a Wednesday afternoon, nobody's going to judge you. So feel free to do that. But let's put this into perspective around productivity. And I have thought about this a little bit over the last few weeks. If we were living in the Victorian era, when industry and scale was starting to appear,

[00:08:10] then productivity and output has always been quite a significant topic during those times, because it was around the time of the Industrial Revolution. So I'm sure there were people going, well, I'm just being asked to do more and produce more and be more efficient. And during that time and over the hundred plus years since that era,

[00:08:36] what we've seen is efficiency, effectiveness, quality improving over and over again. And I think with the revolution of AI that we're seeing in the workplace, yes, we're going to ask people to technically produce more and be more efficient. But I think what we will definitely see is the time saving in traditionally mundane technology tasks,

[00:09:05] allowing people to have more capacity to deliver more value and creativity for their company or the businesses that they own. So it gives them time to develop whilst at the same time administrate the business that they're doing today. And that gives the opportunity to grow the economy, to grow the business, you know, to innovate and think about different things, because the world is changing. There are opportunities out there.

[00:09:32] But giving people the capacity to think about it and be more creative, I think is one of the most exciting areas that AI can help address and deliver. Yeah, 100% agree. And there's an old mantra in IT that you can only improve what you measure. We do hear a lot about AI adoption numbers, but adoption, of course, doesn't necessarily equal success. So what should organisations actually be measuring to understand whether AI is improving productivity?

[00:10:02] Employee experience and overall business performance? What should they be measuring here? Yeah, well, that's a really good question. And if I refer to the point where AI adoption is happening, whether businesses, business owners and those that run and manage organisations like it or not, if I'm transparent, because AI is open and accessible to all.

[00:10:28] And the ability to measure flagrant access, occasional access through web browsers and using Gemini or ChatGPT or Microsoft Copilot or the many other AI platforms that exist, it's difficult to measure what that productivity or output is likely to give to a business.

[00:10:53] So my advice is very clear is make sure that if you are a business, you are constructing the AI toolkit that should be made available to all employees or specific departments or particular use cases. Do it in a structured way with those AI providers and make sure that there are management facilities, software tools that you can get

[00:11:21] that demonstrates the value and the productivity that is coming out of individuals or groups or teams. And a good example of this, HP have a software tool called Workforce Experience Platform. And that tool is there to monitor the performance of the computing technology and what it's doing to deliver supported outputs for people. It doesn't look at what they do.

[00:11:49] It looks at how the technology is providing support and uptime and efficiency for them. And the second point is that with the large language tools like ChatGPT and Clawed and Copilot and others, is that by investing in the business or the enterprise focused platforms, that you get to see that telemetry.

[00:12:14] So you can demonstrate the measurable output rather than just assuming that productivity is happening. It gives you the information and the insights to both manage cost and input, but also see what the benefits are in terms of output. And I think your passion really comes through for the opportunities here and the exciting times are in. And I'm with you 100% on that. And one thing we've got to bring up, though, is AI usage increases.

[00:12:42] We become more reliant on it and use it not only in the workplace but at home. But it's the bill increases too, particularly with token-based services and agentic systems carrying out longer tasks. There's been a lot of noise around this online as well. So are AI costs becoming a boardroom conversation now in what you're seeing and hearing and what are leaders starting to ask? Well, if they're not, they should be. And I say that in a very kind of pointed way,

[00:13:12] not to incite a kind of response. But, you know, if there is a cost to a business and it is a significant and growing cost, that needs to be in the boardroom because it needs to be managed. It needs to be accounted for and invested in appropriately. Now, in terms of the token point, you are absolutely right.

[00:13:36] The rise in cost of token delivery and usage is not going to get any cheaper. And therefore, we need to find ways in order to navigate around that. And the point that I'm getting to is that there are alternatives. And we're seeing many businesses, particularly those that use high algorithmic type models for

[00:14:05] data research or science or coding. They're actually investing in that AI capability in the office place. And what I mean in terms of real-term actions is that by investing in the devices that have the neural processing units within it, which is what an AI chip ultimately is, delivered by NVIDIA, Intel, AMD and others,

[00:14:32] is that by having that capability purchased by a business or by an individual, means that you don't have to divert your queries to a large language data center, which consumes tokens. And if you connect, as you can do, all of those devices together, there's a tool that HP produced called ZBoost or ZBoost, if you are favoring the American language.

