Revolutionizing Customer Service Through AI with Jay Wellings

Mark Smith: 0:32
Welcome to the Power Platform Show. Thanks for joining me today. I hope today's guest inspires and educates you on the possibilities of the Microsoft Power Platform. Now let's get on with the show. In this episode we'll be focusing on mega-app implementations on the Power Platform, or what I like to call anchor apps. And today's guest is from greater Portsmouth area in the United Kingdom. He works at Volato as the head of development. He's an experienced web application developer. You can find links to his bio, social media etc. In the show notes for this episode. Welcome to the show, jay. Thank you very much, mark. How are you doing? Good, good, good, good to have you on and kicking off this series of recordings I'm doing over the next couple of days. Tell me, before we get started, food, family and fun. What do they mean to you?

Jay Wellings: 1:28
Food, family and fun. Put them all together. It's probably the best mixture you could get. Right, take your friends, friends, your family out, have a good time, have some good food uh, for me it's always a a good steak can't complain about a good steak and uh, there's a couple of good restaurants, uh, in the portsmouth area, at south hampton area, uh, that we enjoy partaking in. So that's uh, that's me on that one. How about yourself?

Mark Smith: 1:51
nice, nice. I can't go past a good steak. I've've just been in Vegas and I did three steak dinners across three nights Keep your phone and they were amazing. Yeah, nice little eye for it. So yeah, I can't go past steaks in the US. They do it well. Tell me about your career journey into the Microsoft ecosystem. What's that been for you?

Jay Wellings: 2:20
So I've been doing this game longer than I care to admit. So I've kind of been a software developer for 18 years. I started my career, like many of us do, out of university. Well, if we go all the way back to university, I studied computer science back in 2003, many, many moons ago, and for me that was really kind of I had always enjoyed messing about with computers, playing with computers, but I didn't really know where I wanted to go. So it was like what's the most general place I can apply that trade? And it was a computer science degree.

Jay Wellings: 2:49
By the time I'd got to the end of that degree I really found that my passion really lived in the world of the front end, essentially building things. So I could build with, you know, solving problems visually and I could get that immediate feedback. It wasn't I wanted to just kind of do a backend system and wait for it to process. I wanted to create front ends, solve problems for users immediately. So I was kind of drawn to a web development route. So my first role was as a kind of junior web developer. I worked a support role for a little while fixing books and issues on HTML front web pages. From there kind of evolved my career in the automotive industry, working on many, many kind of information stacks. So we had Java, we had something called ColdFusion back in the day I remember ColdFusion yeah, I think it's still kicking today.

Jay Wellings: 3:40
There's still a loyal but small community doing that good stuff, and it was kind of analogous to PHP, which was quite big back in the day. I remember Ruby on Rails was also kind of taking off and so you know, I was working on that server side of web development. But also really kind of that was when, you know, web 2.0 was really taking off, so interactive web pages we moved from static HTML and CSS and we started moving into a world where, you know, the browser was the GUI and so, you know, we started talking about things like single-page applications, and Google Maps came onto the scene and it blew everyone's mind at the time, and so that was kind of my trajectory of my career up until around about I think it was 2016. I was basically I got to a senior web developer role.

Jay Wellings: 4:34
I left the place that I was working and I was just basically looking around for opportunities and within that I found a role for a head of development within the 365 area vicinity and that was where my journey with kind of you know all things Power Platform, sharepoint 365, all that good stuff really began. So I jumped in, you know, headfirst. Lots of things to learn there, lots of things to get my head around. But from that I've been very much within the Microsoft space, not just kind of like within the Power Platform, but also kind of you know, net, azure, functions, all that good stuff, and we've just been playing in and around the area for the last six, seven years. So that's pretty much where we came from and where we've been and where we are today.

Mark Smith: 5:23
That's awesome. That's awesome. So you know the theme for the show is Mega App Implementations, Just at a high level. Tell me a bit about probably the largest project that you've been involved in that involved the Power Platform.

Jay Wellings: 5:36
So I mean for us. I think probably one of the biggest projects that we've worked on has been for I will be quite coy throughout this interview, but it has been for a financial institution and kind of really you know the the classic uh digital transformation. So taking a lot of, you know, manual processes, you know data that was kept here, there and everywhere, included excel, because we all know the majority of uh the businesses in the world are run out of so many excel workbooks and what we've done with that company is we've fundamentally transformed the way of working for them. So it's not just a case of 70, 80, 90 disparate processes, separate workflows, isolations and silos of data. What we've done is we've kind of come in and rather just like, oh, we're going to deliver you an application.

Jay Wellings: 6:39
We've gone in, as you know, an incumbent partner, essentially an extra limb of that business, and say, right, we're going to help you figure out what your power platform strategy looks like, rather than just say we're going to build you a power platform app and what that's really allowed us to do is really, you know, bake in to that organization using Power Platform as their fundamental technology layer to run their business at a core, core level so that all of financial transactions, movements of data and information within the organization. We've managed to completely kind of transform that from this disparate you know many silos and many data pots and create a kind of end-to-end solution to help them run their applications and run their business. And we've done that predominantly with Power Platform and there's you know bits and pieces where we've kind of jumped out of that and we've used some Azure functions and there might be a SQL database here and there, but for the most part it's been the key proponents of Dataverse, model-driven apps and Canvas apps that have really been the driving force behind all of that.

Mark Smith: 7:51
Nice. So give me some high-level numbers how many people in the organization would use what you've built? And then, how large is the potential customer base? Because I did one of these episodes with somebody that did passports and they only had 600 users, but they had over 5 million external users through the PowerPages portal that they had implemented on their platform. So just some high-level numbers, whether they're around the automations or the apps or the user base, et cetera, just so we can get an idea of scale.

