Copilot Adoption That Actually Works

00:00:01 Mark Smith
Welcome to the power platform show. Thanks for joining me today. I hope today's guests and Spires and educates you on the possibilities of the Microsoft Power platform now. Let's get on with the show.

00:00:23 Mark Smith
In this episode, we're going to focus on copilot origins, features adoption all those great things as it has become very prevalent in our lives. Our guest today is from London, England. He works at Illuminati learning as a senior copilot and L&D consultant. He's a tech enthusiast, drone lover, Team T. I love that Xbox player and FIFA king. He lives and breathes AI and helps business globally navigate the ever changing AI landscape and adopted smoothly in their work flows day-to-day. You can find links to his bio and socials in the show notes for this episode welcome to the show Dan.

00:01:01 Dan Boyles
I'm gonna make one correction. I'm actually head of AI, a company called First AI Group. Now that was my old job title. That was my job title from August I actually I joined a startup back in August.

00:01:09 Mark Smith
So then Wow. I love it. I love it.

00:01:14 Dan Boyles
Called first AI group, so yeah, I'll just drop a quick connection there. So yeah. Yeah. Thank you so much for having me on the show. But yeah, I'm actually head of AI at a an AI adoption company called First AI Group and I've been. There since yeah, end of August.

00:01:28 Mark Smith
I love it and this this is, you know, so timely. But hey, before we get into the AI conversation that we're going to have tell us a bit about food, family and fun. What do you do when you're not thinking about artificial intelligence?

00:01:45 Dan Boyles
Cool food. I've got massive love for.

00:01:48 Dan Boyles
Mexican food and I spent a lot of time in California, a lot of time in LA and I just fell in love with Mexican food trucks there. It's actually my goal, right? So at the end of everything, once I've completed my AI journey, I actually want to retire and just run a food truck selling burritos and tacos. That is. I got inspired by the film Chef.

00:02:05 Mark Smith 
Yes, another another film.

00:02:06 Dan Boyles
That is my end goal for everything, so a massive fan of Mexican family so. Quite a quite a close. Knit little family like my partner in London just called, been together quite a long time. We actually met. Each other at work I took. I was at Tech Talk global headquarters for two years and we met. On a zoom call in in lockdown. And four years later, we're still together. Which is awesome.

00:02:29 Mark Smith
So it wasn't a staff party.

00:02:31 Dan Boyles
It wasn't that, no. It was literally a zoom staff party, I think. And yeah, we we exchanged some messages and and flirted on zoom. And then, yeah, four years later, we're still. Together, making a life for. Each other, which is awesome and fun. I'm I'm actually a bit of a geek, right? So I I paint warhammer, I've got back into Warhammer, so I paint.

00:02:39 Mark Smith
Nice.

00:02:47 Dan Boyles
My little warhammer and I play golf. And that is my stress relievers for the week. Go and smash and balls.

00:02:52 Mark Smith
Around the golf course, he's still playing with the. Drone.

00:02:55 Dan Boyles
No, no, sadly. I live literally under the flight path at Heathrow airport. I'm about 5 minutes away, so all flights are banned from drones. From both. I can literally see the runway from the from my apartment, so I think it'd be dangerous if they let me loose. With drones anywhere near real life airplanes.

00:03:12 Mark Smith
I was living in London. I know there are a few, you know, drones were. But this is 6 years ago, but there were a few incidents out out at Heathrow with with people doing silly things with drones in the the part.

00:03:25 Dan Boyles
Yeah, they don't actually take off. Like if I try and take it off my roof, it just won't let it. I'm pretty sure I'll be on a government list somewhere.

00:03:29 Mark Smith
Yeah. So I'm. I'm not. I don't fancy my chances on that one. I took my drone to the Great Wall of China and I couldn't fly it well. We left off. Yep. Yes. No.

00:03:38 Dan Boyles
Really. No, we think we're free. Yeah, we're bound by technology. We're bound by the constraints of technology all over the.

00:03:46 Mark Smith
Place. Yeah. Yeah. And where I am now, which is is rural but about say, OK and a half to 2K's from my my door. Is the flight. Path for another airport, but I have never flown that way, so I've never been smacked by the, you know. Yeah, yeah, yeah, yeah. Did TI have pretty good at, you know, honouring all those airspaces and stuff so.

00:04:04 Dan Boyles
Avoid. That way, don't go. That way, danger is that way we. Don't want that so that. You can't jailbreak it. You can jailbreak it by just. I just don't want to be on a newspaper. Cover on LinkedIn for you know, accidentally causing an aviation catastrophe.

00:04:21 Mark Smith
Ohh, totally totally totally. But I would have. Yeah, I think bird strike would be probably an ever bigger problem for in a in a lot of situations. Right. And for aircraft that is. Bad news. You know what I think? I mean, just to digress a moment, I think in one of the updates to my DG and I've got a Mavic 3. I think that they might have even now allow planes to break it up on radar. The drone, like there's some kind of flight signature that they are added in. I don't know. As in I just thought I saw something in a change log or something like that. I've not investigated any further than that.

00:05:02 Dan Boyles
I'm getting. I'm sad now. I'm sad I haven't. I'm not going to play. I much drive like 20 miles that way and just. Go. And play on my drone tomorrow. See me in the field just flying randomly.