[00:15:00] What ZBoost does is it connects all those devices that have got AI power and capability, and it delivers the AI functionality without ever having to consume tokens. So the money you've invested in those devices today can help provide all of that AI computing power tomorrow. And that's the evolution that we're seeing because it helps in terms of cost management, it helps in terms of environmental management,

[00:15:28] but it also makes sure that we manage the sovereignty point that, of course, many people bring up these days because they're concerned about where that data is going and where it will be housed long-term. You don't get those problems when you're querying AI on the devices itself. Wow, that is such a great point to make. And it's worth reminding everyone listening that not every AI workload needs to run in the cloud. So how should maybe business leaders listening,

[00:15:57] how should they decide what belongs in the cloud, what can run on device, and where the balance can make a meaningful difference to the cost that we're talking about here, but also performance, security, and control? How do you see that balance? Yeah, I mean, that's obviously specific to many industries and businesses, and everyone will have a different perspective or opinion on that. But the role that we want to play,

[00:16:26] and I think that our industry needs to play, is to provide those range of options. Because there is no doubt the benefits that everybody will be getting through large language model queries and search, by requesting information that takes a greater perspective from what's in the cloud, is really, really important. And the value that we're all getting out of that is very visible.

[00:16:53] But equally, when you're dealing with private data, sensitive healthcare information, content that is intellectually valuable to your business or your person, then surely that's not information you necessarily want to proliferate out into the open market. And that's the same with any form of data or information or output that a company produces. So just getting the balance right.

[00:17:23] And I think that the final point I will make, and this is a bit of a plug to Microsoft Copilot, but it's a free one. I'm not being paid for this endorsement. But the benefits that I have seen from Copilot, that queries my own work and the email infrastructure and the data footprint that I'm producing in my business enables me to be so much more efficient.

[00:17:51] And that is not querying the cloud. It's querying the inputs and correspondence that I've had within my own company amongst my own network. You don't need to put that into a large language model and query a data center. That's being queried from my device, and it's providing value and efficiency to me. So businesses need to make those decisions for themselves. But I think it is predicated on the fact that, you know,

[00:18:19] some information is more sensitive and more valuable than others. And some of that information is very specific to the business that you're doing today. All of that merits more of a localized AI rather than a large language data center query. Your agents aren't producing accurate answers because they don't have a complete semantic understanding of your data. And Denodo is solving this and solving it through semantic consistency.

[00:18:48] Through semantic consistency, your agents can start making accurate predictions in real time. So see what else Denodo can do by visiting denodo.com to learn more. And what role do you see the device fleet playing in organizations' AI strategy now? And the reason I ask that question is it feels like the days of everyone starting a new job and being given the exact same PC as everyone else in the business, it seems to be disappearing now.

[00:19:18] So are you seeing businesses thinking very carefully about which employees actually need an AI PC or more powerful devices rather than approaching refresh cycles as that old school one size fits all exercise? What are you seeing here? Well, I can assure you, Neil, that there still is that sort of large refresh styled project in many corporations. And I should imagine that will continue over the years to come,

[00:19:44] partly because that's how businesses tend to capitalize and depreciate assets. over straight line periods. So that isn't likely to change in the short term. But what we have seen in many, many companies is the creation of personas, different styles of behavior that warrants different technology requirements and applications.

[00:20:11] And if you're a data scientist in a bank, your technology profile and needs is going to be very different to an office operator, an account manager or a client manager within that same banking environment. And so what we do with many of our clients is that we start out with the personas that feature within those businesses.

[00:20:36] And then we apply a standard feature set to each of those personas. And there could be five in a business. There could be 25 different personas in that same business. But I think just making sure that there is some structure to how you're investing. And that also enables businesses to apply those technologies in a consistent way

[00:21:00] that is relevant and specific to the use case that we see within those organizations or the specific industries within which they operate. The British National Health Service is a good example of that. There's lots of different complexities and use cases. And making sure we apply the right persona to the right people delivers very vital outcomes

[00:21:24] that often are the difference between curing an illness or processing patient care faster and more effectively. So we're very, very conscious about that. And I think the team and the industry do a great job of it overall. And for organizations and people listening that are moving from AI experimentation into wider deployment right now, what are the practical decisions that you think

[00:21:53] that they should be making now to better scale adoption without allowing cost, security, or complexity to run ahead of the value that they can create here? It's probably a question you get a lot, but it'd be great to give people listening a valuable takeaway here. If we go back to the start of this discussion, then I think what every organization, every decision maker or technology designer needs to think about is what is the outcome

[00:22:22] and what is the workflow rather than, gosh, I need to invest in some AI technology. Because if you start with that point, you deliver productivity and value and better outcomes for your business, but also the customers and the industry that you work within. And I think organizations, therefore, need to understand where AI creates value

[00:22:48] and where each workload should run, what it costs and how success will be measured. And they also need clear governance and that's focused around data and privacy that I mentioned earlier on, along with the cybersecurity agenda. And then bringing those decisions together allows organizations to take a hybrid approach to AI and using cloud and the edge that I mentioned, and devices where each makes the most sense.

