Navigating the Fascinating World of BizApps, AI, and Tech with David Yack

Mark Smith: 0:37
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. Today's guest is from the United States. He is the CTO at Colorado Technology Consultants. He was first awarded his MVP in 2003. Based on my calculations, that's 20 years ago. He's a Microsoft regional director, photographer, public speaker in Colorado Springs. He enjoys developing applications in the Microsoft platforms, not just the Power Platform, but right across the tech stack, specializing in large system architecture and design. You can find links to his bio and social, as well as if we talk about any things that have links to them. They'll be in the show notes for this episode. Welcome to the show, david. Good to be here. Mark, how are you Very good? Good to have you on the show. I think that, of all the last six years of my podcasting, you're the longest standing MVP to join.

David Yack: 2:00
Yeah, this is an exciting year it was my 20th.

Mark Smith: 2:04
Wow, that's pretty phenomenal. How many years has the BizApps and I know it had other names how many years has that community been running as part of the MVP program?

David Yack: 2:15
That's a really good question. I'm not sure when the BizApps one it really started from CRM right it kind of involved into the BizApps. I actually started as an ASPnet MVP fully on custom software dev, and then I did a stint with Silverlight and then switched over to CRM and then followed on to BizApps.

Mark Smith: 2:41
Wow, a long journey. You've obviously seen a lot in that time. Before we get into the area that you play in today and what you're focused on, tell us a bit about food, family and fun from your perspective. I know Colorado Interesting. I'll give my fun fact about Colorado my understanding. It has the highest bungee jump in the world. Now I've done the highest bungee jump in New Zealand but I understand about once a year there's a festival or something and the highest bungee jump in the world is done in Colorado.

David Yack: 3:13
Yeah, I think you're talking about the Royal Gorge, I believe. Yeah, it's actually a bridge over a creek and you can jump off it. I'm not big on jumping off a perfectly good bridge. I have the same feeling about airplanes and diving out of them.

Mark Smith: 3:33
I've done that too, so tell us yeah. Food, family and fun in Colorado. What about them? What do you like, liz, and what do they mean to you? What's the best steakhouse to eat at? What do you do for fun when you're not doing the best?

David Yack: 3:50
steakhouse that we have local is a steakhouse called Peppertree. They actually cook the steak at your table and they do Caesar salads also. So they bring a little cart and they actually make the Caesar dressing right in front of you, nice.

Mark Smith: 4:06
And what do you do? Do you game? Or what do you do when you're not doing something like it in the IT space?

David Yack: 4:13
Recently we've been playing a lot of pool. We got a new pool table, so we've been trying to become pool sharks Nice.

Mark Smith: 4:20
Nice. Is it like a pool, or does it also do billiards? Is it like one of those super-sized tables?

David Yack: 4:27
It's an eight-foot standard pool table so we do a lot of nine ball.

Mark Smith: 4:33
Nice, nice, hours of entertainment.

David Yack: 4:35
Yeah, and sometimes Julie wins, sometimes I win, sometimes we have guests and they kill us. Nice, nice. Next time you're out this way, you'll have to stop by and play a game.

Mark Smith: 4:45
Well, are they ever going to run a conference in Colorado?

David Yack: 4:50
DynamicsCon, I believe, is here next year in Denver.

Mark Smith: 4:53
Wow, okay, that's cool. That's cool. Who knows what will happen between now and then. I told two weeks ago I didn't think I was doing any conferences this year, and now I've been told I'm going to what is it? Microsoft Partner Power Platform Conference in Vegas. Then I've got another one in Slovenia next year coming up, I think May the Dynamics Mines Conference. So All of a sudden, the the world seems to be getting back into pre-covid gear.

David Yack: 5:23
Yeah, I think things are starting to open up again finally and people are starting to do some in-person type stuff. I was at build and the MVP summit. Those are my recent ones that I was at and I had a good experience, certainly at both. Got COVID at the MVP summit, which wasn't one of my highlights, but I wouldn't change anything for going, so somebody happened to have it there, got it, check the box and moved on.

