Unlocking AI’s True Power: Beyond Copilot Basics

00:00:06 Mark Smith
Welcome to the MVP show. My intention is that you listen to the stories of these MVP guests and are inspired to become an MVP and bring value to the world through your skills. If you have not checked it out already, I do a YouTube series called how to become an MVP. The link is in the. Show notes. With that. Let's get on with the show. Hey everyone and welcome back to the Microsoft Innovation Podcast. Today, we're heading to Telford, England to meet someone who's been shaping the Microsoft ecosystem from the inside out. Please welcome Alex, Chief strategist for Microsoft at Softcat and a long stand, Microsoft MVP, Alex. Spent years helping organizations and especially schools get the most out of Microsoft 365. Whether it's streamlining classroom management or rethinking how AI fits into the modern workplace, you'll find them right at the intersection of strategy and innovation. Lately, he's been driving diving deep into Co pilots, real world impact and how governance frameworks needs to evolve to keep up. Fun fact, he wants maps, a version of teams becoming an operating system, literally full links to the are in the show, notes Alex. Before we get into AI and ethics, what's one tech habit you've picked up lately? That's. Even surprised you?
 
00:01:29 Alex Pearce
I'm I'm well. First thank you for having me on the show. Just to say that. So thank you very much. I I guess to answer your question, the, I I love Claude Claude is my preferred AI model to go to and I'm sure people prefer Judge TBT. Some people may well prefer copilot. Both the personal version or the corporate version, but mine is Claude. And to go to that and to be able to ask it not to necessarily just go off and answer a question for you or to build something, but to give it to you as an interactive HTML page. So think about the kind of questions you're asking. Can you ask it to go and use a JS model? Go off and ask it to build you a single HTML page and the interaction and the way you can learn from whatever that content is is so much more powerful than just a prompt or just an output. But that interactive web page that it can go off and build. It's really surprises me and the ability to be able to go off and do that. As an example, I did Microsoft licensing build me a network graph based on the Microsoft licensing between M365 business premium E3 and E5, and you may well see something that's quite.
 
00:02:34 Mark Smith
This is amazing. This this is what makes the like I'd happily hang up the call from here because it's tips like this that I find. So powerful. So let me just unpack what I understood you just said. When you do your prompt, you ask it to. If you're going to learn a new subject or wanting to think about something, you're asking it to create an interactive. Web page that you can then learn the topic or the concept on. Is that right?
 
00:03:05 Alex Pearce
Yeah, absolutely. So there's a a YouTuber that I watch called Where's Roth slight tongue Twister for me there, but he's absolutely brilliant because he does a lot of testing and teaches me. All sorts of stuff I think you can go off and do from a coding point of view. Now I don't mean Visual Studio code and canvas and all that kind of stuff. This is directly within the AI model. So yeah, you can go off and you can say build me a HTML file on that sub. Yet do me a one of them. I played with recently. Do me an OEM, so a devices, so Dell Surface HP do win in a report and put that into a set of graphs and put it into a HTML page. I can then save that as a PDF file and I can send that internally to wherever I need to be able to go off and do it. One of the ones that. Wes did he asked it to build out an Ant commonly, and how it would find a piece of food. And all of a sudden you can see how ants wear. They go off and they scatter. One of them finds a bit of food and gradually over time you get to see how the ants go off and start collectively and working to build that food and to bring it back to the nest. And then he said, well, go off and add a make it 2 * 5. Times quicker and add more. Ants had less ants and that's kind of the entire web page. But think of that as a resource to be able to teach kids. To go off and build things, but it's not about I. I did an article. Interestingly, on my newsletter on LinkedIn about should we be learning more about web standards as part of prompting? Because think about the outputs that you may well get from ideas that you've just got now. So getting to track GBT, getting to Claude the reason why I like Claude as well though is cause got these artifacts features  So the artifact is when it writes the code. You can then go off and you can view it within cloud rather than having. To take it to an external web page.
 
00:04:55 Mark Smith
I like it. I like it. I've got subscriptions that go across most of the main tooling out there. Definitely I have one for Claude and I use a GPT, a lot of and of course, copilot. And I'm even using crock a bit lately. I do find each one has its unique place. I suppose the one I'm least deep in is Gemini.
 
00:05:18 Alex Pearce
Do you know what I know exactly the same, but I actually want to get. Into. It it's one of the leaders out there, I want to know more about it and I I guess time when we we we've both already chatted before this. We've got young kids and certainly the young kids take away from the time we've been able to do a bit more innovation and be able to go off and do that bit more deeper. Research into things like Gemini and what what Google are doing, but what Google are doing are making me think maybe I shouldn't have an iPhone anymore because the AI and what they're pushing. Makes me go. Actually, I'm not quite excited by that. Wherever I see nothing from Apple point of view. 
 