00:05:12 Mark Smith
You know the freakiest place I've put my drone up was in the middle of Russia in a place called Lake Baikal. And every time, well, I had it in the air, I was worried about being picked up, you know, as a like as somebody ever come because I'm pretty sure I wasn't allowed to be flying there.

00:05:30 Dan Boyles
That's gonna say it's pretty brave. I've I've spent some time in Moscow and Saint Petersburg, and there's definitely a list of things I couldn't take with me. And I think a drone was one of them. So brave, mate, I appreciate.

00:05:40 Mark Smith
But yeah, I definitely got it up. Got it out and and and landed it and gone. Ducked away. Right. I have once been intercepted by federal police for flying. This is in Sydney, Australia. This is when drones are brand new. Like I got the first, you know. DJI Mavic that you know came out. And I was in a big park, but right next to the fence and I was literally next to the fence was Government House, which have got permanent federal police on site as I put it up, taking photos of the harbor. But, you know, video of the harborage and I put my camera down and I can see these cars. And like people pointing inside this compound coming to me, so by the time I landed the drone, the police were walking up to me. What are you doing? Blah, blah. So no fly zone. That type of, you know. I need to watch you delete the cause, and of course they weren't smart enough to know about the how the case worked and that I can delete off my phone, but it's still on the SD chip on the drone. So yeah, I didn't lose anything.

00:06:50 Dan Boyles
Fair play, fair play. I mean, yeah, I'm. I'm just gonna go and flow. I'm going to tomorrow. Gonna book myself on a meeting for an hour. I'm just gonna drive until it will take off. And then. I'm fly about.

00:06:58 Mark Smith
Nice. Nice. I love it.

00:06:59 Dan Boyles
Reminisce but but I'm really happy we finally get to talk by. The way and. I've been we've been flirting with this.

00:07:03 Mark Smith
Yes.

00:07:05 Dan Boyles
For about 6 months now, so we've. Actually got it right. So.

00:07:07 Mark Smith
Yes, yes, yes, I love it. I love it. And on the side on this topic I have. I've got a book contract. From Microsoft Press to write on copilot adoption. And so it's even more timely to get your insights and thoughts and stuff in this area. And I've been working with Microsoft internally for the last year and a half on adoption. And what makes big projects successful from our particularly copilot first and foremost.

00:07:39 Mark Smith
Let alone getting into things like Herbalife Studio Foundry and and everything else. You know that's particularly available on Azure from a a wider set. So yeah, very keen. First of all, tell us about what's your journey been into this whole AI space over, particularly the last three years. I don't know if you're dabbling in it more intently before generative, you know, became such a big thing on everybody's radar.

00:08:05 Dan Boyles
Buzzword. Everyone loved it. Yeah, I mean. I've I've got a long history of of Microsoft as part of the Global Master training team for a long time, specifically on the commercial side. So I was looking after server HDI stack when that. Launched. That's fun. Obviously to lock down took a dive out on on Korea, moved to a different company, TikTok for two years. Head of training for a region there, then came back over into the world of Microsoft and I think it was a case of the right place at the right time. I was doing a lot of M365 training. I've I've been in end user adoption specialist for a very long time and it just so happened that copilot came out and the company I was working for, the previous company they had the CSP masters contracts to deliver. CSP Masters training to all the deities from America right through to Dubai, on copilot and wow. Look, I've I've been interested in as soon as you know the beta testing of of ChatGPT was out. And then yeah, when it kind of came mainstream 2122. I've just been an early adopter, but I love AIC. You know, I missed the.com boom. I'm just young enough to have missedthe.com boom and not. Being. A part of it, but I see this as it's generally the biggest technological advancement we've we've had possibly in the last 1015 years. So when it came out, I just threw myself into it and I learned everything. I was building everything I possibly could, taught myself a lot more coding, a lot more. Kind of involved and then. I guess I was just spending all of my time helping. And most of the the Tier 1 dusties build their copilot practices, so I've got quite a tech. A tech background. So I've been doing really, really interested in the technical, the pre set up the security, the architecture, all the stuff that goes into making it work like the engine. But for me adoption is the the thing that I love doing because you. Know we spend so much time. Trying to change people's lives with technology, this actually can. This can touch every single person at every single level, whether it's C-Suite or to to to an assistant or or front of house. And it makes us good at something we're not good at. Yeah, which is for me, it's the ultimate dream. So. But that's been my journey. And then? So that was. A company called illuminate learning. I then joined start up. I had an offer from quite a big global partner to go and head up their. Copilot practice so. As I call me to go and do that, I'm gonna live and breathe, copilot, and then start up. Approach me called First AI Group and I've joined them and we've actually got a really nice business. So we're solely focused on copilot adoption. That's all we do.

00:10:26 Dan Boyles
But we hire, we focus on diversity and we focus on taking graduates straight out of university, training them up to be copilot experts, copilot adoption managers, and then we get them ready for a really good career and we put them into global business. So I've got a really nice mix now because I get to work, you know, our clients will range from 50 users, right? Our biggest ones at 9 1/2 thousand seats deployment. So I get to be as hands on as I want to, but also get to get the next generation of workers ready to have an amazing career in AI. So that's that's my journey outside of work. I mean, I've build. Harvest from Iron Man to running my house and on my smart speakers. That's the level of geek that I am. So yeah, AI is is everything to me.