Mark Smith: 5:48
That. That is unfortunate. Tell us about what's on your radar at the moment. What are you focused on? What's what's 2023 shaping up to be for you? What are you excited about? And tech that's changed.

David Yack: 6:05
You know there's a number of things I'm excited about. I mean I think we've gotten to where we've got, especially in biz apps, where we've matured enough to where there's an actual Viable product that is being sold and deployed and used by people and all kinds of people are able to build stuff. I go back to when I wrote CRM as a rapid application development platform and you know it's been a number of years since then, but we've kind of really made that truly a skew that can come to life on there and so really excited about that. But I'm also excited about a lot of the AI stuff. I mean, everybody's excited about AI and and I think that excitement bleeds off on everybody- yeah, Absolutely.

Mark Smith: 6:49
It's interesting that you just said CRM rapid was a rapid development platform. Yeah, so this is the crazy thing We've had such an index on, you know, low code, no code, with a power platform and and people thinking that the workloads that you can build on it are Just the simple things. Right, maybe the Excel replacements. Maybe you know what I call clipboard based apps and yet through our careers and I've been 20 years in the dynamic space I've Built some big apps. You know that that that Involved tens of millions of dollars worth of revenue. That evolved around the apps that we built and it was on Back in the day we called XRM and was allowing us to build anything like you know I was talking recently of built an app on Dynamics 365 sales or Microsoft sales back in the day that controls Roding assets 11 times the size of the United Kingdom, right? So this is big roading infrastructure in WA and People. I think there needs to kind of be a pivot back to. You know you can actually can create enterprise scale Applications on the power platform. So that book you wrote back then really holds true to today.

David Yack: 8:11
Yeah, it certainly does and and I'm actually really excited we did the. Microsoft just published a app modernization White paper and maybe you can provide a link to people in the podcast. But it really drills into a lot of that and I was happy to participate in some of that just recently came out and it really highlights some of the things. You may not build a stock trading Application on it to track the trades, but you can certainly build everything around it and I think a lot of people turn off the thinking, like you said, that it's only for you know, maybe a simple business process, but you can do a lot more than that.

Mark Smith: 8:47
Yeah, absolutely. I had another one recently which was for a power utility company and the app that wanted built would manage End to end the entire sales lifecycle but also the delivery, implementation Lifecycle and validation of solar systems on homes, Right. So it's quite a an app that does a whole bunch of stuff. It has you know digital signatures at different points. It has you know stage gates of things. That is needs to be a property assessment. There's a you know contractors go out to the property. All of that built on the power platform end to end and and it's a tire business unit. This is a system that runs the entire business unit you.

David Yack: 9:32

Well, and if you look at Dataverse I mean Dataverse used to be strictly relational data you now have unrelational data, that's, non-relational data that you can store in there, the elastic storage. We were just looking at that for a scenario today to see if it was applicable where you might otherwise look straight to Azure Cosmos. You can do some of that within the platform habit, tied with the one single API access that you're working with.

Mark Smith: 9:57
Yeah, tell us about 365.training, which, by the way, if you're listening to this, if you go to 365.training, that is the URL to access the site that Dave has. Tell us about that.

David Yack: 10:12
Yeah, so 365.training is all focused on Power Platform. The idea that we had is we want you go to a number of sites like I won't name them, but you know the ones that have just everything under the sky and you get lost in the sea of content. We wanted to focus on the Power Platform, have some free content, some paid content, kind of a mix of both, some community stuff out there We've been recently. One of the things I'm really excited about is we've recently been adding speaking of AI. We've been adding that into the platform alongside the training. It's not going to go learn things for you. We want to help you in that process. Think of it almost like a co-pilot for training.