00:05:52 Mark Smith
Ohh, totally right. What a? Let down apples been in the last 12 months. I usually upgrade. I'm one of those people that upgrade every year and I've not upgraded because I'm like I can't see any benefits to upgrade. There's no net new. You know, for me it was always the camera it's you know brilliant tool for video and. And images, but definitely let down now that nowadays with that Apple.
 
00:06:17 Alex Pearce
Indeed, Apple have got big conference next week. I'm expecting a lot, otherwise they're gonna lose a customer. I'm afraid I probably still have my iPad, but. There's some things I can't get rid of, but the AI.
 
00:06:29 Mark Smith
Rate I use that listen nowadays though iPad even.
 
00:06:32 Alex Pearce
Yeah, no, I agree. I use it more for styling and maybe if I'm doing some cooking, got YouTube playing or or something like that. And so it's not there necessarily from a an AI point of view, it's there to consume content more than necessarily anything else.

00:06:47 Mark Smith
Tell me about AI productivity and I I find the word productivity can. Generate emotion in some people quite strongly and that you will have people that say that I'm super productive. Why you trying to make me more productive and using AI to make me more productive? How about you? Let me just get home to my kids at a reasonable hour rather than, you know, the amount of overtime I'm doing just. Getting through the repetitive, mundane work, and so I've seen this move from the concept of productivity then to let's talk about efficiency. And then I was at a conference in Portera last week in Slovenia and a Microsoft Copilot person presented. And she said, well, it is efficiency, a higher enough goal or should it be effectiveness? And I'm like, OK, what do you mean by this? This is a real interesting nuance in play here. And she was saying. So let's say you had an all hands meeting every fortnight and the team comes together, have an all hands run Q&A. As part of that and we decide to implement a chat bot to replace the All Hands meeting. Now, is it more efficient because anybody can ask their questions anytime it's more efficient, but is that more effective? Because now you've moved the whole people relationship that was happening in that event and I was like. That's such a an interesting nuance is that should we be aiming for effectiveness and then? I've even gone beyond that as I've unpacked it on the way home and said how should we be looking at how AI becomes part of purpose for an organization. And rather than just, you know, productivity rather than just efficiency effectiveness. But really, how do we create more human purpose by using AI in the mix? What are you?
 
00:08:43 Alex Pearce
Yeah. I mean, the two things come to mind there. The 1st IS, is that you've got individual productivity and you've got business productivity. So spin out to spit out the two, I think is quite important that individual productivity to be able to go off and to improve somebody's work, work life balance, all those kind of things is is 1 is a very quick. Memory and that business productivity, how can I do things better? How can I increase my uh, what I do from a user base point of view, my customers, my business processes? Can I improve those? Things and AI isn't the first thing to come in place to be able to say. Let's make it more productive. Interestingly, yesterday I went to a garden centre, I took my 6 year old with me there and she's like, what's that? Daddy and I was like, that's a typewriter. And she was actually like, how is she, like, pressing the individual buttons and like, couldn't, couldn't understand the concept of it. So we've gone through productivity. Before where we've gone from a typewriter to a computer and obviously we've got 30 years worth of history, at least of of computers and and typing in from a keyboard point of view. And AI is just another transformational tool to support the individual. To poop yard to do things better. I've heard Microsoft recently talk about how they want to build copilot and AI just so it's like another set of features into word and Excel rather than something that's built on top of or we've just added a bit of a on top. No, we want it natively in the applications, and we want you to see it as if it's a native set of features within the productivity. Within their productivity application. So I think we see it more from that user point of view and users will go off and take advantage of that AI side of things. You can go off and do teams meetings, recordings, yes, great you can do these things and those are the standard things. And I I kind of call them the foundations now. I think everybody knows that people don't want demos of those things. Now, what are the true productivity gains? Now? This is something I'm a little bit more passionate about, looking more at that agentic AI. How can agents actually help you to be more productive? Now you're working on a iOS ISO standard. Let's say it's O 42,000 and one that's the AI standard. They're out there. Would you build an agent to help you to have a conversation to know with hang on a minute. I'm doing this document. Can you tell me whether I'm hitting this particular set of standards? That's a true productivity gain. If you're building a gentle Ki that goes off and does tasks for you, or even autonomous agents, that is where we're going to see real true. Transformation, productivity, efficiency gains support both the individual but also that business side of things.

00:11:19 Mark Smith
You mentioned the ISO standard, which was music to my ears, because you're probably the first person I've come across it that's brought it up rather than me bringing it up in conversation in the last six weeks. What are you doing much in that in that space with that new ISO AI standard that's come?
 