00:11:08 Mark Smith
Yeah, I'm there with you. Man. Everything in my home, if I say she who shall not be named movie. Time I have got three big window shutters that come down. My projection screen comes down, my projector turns on, my NVIDIA Shield turns on. And my kids are just like, this is like they're trying to work out the command, and they're only two years old. And and four years old. And they still haven't got the the tone right to be able to do it. And then at the end of the movie time, I say. That's all folks and ohh my big blinds which I'm talking about big like ranch like a tight floor to ceiling windows all go up my projection screen, everything shuts down but like I've even in the studio here when I come in in the morning I just say studio go is my my trigger and all the LED's everything lights on computer. Mouse up sound, disk everything and then our same command on exit.

00:12:06 Dan Boyles
And that's it, right? It's most basic form is it's automation, but it's automation that is there to help and it's that's what I've been interested in for so long. And I know it sounds geeky that, you know, different people get what the word turned on, but they get enthusiastic about something and it's how can we use. A little bit.

00:12:19 Mark Smith
Yes.

00:12:20 Dan Boyles
Tech to help people.

00:12:21 Mark Smith
Ideas turned on.

00:12:23 Dan Boyles
Turn on, hey, it turns me on I don't know. I don't know my girlfriend’s gonna hear. About me saying that, but so it does. It's generally that you know it's automation, even copilot studio just power, but so power virtual agents and then it's just turned into copilot studio power virtual aspect, whether it was called before but. And it's it's such an exciting time. It's so. Cool, I love.

00:12:43 Mark Smith
It it's funny because I'm in two Minds 1 I feel that the whole agentic story that has been touted in the last six months, that's ramping, you know, louder and louder and particularly autonomous agents. I'm like, I think you're using what you're referring to is hyper automation right as a as an. For me, it's not autonomous yet autonomous is I set a command that has many different steps that I haven't even thought of. And you can work that **** out and come back to me with the output that I'm kind of seeking that's much more autonomous than I'm thinking when I think of cause for, you know, if you look at agents, there's there's rule based agents. If then type statement agents. Yeah, which I think that's. Really where we. Are but the marketing tells us they're autonomous, and I'm like.

00:13:29 Dan Boyles
This is it. I've got a problem with the marketing around AI, not Microsoft marketing. I love Microsoft marketing. They're very clear on. What it does, to a degree, some stuff they could. Maybe be a little bit clearer on. Everyone thinks that AI is just this magic stick that you wave over it. It fixes and it does everything and. It's not, yeah. And I had this exact conversation with one of my new starters earlier on and it was. What's the difference between a prompt or workflow, an agent and an autonomous agent, and the easiest way I can describe it is if I build a role rate or a a rule based agent or a prompt based agent that is like me getting in a car following a sat NAV and I can never deviate from that sat NAV. I'm going to I'm driving to the airport and I'm going that way. If there's traffic, I'm gonna sit in that traffic and I'm gonna get there when I get. True AI agent autonomy would be I get in my car and it does whatever it needs to do to get me to that destination with minimal input from me. It just knows I need to get there if it needs to take me off the highway and put me on a back Rd. it does. If it drops me at a train station and tells me to get on the train, that's where we are and I think there's a bit of conflict between what we expect AI to do and what it actually is. People of which, when we're then speaking to businesses, causes an issue sometimes because they think it's, oh, we can just do this and you're almost. Like, hold on, start with the basics. Let's let's do this.

00:14:41 Mark Smith
Yeah. You know, if I hire an EA to work with me and and help me, right, and I give them a job description, I say these are the things they're gonna. Blah blah, blah, blah, blah, blah blah. Here's the list right after one month, they never refer back to that job description, but they also know they probably do 60% more than my job description because I didn't document it in it because one, they've learned like, if you're gonna book me a flight, I prefer to be on an aisle seat. I prefer to be in an exit row. Like I didn't document that, but. After one or two interactions when she's booking, she goes. Well, in the future, I'm not gonna put him on the window cause he doesn't like the window. He he likes to not have to climb over people or, you know, to get out, like. And that's what I call kind of like tacit knowledge. When you're in an office, you know the dynamics, because you feel here at that type of thing. And when I think of AI being autonomous, I'm like it's gonna pick up almost nuances about me. Over time, and that's why I think memory is such an important construct that we're going to see a lot more investments in and and particularly in long term memory and going. Hey, you know, like for me and you know, I've worked with a lot of virtual assistants over my career, had them work full time for me and that type of thing. And and I would have a protocol which I I I I refer to it now as a standard operating procedure if this happens. Here's the kind of. So they knew that if I was flying to another country. And I had dead time in my diary. They would know that I. Like to do food to. So I didn't have to tell them. Did you find me a food like that would go? Hey, is this food too? Do you want me to book? You in there, I mean them. Sure, cause they've they've seen that behavior in the past. I didn't have to write it in to the command or the rule or the automation, and that's why I feel that we're just not there yet. But I do believe we're on a pathway to.