[00:23:18] And it will enable them then to scale with greater control and measurable business value to go with that as well. We've both said a few times today that it is an exciting time at the moment. So from your work at HP, what excites you at the moment? What makes you want to jump out of bed in the morning and head to the office? And what excites you about where we're heading with this technology? Well, everything makes me jump out of bed in the morning.

[00:23:44] As you might be able to tell, I enjoy the positive side of delivering outcomes for our industry. And collectively, regardless of where you work and what you do, I think we all work hard to make sure that we deliver outcomes and raise the profile of the technology industry. That's what gets me excited in the morning.

[00:24:09] And importantly as well is actually having a vision over the next 5, 10, 15, 20 years and being able to see how this technology is going to evolve and the benefits it will deliver in terms of elderly patient care. We're all living longer. And I think that this type of technology can deliver such material benefits in terms of longer term, longer life health care.

[00:24:39] The effectiveness that's going to have in terms of delivering a quality for supporting children that might have specific needs in education, better quality assistance and guidance in the classroom in a world that's pressured to deliver more higher quality and attentive education on a pupil by pupil basis. And then looking at the opportunities in the home as well

[00:25:06] and how AI is helping to solve great family debates around the dinner table and delivering queries on hand about... I'll give an example at home. My mother-in-law presented this old piece of jewellery the other day and said, well, I've got no idea what this kind of long number is on the jewellery. And, you know, a quick bit of AI and a photograph uploaded

[00:25:34] indicated it was some kind of serial number from a jeweller from yesteryear. And, you know, we're able to put these very simple things to close and it creates a lot of family unity and discussion. And I think all of these great things are just small opportunities and examples, but, you know, opportunities to improve what we're doing, how effective we are. And if it brings a bit of family unity with it as well and debate, all the better.

[00:26:05] Yeah, 100% with you. I also saw recently as a friend that they dropped a very tiny screw on the carpet and then someone pulled out the phone, took a photo of the carpet and said, where is the screw? And he found the screw. It was a great debate around that and such a funny moment. But for everyone listening, I wanted to learn more about all things HP. And a lot of the things we talked about today, where should I be pointing people, especially if they want to connect with you or your team as well? Yeah, look, HP was founded in 1939.

[00:26:34] We created Silicon Valley. And we're very proud that we're coming up to our 90th year of trading over the next couple of years. And, you know, we've been around a long time. We're a very trusted provider into both the corporate space, businesses around the world and homes, as I said. And, you know, for us, if you want to find me on LinkedIn and drop me a line,

[00:27:01] really always happy to respond and make sure that we point any query or future discussion in the right direction. So we're a very visible. We're a very personable organization. We believe in building long-term relationships. And hopefully that's a good reason and a good example why HP has been around for many, many decades and for many, many more decades to come.

[00:27:27] Yeah, and I'm very fortunate that I managed to visit the garage there on 367 Addison Avenue in Paolo Alto a couple of years ago. And it is phenomenal. It's easy to see why we both get excited on this. And you go back to those humble beginnings in, I think it was 1939 or something like that, that garage, and to where we are now and where we're heading. Exciting times indeed. And a big thank you for sharing your story today. Well, thank you indeed, Neil. And great to meet another Neil, as always. So thanks for your time.

[00:27:57] So many great things to come out of this conversation. I love how Neil advised, start with the outcome and the workflow, and then decide where the workload should run, understand what it costs, and then agree how success will be measured. And yes, that does sound straightforward. And it's easy for a company to begin with just a model, a device, or a cloud contract and work backwards in search of a problem.

[00:28:22] But I really appreciate the point that productivity gains should create room for better thinking, creativity, and service, rather than just raising the workload for everyone. So a massive thank you to Neil for joining me today and surviving a conversation with another Neil, without either of us being called Ian as well. So you can learn more about HP and connect with Neil on LinkedIn. All the links will be in the show notes. But over to you.

[00:28:49] How is your organization connecting its AI spending with outcomes that employees and customers can actually recognize? Love to hear your thoughts on this one. TechTalksNetwork.com. That's where you'll find me. 4,000 interviews. Lots of ways of working with me. Or you might even be able to meet with me if you take a look at the events page. But that is it for today. So thank you for listening as always. Bye for now.

[00:29:20] Bye for now.