Mark Smith: 10:54
See, I love this concept. Internally, I'm working on an app that basically will interrogate all of Microsoft Learn and give you. Let's say, you want to replace an Excel spreadsheet, what do I need to learn, if I'm a citizen developer, to go and consume that? So, on your platform, are you saying a similar type thing? You can go, hey, I want to learn XYZ skill, and then it will give you, of your entire library, which is an extensive library, the path of things that you need to learn and do.

David Yack: 11:25
Yeah, and what we've done is we've paired it with the course topics. So if you take a PowerFX course, you're going to be able to ask PowerFX questions and it already knows that you're dealing with PowerFX Because each of the co-pilots have context information about that. We've injected in and fed in some extra information so it's not just like going to chat GPT and getting a stock answer. You can go to chat GPT or even get up co-pilot and ask it a question, but it doesn't know the context that you're in. That's one of the things that I think is really important about AI. I actually was looking at co-pilot for sales and that's a good example of it's getting smarter, but it's not quite there where it knows the context of the opportunity. It's those context things that make AI really useful for somebody working with it.

Mark Smith: 12:18
So let's unpack that concept of context, because I don't think a lot of people understand what it potentially means. So we have a large language model. I understand and not to my understanding, and you can correct it is that the context piece is your data coming to the mix? But it cannot be a bunch of things as well. It can be preliminary data, it could be metadata, it could be the fact that it knows where you're logging in from, who you are, what you've done to date, the context of the training that you're in. It is aware of all these parameters and so when you ask it a question, it's almost like it constructs the correct, prompt engineering to hand back to the large language model to give you a very precise, accurate answer with not too much hallucinations. Is that about right?

David Yack: 13:10
Yeah, I like to think about it. So we just published a course on power pipelines and if you take pipelines in itself, you said what is a pipeline? And you went to chat GPT and asked that I would imagine there's a good chance you might get something about oil in your answer. Whereas if you go to our site and you ask what is pipelines, it already knows that you're dealing with power pipelines because through the prompt engineering we've already instructed it to that scope. We also have kept it to say don't go talk about oil pipelines, talk only about power pipelines. We've also because large language models are trained at a certain point in time and that's the knowledge that they have. They don't have the ability to gain new knowledge unless it's injected in as part of the prompt that goes in. You see, on some models, the fine tuning. For example, on the 3.5 model they just released the fine tuning, but that's not always necessary. A lot of times you can just tell it along with your prompt, give it some additional context when you ask us about that. It doesn't know about power pipelines because it was released after the model was trained. It wouldn't be able to answer those questions. But we can give it some baseline, some grounding data. That helps it understand how to answer some of those questions. It's really good at reasoning, understanding what you're asking for, formatting, and that helps it give a more intelligent answer to the question. If somebody gives us feedback that hey, that isn't a good answer, we'll go look at it, take that feedback into mind, maybe give it some more grounding data to give it some more basis for the answer it gives the next time. What's?

Mark Smith: 14:53
your learning curve been like since November, december last year, where open AI splashed onto the world stage. Everyone became aware of LLMs, if you like. It became noticeable in the public consciousness much more than the AI has been around since the 60s. But it's now this whole area of generative AI For you and applying this to the Power Platform. What's that learning curve been for you and where do you find your resources? Because a lot of people, and I'm probably. I know that you're technical by nature and you look at it from a technical lens, but how do lay people get started, do you feel, with the AI journey and in the modern context of how we know AI?