00:16:32 Dan Boyles 
100% apartment there, but I think also as well. I'd probably say. 60% of companies I speak to aren't actually ready for the basics of AI. There's a lot of stuff I think that there's we, we've almost jumped a step. You know, we get ready for cloud. We were forced into cloud computing through lockdown a lot, but that was the biggest change in IT approach. I think I've seen. Yeah, a lot of businesses took a cloud first. Coach and then chaos ensued with SharePoint and team. And you know, security groupings and obviously Microsoft have done a lot around the intra security, so you know P1P2 your add-ons E35, whatever you've got. Yep. The securities big. Part, but we've got a lot of different playing fields and cause every every company is a little bit behind. I would say on the security you find the odd gem that are going right. You're ready for it. But I think a lot of the conversations we have is.

00:17:14 Mark Smith
Yes.

00:17:16 Dan Boyles
Let's focus on your infrastructure first. Yeah, the basic let's get back onto the basics of it. I always think this as well. We're light years ahead in our personal life and we are in our business life.

00:17:25 Mark Smith
Yes.

00:17:26 Dan Boyles
So in a personal life, we. Use biometric security day for every day we use. It banking logons. Nathan, businesses I speak to you still use the password they've got written down somewhere. Authenticate has just been made. You know it has to be done now, which is great. And we live in an actual world in a cloud world. If we lose my phone within two minutes, I can have another phone. You can buy it. I've. Got everything, phone, everything but in our personal in our business life.

00:17:50 Mark Smith
Hasn't it always been that case? Right? Your home machine is always way higher spec than what your business machine ever was, right?

00:17:58 Dan Boyles
I think I've got. I've got this same as yours. I've got the same lights on there, but that's it. But now in our personal life, we're light years ahead of where we are in our personal life with AI user. And I use my mum as a litmus test. My mum is probably the least technological person you'd ever meet, but she uses AI and she understands how to use it, so it's transferring that mindset of, you know, Sue, bless her soul. How do we get sues mindset into businesses? And it's just a bit of catching.

00:18:20 Mark Smith
Up sometimes I think, yeah. I've always looked at it from, you know, we can make gaming software addictive, you know. I've there's been times in my life that entire nights have disappeared and I didn't touch the pillow. Right cause I was gaming and we can do it and watching movies or whatever, you lose this concept of losing track of time and I've always seen how come we can't great business software. Then all of a sudden you clock in at the start of the day and you're like what? It's 5:00 already. Like, man, it has been such a good day.

00:18:48 Dan Boyles
Find me that job, my friend. Find me that job and. I'll be there. Yeah, right. I think that's what AI where we're at is if we think about mobile phone or mobile phone, the first thing I do when I wake up is check my mobile phone. The last thing I do before I go to bed is check my my because it's ingrained into everything. It's ingrained into my car, my shopping, my money, my banking, everything on there. I think we're on a journey for AI to be like this.

00:19:09 Mark Smith
Yes, 100%.

00:19:09 Dan Boyles
AI is gonna be there, but it's not there yet. Yeah, and it's two people like us being able to guide through to to get that done.

00:19:15 Mark Smith
So let's talk about adoption. I one thing, one pattern I have noticed over the last year and a half when it comes to copilot adoption and and even maybe broader use of AI when you know custom AI's are built for an organization is I've never seen the emphasis on adoption.

00:19:36 Mark Smith
 From the get go, you know, I came from a power platform dynamics background for the last 21 years. And fought in many cases with customers about the need. You know, I come from a pro side type background and doing a formal implementation process. They're kind of and the salespeople like how do we sell it, it's. Like not like. The data shows that 70% of projects fail because people weren't taken on a journey. And a formal adoption program was put in place. What I am noticing? Is that there's much more or greater appetite to get adoption sorted out before the launch. In other words, it's inherently been built in, and then the odd I I've got 2 use cases in my mind which I won't mention the companies names which are asked for their copilot licenses to be removed, large volumes of them, because nobody adopted in both cases. 80 switched it on flick the switch and then 6X. Laid out what? What? What are we paying for this for? Right. Like, why didn't people just adopt it? Well, no adoption program, no like and and. And then one of the challenges is that I show is that. The Google search box. If I take that and let's say take the ChatGPT. Prompt. They look identical. They're. On a microphone at the right. End in for voice. And you take people and they go, oh, I can ask it a question because that interface reminds me, and I don't logically think, right. It's a inbuilt learned response that this is just a question till they ask the one question which they think is. Bumped. And then go, oh, I didn't like the answer or wow, that was magic. Now I'm gonna ask it. Some crazy stuff. And don't get that this is a conversation. This is a multi facet like I can easily can consume an hour in dialogue with AI.

00:21:25 Dan Boyles
100% I'm ready, right? I'm. I'm glad to hear from your side of the pond that's happening because it's happening here. Every company I speak to, it's happening to so. I guess would you like to hear? How we approach adoption and what we do? Which kind of?

00:21:37 Mark Smith
Yes, I'd love to hear.

00:21:38 Dan Boyles
Cool. So we're not an MSP, we don't sell licences. All we sell as adoption. So we've got partnerships with a lot of the MSP's, Microsoft we. Work close with everyone quick.

00:21:47 Mark Smith
Quick question, how do you handle ISIS? Oh, do you know?

00:21:52 Dan Boyles
Don't touch it.