00:11:40 Alex Pearce
It's only something we see, so my my job is to go off and speak to customers. I'm almost an evangelist as a job, so I get the great benefit of talking to not just education establishment now, but legal firms and all sorts of different types of businesses. And I think because the rate of AI is going so fast at the moment to do a particular. ISO standards related to AI it's quite difficult because hang on a bit, that's something new. That's something new. Where does it go off and do those kind of side of the? And just to give you as an example is that if you started looking at all the applications you've got on what's got AI, you can probably quickly go. Yeah, they've got an agent. Yeah, they've got some that's built into it, but you need to be assessing what models that they're using and where are they using that model. So let's just say there's a a company out there. They've built a SaaS solution. It's holding your personal data, but they're using clouds to be able to. Go off and assess it. While Claude got an API, he's got the API key, it goes off, but it's a consumer service using. Now your personal data to be able to go off and assess it. So should we be going off and asking what models you're using? Where is it being hosted? How is it protecting my personal data? So to go off and deliver an ISO standard when people don't have that understanding that they need to go off and do that. But from a legal point of view and a technical point of view is a challenge. So when I give that challenge to customers as part of my call, they're going, we're not ready for ISO standards yet. I AI is too big at this moment in time. We need to get a grip. Of application standards, application policies. What does it mean to use different language models, agentic and so many people don't even know that you get copilot studio with your Microsoft 365 Copilot licence. So when you give them that, they go and I'm gonna. I'm gonna go and resolve that issue now rather than this is everybody looking at the ISO standard. So that's my experience. But love to hear yours as well.
 
00:13:32 Mark Smith
Interesting. Interesting is and the reason I'm deep in it is that we're building an AI management. Custom designed for enterprise customers to manage all the as I'm going to say, manage to be able to report on show due diligence across and you know trustworthy AI at the end of the day, no matter what the workload. So being able to. Audit themselves maturity model and then where are they using it? One of the pushbacks I've seen on Microsoft Copilot from organizations is that there's no clear what model is being used behind. Copilot and therefore companies are going in or what we'd rather build our own chat solution. Let's say in something like foundry in Azure, because we at least can choose our models, we know what they are, that type of thing where Microsoft is really because it's a SaaS solution that you're getting with copilot is a. There's a lack of. If you like traceability on what's being used for the different, you know components of of copilot. The native M365, copilot I'm talking about.
 
00:14:44 Alex Pearce
Yeah, absolutely. And and this is The thing is that when you're going off and you're using Microsoft Copilot, they're looking after the model. They're looking after the orchestration layer. They're looking after the safety and all that kind of side of things. But that's where agents sit in really nicely cause you can bring your own agents into work, into PowerPoint. I think you can now do it in XLS well. You can build your agents as you mentioned inside Azure AI Foundry. You can state this is the model. This is where I want the model to be hosted. Just as an example, speaking to a legal firm, let's say that there's a set of solicitors that are dealing with. Child. Matters. And you asked copilot to go and assess those files it. Not it's got content in it that cannot be assessed by their models because Microsoft has set that orchestration layer to say we're not allowing that content. Therefore you can go and build an agent inside the Azure AI. You can set the content layers all that kind of stuff. You gotta build the compliancy as well because you want to make sure that they're using it an ethical kind of way. But it does enable you to be able to go off and do that. So yes, I've been challenged by organizations of what model is it using, I'm saying. Don't worry about that. I don't wanna have to say that to you, but that's Microsoft dealing with that. They will bring what they think is best and the right things. They'll go off and do testing and they will bring it now to the box feature if you need to be able to have more of that granular control. Azure AI Foundry and the other challenge that I get around that is that I need it to only be UK focused. I'm in the UK live in. The. In England, speak to defence organisations as an example, we need to only have this only working in the UK and Microsoft say well, it's within GDPR. That means it could go to a European. Data centre to be processed. And we can't have that. So that's where again, AI as your I found you comes in because you could be more precise.
 
00:16:29 Mark Smith
Now you mentioned an EU standard there, like GDPR and of course the AU AI Act, the sorry EU AI Act has come out and explicitly. How do you think UK businesses are looking at that? Are there like, well, not part of the? EU so we. Don't care what's the. What's the? The play from a, you know, cause I know that the UK are getting their act together. They've released. I know in the last nine months or so some detailed. Documents. I wouldn't call them white papers, but thought, you know, pieces on how they're thinking of of, of AI. What are you seeing from UK businesses?
 