David Yack: 15:43
Wow, that's a lot to dive through there. You may have to pick up the pieces, as I missed some of the things you asked there. As far as for me, the journey, I was a little bit taken surprise. I followed AI for a while and it's always been interesting. But there's a high bar to entry. We did some neural network stuff years ago before it was even popular, and now it's gotten much more in the vogue. But you're seeing AI become a lot more approachable by a broad set of people. But at the same time I think you're also seeing the hype of it exceeding anybody's expectations. It can solve everything. Ai can't solve everything. In fact, it'll probably create some more problems for things that weren't problems before. That we'll have to figure out and grounding people's hype in that AI can't solve everything. And I think where people get disappointed with AI sometimes is they think it's this magical thing, that's unicorn, that's just make everything. They ask one simple question. It'll do everything for them and it takes work. Just as I might ask you for feedback on an email or help me compose something I might ask AI to help with, that Doesn't mean that I'm going to take what you give me and just slap it in and send it without doing any editing on it. Right, yeah, I want to put my own personality on it, and those are things that I think you're going to continue. What I'm excited about at AI is seeing that personalization, that tailoring, that partnership come to life in the different products and things that you use every day. But at the same time, I'm realistic that it's not magic. There's technology behind it that does things on a Rational way of doing things, and you have to not expect it to do the dishes for you.

Mark Smith: 17:44
No, no, that's good. I like the explanation, but what it sparked in my mind then is how soon do you think? Because you gave a bit of understanding of what it is and what it isn't. For the last 20 years, I've been collecting all my personal data. I'm talking about scripts from going to doctor's appointments, all my insurance policies, everything, all my e-mails, the whole lot. I want to get to the point that I can feed that into a large language model and it knows me. I can interrogate it to find out what I haven't thought about, what I know and things like that. Do you think we're there? It's kind of like a digital twin, but for a human, not an IoT device. What are your thoughts?

David Yack: 18:36
I'm going to stop you right there. No, we're not there yet. The reason for that is that's a bucket load of data. If you think about AI's memory currently, that it's capable of the context that it's able to talk to and remember that you tell it, it's much smaller than that bucket that holds your life knowledge. That's the thing that will evolve over time, I believe, is how much of that it will be able to know without having Because you don't want it to be trained on your data. You want your data to be used as necessary to answer questions, but it needs to have a large enough capacity that it can pull the important things. In your example, it might not need your financials, but might need certain things from medical. It would know how to load that in, use it and then not be trained on that, Because you don't want it trained on your data right.

Mark Smith: 19:33
No, I don't want to train the model, but I want to anchor it. My vision in the next two years is that everyone will have an LLM on their phone. There's already been predictions that'll happen within 24 months, that they'll condense the latest LLMs down to around 2 gig in size and therefore easily can be stored on the phone. The concept of basically then being able to use that against your data sets. I could go back and go okay, where were all the dumb financial decisions I made in my life? It was there, there, there and there, yeah.

David Yack: 20:10
Yeah, I think that's. What's interesting is, people are going back and I think companies need to do this. Individuals need to do this on their own. How do they do stuff on a daily basis? Be realistic about it, but look at where the AI can help them and where it won't be helping them. For companies, that means re-envisioning some of the things of how they do business that weren't practical before. That will be practical going forward because of what is enabled. You also have to realize that that's going to rapidly change. Even how you do that will rapidly change. You're going to need to invest wisely in where you spend your resources implementing those. At least, that's my opinion. I don't know what your thoughts?

Mark Smith: 20:54
No that's good. That's good. Listen, just on your website right now because we're at time to wrap things up. How do people get access to this new AI functionality on 365.training.

David Yack: 21:06
It's on a number of the courses. You'll see 365 AI in the course details. Power FX is a good example of one. One of the things I like to do with Power FX is ask it to generate some code for me. I give it back to it. Ask it again what does this do? And it will explain what the code does for there. But there's a number of courses that have it. Appliance has it. Dataverse has it. The Dataverse one actually is infused with my Dataverse book, so when you ask it a question about it, it will look at the Amazon book content, look it up, include that in the answer, blended with what it already knows.

Mark Smith: 21:51
Hey, thanks for listening. I'm your host business application MVP Mark Smith, otherwise known as the NZ365 guy. If you like the show and wanna be a supporter, check out buymeacoffeecom. Forward slash NZ365 guy. Thanks again and see you next time.