00:21:52 Mark Smith
Gotcha. Yeah. I'm. I'm. I'm in the same place. Right. The company that I work for, we don't qualify under any of the cloud competencies because we only do strategy. Right. We bring in partners that do the implementation work because we decided we weren't going to have an army of consultants that every time we do strategic consulting, the company goes, yeah, how many people you're trying to sell us right to come in behind. So we do our arms lengths separation from that.

00:22:21 Dan Boyles
We we partner up. Yeah, we we've got. A wide network of people.

00:22:24 Mark Smith
But we don't then qualify and Microsoft on the one hand is going to us, hey, listen, if you could be esif, we'll give you the launch funding to get the strategy work in place because we our results show we get large builds out of it, but it's not on our letterhead, it's on another partner's letterhead. So therefore, we don't qualify for those type of. Yeah.

00:22:45 Dan Boyles
We we we just have to partner with those that can and then either put the work through them or you know approach them and strategic partnerships like we're we're a sole adoption business. That's all we do. We don't, yeah, we can advise you on your pre sales work and but we just get involved, the journey should be we we get involved quite early but we're not involved until later on. Our approach is really simple Sir.

00:23:06 Mark Smith
Yeah. So tell us should be persist. Tell us. Versus.

00:23:09 Dan Boyle
So the process is before we do anything straight working, if the companies right fit for us. And if we're a right fit for the company, depending on what they want to achieve, the first thing we do is a leadership meeting myself. They're senior leaders, they're you know, creation of a a senior leadership champion program who's paying for it, who's responsible for success, who's managing it, identifying all these things and just sitting down and going. Right. If we split this into 3 levels right at the top, we've got your organization. What is the organization's goals for adopting AI? Are you doing it for a competitive advantage? Are you doing it because your competitors are? Is there things such as? As are you going through a merger and acquisition, have you got headcount challenges, rapid expansion that gives me like a business use case for doing it from from a top down then we go department by department. So again what the challenge is in specific departments is this that you have a skill, a lack of skill, is it you're over skilled, is it you need to expand very rapidly on there and then to individual users actually what? Is, you know. What does success look like at all levels? Build a strategy around it. We go to them with a proposal. And. Say, look, don't turn it on for everyone, or once less focused department by department. Let's focus on what you need to get done. Let's build a test user case and we'll start with this. And no matter the size of the. Business we'll take between 10 to 50 employees. And we'll do a 10 week rapid adoption with those before the global switch on. So we'll take two from marketing, 2 from HR2 from sales, 10 from operations. And we sit side by side one to one for one hour a week and we go full needs assessment. So we did one for a legal firm recently 6 and our thousand lawyers we sit down with a test group and go what's your bottlenecks. What is stopping you getting work done? What is, you know, let's declutter the noise from the signal. What's the process is? What's the applications? Where did your data live? We go off on the back of it, we start building custom prompts, proof of concept agents and workflows, and we test it over 10 weeks and at the end of the 10 weeks we come back to the business with a focus group and say, OK, you've represented across the entire business. We'll give you an 80 page strategy document that's got 500 prompts that you can use across the business. Here's where you should focus your investment. Here's where you're saving the. Most amount of time, but key to it. What are you gonna do with that time off the back of it? Now, how do we protect a bit of our lawyers? Lawyers don't wanna turn it on because they think they're their billable hours are gonna go down. But actually, we give. You the strategy? To make more money with it, which is why we're getting some really nice contracts is because we we're in that Gray zone now. We're playing in that zone of. Helping you do that, so once we've done that, look, we've done our our 50. Use. Focus group. We then take it global to a customer and that could be multiple countries. It could be 200 people, thousand people. It doesn't matter we. Then do group training. So up to 102. 100 people can rock. Up to the trainings, we use prompt buddy. I live and die by. Prompt buddy, I swear it is the best tool.

00:25:40 Mark Smith
Never heard of it? Never heard of it? What's prompt, buddy?

00:25:42 Dan Boyles
So point buddy is. A power app that you can deploy on teams. Anyone can click on there and it goes right. You click on the application and it's got all of the prompts that you use in your business. If you go by Persona, you can go by business group. You can say what do you wanna do and it will give you the video of how to do it. No. Yes.

00:25:57 Mark Smith
Wicked. Is that your product or somebody else's?

00:26:01 Dan Boyles
I know it's it's on GitHub Microsoft shout out to the creator that that made that I can't remember his name. But it's a. Great tool. It's a free power app probably. I'll send you the link.

00:26:09 Mark Smith
To have a look at. Yeah, I do want to have a look at this.

00:26:12 Dan Boyles
But we drive through that, we run a help desk, so we run regular drop in Sessions department drop in Sessions, department training on demand, learning material videos and that's how we approach it. But the key for us is that human understanding to start with, which is what does your business want from it because if we can't align with what your goals of success are, it's pointless, have an AI on.