00:17:13 Alex Pearce
They're just consuming it is what I'm seeing. The innovation is going on outside of the UK and let's be honest, it's probably happening outside of the EU. What you're seeing from an innovation point of view is in a war between America and China. And and it's interesting to see a lot of the Middle East now making agreements that go off and enable them to be able to use more of these AI tools. It's really interesting to hear that the UAE are gonna give everybody in their country ChatGPT Pro, right? Wherefore you're never going to see this thing, right. I think it's amazing. I think it's it's going to completely. Main job is go off and do it. I I guess my question is who's actually seeing what prompts they're making, and does it go to the government or not? But that's something I've not done my own research into, so I can't really form an opinion on that from an EU and the UK point of view I I guess the people that are looking at that guidance are more public sector. What is it that the UK is saying? What is it that we consume from an EU point of view? Because there's all sorts of agreements between the UK and the EU that we still consume certain things. One, as an example, autonomous cars still can't do autonomous cars inside the UK like you can in the US because the EU have said, well, we'll just follow them, follow sort of things. So there's a a big, UM kind of understanding of, you know, what we'll follow what the EU we're doing, we'll follow to see what they're saying. The UK will still come out with something and do their own thing and say what it. Is. I think we've got bigger challenges that because I think we've got the knowledge to be able to build AI our own. AI solutions, it's just too expensive inside the UK, with the most expensive country in the world, from an energy point of view costs, that's not going to drive innovation as much inside the UK or inside of the EU. So yeah, public sector are the ones that keeping it don't really hear much from the private sector.
 
00:18:57 Mark Smith
Yeah, we're about three years into the Gen. AI story and you know, copilot from Microsoft side is about what is it? But a year and a half old, maybe. And and it's been a market. What are you seeing from organizations around their maturity or their capacity to you know, is it still? They're worried about being left behind. So the fear of missing out where what are you seeing in the market as? You know, I mentioned in the intro there around frameworks and and the need for. Ways of thinking about how AI is implemented in organizations. What are. You coming across?
 
00:19:40 Alex Pearce
 Yeah, there's there's kind of two or. 3 trends. I've got some organizations that are really kind of Fast forward thinking. They're saying to me, Alex, my Microsoft so slow and because they're building these solutions, they've got to spend time. They're gonna make sure it works with their current code base, put it in the right governance controls and support Microsoft to be able to go off and say, you know what, when we deliver it to you, we're going to deliver it to you in a GDPR. Standard, but it means that we're not gonna deliver it on a day one. That's kind of my first customer there, a smaller set. I've got some customers that going off and just using copilot using in Word using Excel. And they're probably sitting there with 50 odd users going. I'm not sure if I've got any value out of it, so I jump on the call. I talk to them and saying, well, what is it you're spending? What do you know about the product set? And I don't talk about copilot necessary. I talk about Microsoft AI. Architecture. Because we've got to talk about copilot plus PCs. What does that mean? For the end. Price Lenovo have told me that 60% of the devices that they're gonna ship at the end of 2026, so we're talking 18 months away now, 60% of them are gonna be copilot PCs, AI focused machines. So potentially in three years all devices are going to be AI. Focused devices. So what is that as part of your policies and processes moving? Most people aren't even thinking about that. Don't even aware that. Ohh we'll we'll leave that for now. Well, you can't leave it for too long because the market's changing to be able to go. Off and deliver. That as part of your copilot, are you aware of actions and notebooks and copilot studio and agents? And their answer is no. What's that? And I'm like, well, it's part of the $30 per user per month that you're spending. How do you not know this? And that's one of my criticisms in a nice way towards Microsoft is you talk about what's going on with Microsoft 365 as a copilot, as a set of products within Word and Excel and PowerPoint, but you're not telling anybody what you get within the full licensing stack, which includes that code, pilot studio and all that. So that is very much. Probably more my I'd probably say about 80% of my conversation is giving them that understanding of. Let's go from the foundations of talking about your data and what you can do and how you can talk to that all the way through to what it is to have the researcher in the analyst. The two big agents that have come out and gone generally available for everybody in the last 48 hours around the world through to building your own agents and that maturity kind of side of. Things. And then I guess that last few ones is those organizations, I think we're going to do AI now we we feel ready for it. We feel that we understand what's going on more in the marketplace. And they've got a bit more of that kind of confidence that, you know what we are now ready for an organised. And let's start moving on and understanding what it is. Let's take down the barriers that we put in place, what we need to understand what it is and how it works. And you still have to go off and tell them that actually their understanding is wrong. That's me being polite. But you've got to ensure that you know what the product is that all Microsoft will be training our data. It's going off to consumer. Service. No, it's not like what it is because that's what it says with copilot pro. And I'm like, yeah, but copilot Pro is a personal set of features that you buy as an individual. Your corporate versions are very different. So using that same name still causes confusion as.
 
 00:22:58 Mark Smith
Well, Alex, we're we're over time. It's been an absolute pleasure to talk to you and I hope we have many more conversations. In the future.
 
00:23:06 Alex Pearce
Any time more than happy to to jump up.
 
00:23:13 Mark Smith
Hey, thanks for listening. I'm your host business application MVP, Mark Smith, otherwise known as the nz365guy. If you like the show and want to be a supporter, check out buy me a coffee.com/nz365guy. Thanks again and see you next time.