00:26:32 Mark Smith
This. Yeah, there's there's a whole bunch of this. I want to pick apart with you. I. Because I come from a power platform 10 year back end. That type of thing. One of the things that you know and this has been prevalent a lot of different companies, is that a company would establish the centre of Excellence and the Centre of Excellence has always been a bit of a rub on me.  Particularly in the last I suppose. Five years because I've seen customers go, oh, we don't like that word. That sounds like that's an elitist group inside the Organism. Nation, you know, IT has a history of making common people feel like they're dumb, right? You know how many times you know? Earlier, 20-30 years ago. You call help desk and you go. I'm trying to do this on the keyboard and it's not working. And a server, a service on the network has stopped the IT. Person identifies that while you're on the line, they start the service and they go OK. Try that again and you go oh, it is working now and they're like, you know, like, yeah, it was you. When they know they fix the problem, but they're not gonna look like knobs, right in doing. And so this history has been built in culturally into organizations, and then all of a sudden along comes AI and OHS tech. So therefore it sits in the IT department and what IT do they.

00:27:49 Dan Boyles
They kill the projects, IT kill the projects.

00:27:52 Mark Smith
Exactly, exactly. And so one of the things that I've noticed is that. Big, successful projects always get executive sponsorship and I was interested to see that you got the leadership team. And because the other thing is guess who has the budgets, the executives?

00:28:06 Dan Boyles
Right. This is it. There's three people. You've got the person paying for it. The person that's managing it, and the person that wants it in the business. So I normally find that's the. CEO and CFO and CTO. If I can get those people in the room right, it's good. Because where I've seen copilot fail and look, I've been a network manager. I'm. I'm about to throw some shade on network managers and IT managers. You give it to our team managers and they go. It's not really that good. I'm just not gonna push out to the rest of the business. Yes, give it to the right person. And if you get their boss saying that, you are gonna. Push this out and you get. The buy and that's why that three pronged approach really works well for me. If I speak to someone that goes, I are just gonna handle it. I. Tend not to. Try and work. With our business or I would just give them a real hands off, here's the stuff we're gonna make sure you get it. But. The hands on approach needs to be there. There has to be buying from from that champion the the top level champions. Because it just goes wrong.

00:28:54 Mark Smith
So. Is a pivot to Coe and I see Cole as a subset of this concept here, which I call a centre for enablement because it's just a little mindset difference. We're here to enable people, we're here to help people come up with ideas innovate, you know that that type of thing and it's that little nomenclature and then you've used this word a couple of times, which is champions. And if we look on the the curve of adoption, the bell curve of adoption, you've always got these people you or I in that category that before anyone tells us to get on the bandwagon, we're on it. We've been using it for ages, that type of thing, right and so. I have found that if you can tap that audience, find those people in the organization they are going to be your vocal mouthpieces, right? But then you've gotta go into that. The first adoption part of that Bell curve, which you need to formally. Take people on a journey. How does that look for you when you think of a champions program? What are the artifacts? Are you setting up a SharePoint chat environment at teams environment? How are you kind of and and I'll just actually give you a bit more reference. When I was in London, I interviewed the the lead for Virgin Atlantic and. She is now Microsoft employee. But what she did is she established what I would call a centre of enablement and that she was very much a technical person. But she understood, you know, organization, culture, personas and stuff. And she ran regular luncheon learns, she ran app showcases off some part of the business is doing great.

00:30:20 Dan Boyles
Yes.

00:30:31 Mark Smith
She would put on her. Hey, come along and ask you questions. And quickly got to the point that she had a backlog of three. 100 Project wishes from the organization that are adopting everywhere. Microsoft loved it, of course, because Power platform went through the roof and this organization and then, you know, she went to Microsoft and and created a lot of tooling around the name billing organizations to do this. Coming back to what you're seeing in your champions program, what is that underlying tooling? Which really allows you to create a mouthpiece, A megaphone, a communication pattern protocol that makes people go. I wanna be part of this. I wanna be recognized. And this I wanna be show that like I'm the go to person and every company has those go to people, right.

00:31:15 Dan Boyles
Yeah. I think first of all, when I when I when we set a champion program up, you know you start who. Wants to be. A champion. If you put your hand up, you're a good candidate. You might not be the right person, so these needs assessments that we doing the business identifies who's going to be good for the proof of concept. Cause ultimately you don't want someone at sea level. You don't want someone at senior management level. Because they will have assistance, they won't be doing the day-to-day work through it. So they'll be a mix of people, but our champion program I like and. To all out assault. I will use everything at my. Disposal to shout and to talk because there will be people that will see this as a blocker. They need to change the way they work. So the first couple of weeks of quite a closed environment and that's where we just investigate the business and see how we're gonna approach this. But we have everything. We have teams, channels, we have business wide emails that go every Monday saying here's what we're planning this week. Here's the business challenge. That we're tackling. Here's the KPI we're working on. Here's the success story from last week. Here's how much time they've saved. Here's what they've done with that time we do on demand videos that we send out, and we push out through teams, update channels, direct emails, and we've got a website we've got. Help desk most of our work is remote so we don't have face to face the adoption manager that we put in there becomes their best friend to everyone and it's all about celebrating successes. All the cool stuff. From the front. And now on the back end, what we're doing is actually being quite clever is the people using our champion program is the best case your business will ever have to get an AI. So they will be the person with the maximum returns. As time saved for money or revenue generated. And and I've also learned that. People don't wanna do it themselves, they just wanna. Have it done for them. So like I say at the end of that 10 weeks or. The eight week. Period. We will go to the departments with 200 prompts that we have tested 1000 times that we know work, that have a benefit and we'll put it in your library and go there you go. Here's a video on how to use it and it's the regular check-ins. On there. So I'm on about everything. There's podcasts, there's videos, there's teams, there's emails, there's shout outs, there's lunch and learn calls. There's help desks that you can speak to every working day, every working hour, everything. Why not?

00:33:16 Mark Smith
I love it. I love it. I love. But personally, how do you manage your prompts library?

00:33:22 Dan Boyles
Prompt body properties. How I manage it so I'll show it to you prompt buddy as soon as I've got a good prompt to upload it straight onto that.

00:33:29 Mark Smith
So is that, but that's copilot specific, right?

00:33:33 Dan Boyles
Like I use it for anything I use it for for all my stuff I've I've actually got a loop where I've just got everything dumped in there that I built, also built on a GPT agent that I've uploaded all my stuff to that I go give me the.

00:33:39 Mark Smith
Yeah, look.

00:33:45 Dan Boyles
Best I'm trying to do this. I know it sounds really bad, but I tend tend not to use it too much during work. My biggest stuff is all about recapping of teams meetings like my pain points at work is the amount of communications I have and the amount of meetings I have. And also I'm really forgetful and I've got ADHD, so being able to I'm a bit scatty to to do stuff. So you know my my most used prompt of 4:00 everyday is what is outstanding. That's important. I have one hour left in my working day. You have to do this.

00:34:13 Mark Smith
Interesting.

00:34:14 Dan Boyles
Prepare me for tomorrow Morning. Here's. The look for tomorrow morning and. The other thing is prepare me for this. Meeting it's my biggest prompt. Though is obviously in a. Bit more depth than that but. That's the the gist of what I use I for.

00:34:24 Mark Smith
Yeah.

00:34:25 Dan Boyles
If I'm coding or building yet it can help me code quicker than I can if I'm, you know building nice customer facing training material. It doesn't touch it cause I can do that better than. Than. Now I can, but for me my weak point. Is Is organization and it is communication, so that's what helps me for. The most I'm. Not the best person to use that in the. Business because I'm very simple with what I use it for. Yeah. Compared to some lawyers that are comparing you. Know. 500 paid contract analysis for this. That's really cool. That's interesting, that's. Not me. Yeah, I'm just simple so.

00:34:53 Mark Smith
Interesting as.  And where I found that I repeated using prompts. So last year I set a goal like I do extreme things and they're in the spirit of the minute and I make commitments to myself to as I'm sorry I I buy leverage to hold me accountable. And so I made a commitment publicly on social media. That I would shoot a TikTok video every day for 100 days and if anyone discovered I missed a day, the first person I'd pay them 1000 USD. Right, Yep, that buys me motivation, right? And so I did, I'd for 110 days or something like that. And So what I was using prompting for, I would go. I have a concept, blah, blah, blah, blah. And I just recorded and to chat should be T in this case. And then I would have the prompt which would. Take this. Identify the hook, identify the call to action. I don't want it to be any more than 30 seconds, 90 seconds. And don't take out any story that I've given you. I've I've given you a personal story. Don't play with. Don't you know, reinvent that story for me. Give it to me. Let. It be my and. So I found that I would want to. Do that over and over again. And so I used a a tool that I've used for years called text expander, which is a little app. It's a it's a cloud subscription, but it works in all my browsers. It works in. Everything and that enabled me to just and it's a keyboard that works on my iOS phone, so I just go to the keyboard and go TikTok video and it and that prompt. I adjusted massively over time right as I developed the concept, yeah.

00:36:38 Dan Boyles
But that's the evolution. That's that's the evolution of Crompton. You try it and you. Change it and and yeah.

00:36:43 Mark Smith
Yeah. Yeah. And it's like, I'm always in search for those kind of tools that are gonna save me mass amount of times. And This is why I go back to memory that I talked about before because. Some of my prompts were literally pages and pages and links because I am preloading the prompt like who's my target audience? What's my voice like? I did a whole exercise of what is my my voice. You know, I'm a bit irreverent. I swear sometimes. Like, don't, don't, don't dilute me. Right. That's the way I am, yeah.

00:37:21 Dan Boyles
I'm glad you said that because. I think yeah, the whole thing went from. And. I think they're with AI good in good out. If you put good data in, you get a good data out. If you put a good prompt in, you get a good answer out. I think people are shocked when we talk about prompt engineering and and kind of the level that we go to an adoption that we don't do top line we we drill down into your soul and give you an answer. I've got problems at like 1300 words long that we spend hours. Often, but it does one very specific thing. So actually when I think businesses, when we're saying look, you know you can take two of our staff full time to do your adoption and that's what you need to get this true adoption and we go to them. I say that playbook I think the. Longest one we've done is about 2 1/2 thousand. And that is and that's for a legal firm. But the the result of that is that took something billable that takes 2 1/2 hours down to one single prompt.

00:38:08 Mark Smith
Yeah. Amazing, right.

00:38:10 Dan Boyles
And that's worth the 20 hours open, but that is probably, I don't know how many iterations we did, but it's iteration after. Iteration. And. Much like anything, it's much like the controls on. Teams. It's the. Controls on everything is that it evolves over time. The prompt that we use just now which goes ohh. Can you summarize this by the way, summarizing PowerPoint is incredible. Summarize this into a Word document for me. Makes me sound like I know what I'm talking about on the latest. Update that's correct, but actually the true adoption that he said is the thing that's more than 10. 1520 hundred words long on a prompt.

00:38:39 Mark Smith
Yeah.

00:38:40 Dan Boyles
That's cool. I'm just geeky, so sorry.

00:38:43 Mark Smith
I want to pick up on something you said earlier, which is the there are three things that you do to initial kick off and engagement. You mentioned data. How do you have that data discussion? In that one of the themes that I have seen more and more is people going, you know, they use the word hallucination and they're like ohh it got it wrong and I'm like. Hang on a second, let's and and as and another gentleman forget his name. That gave me this illustration, he said. If you take an average size organisation, let's say it has 10 million data aid effects, just as an arbitrary. And then you give that to AI and say there you go. I've found it in my information. Right? And and AI starts, you know, it goes through all that and it sees the Excel spreadsheet of which there's now let's say, 30 copies in the organization. And the person that originally created it made a mistake in a formula and. It's now there's 30 copies of the mistake in a formula. No one's. Connected it. This could be a legal document. Somebody had changed a terms and condition. They didn't remove the old one. It's been superseded now. The contracts confusing and we've created 50 copies of that. When we onboarded 50 customers and you give that as training data and then you go, Oh my gosh, it says it's got this wrong yet you trained it on human. Sierra. Multiple times in the organization and you're saying the AI is wrong. The actual data you gave it just reinforced the error. It doesn't know whether it's right or wrong, right? And and and so I talk about, you need to take that 10 million you need to distill it down to, let's say 5000 for the use case that you have. If you're going to use a rag process. And then you need to refine that to maybe only 500 artifacts that you're going to use. As as the knowledge for whatever you're doing, because then you're going to take those 500 and you're gonna sanitize them, you're gonna clean them, you're gonna understand. And that's what you're gonna give out. And of course, all of a sudden hallucination drops right down because garbage and garbage out. Right. You've validated what you've put into the system. Do you do any work around that area?

00:40:50 Dan Boyles
We advise on it. I would just hope that the the partner that has turned copilot has done it. Now we want our own assessments. We'll we'll run over sharing assessments. My favorite 1 is just putting the documents somewhere that we shouldn't have. Access. To and going right, I'm gonna try that as a guest user or try it as you know, an entry level user into this and it will just have one specific keyword and then we'll run the over sharing based on DLP and label and everything else. UM. We're not in the business of making changes to it, but I will make recommendations on how to do it when we do our proof of concept, we'll actually just do the the, the setting where you can shut down entire archives of of SharePoint and only let them see a certain SharePoint to train the data on and to train themselves on. I've actually seen it gone really wrong when you can see documents you shouldn't, and so we advise on that. We'll give a strategy around that. We won't do the work, we will not. Press the buttons on that, yeah. Yeah, but it it's definitely interesting. The company is working with and the SharePoint was just. In the end, we actually end up putting the project back by 6 months. It was easier for the for a company I was working with to rebuild the SharePoint from scratch than it was to fix it because in lockdown when everybody started having access to teams you had like 1, I think 1 SharePoint about 95,000 different permissions for of access and. You're like, no. Just rebuild it and also cleanse your data. Copilot doesn't need to look at 10 years worth of documents, just give it what you use. If you've not accessed the file for two years, sack it off onto a SharePoint somewhere else. Don't let copilots sit.

00:42:16 Mark Smith
Called storage, yeah. Exactly. Listen, we're well over time and it's so fun talking to you. I think we should book a A reschedule in six months time and see what's changed. Where? Where are we at in the world?

00:42:28 Dan Boyles
Let's do it, man. Thank you so much for this congratulations on the book, by the way. I'll be really interested in reading that when it comes out.

00:42:36 Mark Smith
Thank you. I might send you a pre release. Get you do a review of it, a lot of review but like you know as a what do you call it? You know when you're actually? Yeah. If you came, I'll put you up for a as a a reviewer. So before it even you know goes to.

00:42:51 Dan Boyles
I would love to. I too am writing a book on AI adoption. Ohh good. I've not been. I've not been commissioned by Microsoft to do it, but I'm I'm doing it as. Part of my. My I just. I just wanna contribute towards the channel right and and help. People do it. Right. So love it. So I will send you that. I said I wanna write it. I've written the contents page and I've written probably about half of Chapter 1.

00:43:03 Mark Smith
1.

00:43:10 Dan Boyles
Between what I do, but yeah, I'll. I'll send it to. You when I'm done as well I.

00:43:13 Mark Smith
Love it. I love it. We'll stay in touch then.

00:43:17 Dan Boyles
But I mean, it's been really refreshing and it's always good to speak to people. That share the same. Passion for it and it's clear. Look, you know, with what you do and and you talk about in history, it's. Just been a. Pleasure chatting to you. I've really enjoyed it. Excellent. Thank you.

00:43:29 Mark Smith
Hey, thanks for listening. I'm your host business application MVP, Mark Smith, otherwise known as the nz365guy. If there's a guest you would like to see on the show, please message me on LinkedIn. If you want to be a supporter of the show, please check out buy me a coffee.com/nz365guy. Stay safe out there and shoot for the stars.