Jay Wellings: 8:27

So I mean, I think, the entire organization using the systems as a whole. We're talking in the region of hundreds. You know it's probably in that kind of 200 to 400 region in terms of, like, the organization that's using it, but it's that business is supporting thousands of customers. Now, are those customers front end to Power Platform? Most probably not, but the way that they interface within to that organization, you know through, you know your traditional websites, through telephone, through whatever it may be, through their apps and all of that good stuff, that is all supported underneath by the Power Platform solutions that we're providing to our customers.

Jay Wellings: 9:07
So for me, there's always this balance, right? Because when we're talking about mega apps, it's you know, what are we bothered about? Are we bothered about impact that we're doing, or are we bothered about numbers of licenses that we've sold? Because if it's impact of what we're doing, streamlining that workload for that 200, 300 people that we're doing for that organization, is having a seismic impact for however many customers that they're working with, and that's uh. For me it's that. That's what I focus on less. So you know, we've sold a million licenses for microsoft, but someone might slap my hand for saying that out loud yeah.

Mark Smith: 9:47
So I tell you what. Why I come with this question is because it's not about the license sale, etc. Et cetera, but for me it's about understanding whether it's answering this question which I've had. Enterprise architects say that the power platform is not an enterprise platform. We'll use it for rats and mice type business If citizen developers can work on the platform. Therefore, there's obviously no strict ALM strategy, there's no security strategy, there's no. You know, we're not going to use it as an enterprise play. And that's what I'm trying to address on this podcast is is the power platform an enterprise platform where you would put mission critical workloads that if we switched it off in an organization, there would be major ramifications for doing that. So in that context, what are your thoughts?

Jay Wellings: 10:40
So, I mean for me, obviously you know cars on the table I'm going to be biased, right. For me. I think the Power Platform is absolutely fit for enterprise, absolutely fit for purpose, to be able to scale up, you know, from a small amount to a large amount of users, throughput and all of this good stuff. Being around the industry as long as I have, you often see people get very protective about their technology and a lot of enterprise people will say, oh well, you've got to use Java, you've got to use NET, you have to use this, you have to use that. It's the only platform that can support this, that and other. And what I typically find is 90% of that, 80% of that kind of bluster, is just emotion and feeling.

Jay Wellings: 11:28
Every now and then 20%, 10% of the time absolutely. You're not going to get anything running faster as a piece of code than some assembly that's written very specifically. Absolutely. You're not going to get anything running faster as a piece of code than you know some assembly that's written very specifically to run a very specific use case and business case. However, 99 of what software needs to do in the modern age is not run on bare metal at that kind of pace. What you need is software systems and solutions that allow developers to create fixes and software for clients and businesses quickly.

Jay Wellings: 12:02
And if we get so, you know we get caught up in a religious war for the one of a better term over. It has to be a java or a c++. You're. You're missing 95% of what's important and that's we need to, you know, solve problems for our customers. And the argument of yes, you know Power Platform it's. You know I've heard people make jokes about it being the kind of the Fisher Price. I don't know if that's a brand that you're familiar with, but it's a toy brand in the UK and it might be.

Jay Wellings: 12:36
You know Power Platform has been described as that and I think that's a very unfair designation. I think what Power Platform gives you is a fantastic entry level for non-technical people to implement solutions very, very, very quickly and very, very simply. But that doesn't mean that's where Power Platform begins and ends. There's plenty that sits underneath it and the nuts and bolts of it, mark, if we talk about it, is it all sits on Azure, right? Everything is powered by. You know, I'm not sure who's winning the cloud race at the moment, if it's Microsoft or if it's Amazon, but you know it's essentially the biggest cloud platform in the world that supports the majority of the biggest services, uh, that the entire world uses. And so to kind of just hand wave the platform, power platform away and go oh, it's not big enough I find that very short-sighted. Personally, and I think you know, microsoft provide many, many case studies of Power Platform being, you know, fit for purpose and ready to go to service these enterprise customers.

Mark Smith: 13:44
Yes, yeah. So the thing is is that you know, when enterprise customers are wanting to look at a technology, that's one of the first things they're going to look at is a security and being in the financial sector that this use case that we're discussing is about. How did the power platform address their security concerns?

Jay Wellings: 14:04
I think a lot of the security concerns were kind of just like baked in already, so it's not like there's not this huge kind of commotion about you know where's the governance of this, where's my code going, where's my data? You know where is this stored. None of these questions have to be answered because the platform manages all of that for you. So you know your environments. They're on your tenant right, your global admin and your security people. They can go in and they can audit everything that they need to audit. They know that out of the box, microsoft is providing that continuous backup of the data and the environments without anybody essentially having to do anything. And obviously we can layer additional levels of protection in and above that through the implementation of what we're building.

Jay Wellings: 14:54
But the proposition that Microsoft are providing the customer is security by default and essentially over the last six months Microsoft have changed their position internally to be kind of security first is everything, and those kind of things build confidence within the client because they know you don't have to wait for Jay at Valto to make sure everything's perfect, because you know Microsoft are giving you that footing right from the get. Go to make sure a lot of the security concerns around. You know where does my data sit? What regions is my data living in? All of this stuff is configurable for and by the end user, and it doesn't need a discussion with a third party like us to make that happen. And then, if someone's made the wrong configuration choice, it's going to take six months to move it. It's all kind of there in a box. I'm ready for you, I'm good to go.

Mark Smith: 16:31
How did the customer come to choose the Power Platform? There's multiple options out there. They could have used SAP, they could have used Pega, they could have used Service. Now, they could have used, you know, for automation, UiPath, Blue Prism, any number of technologies out there. What was the decisioning thought around the Power Platform?

Jay Wellings: 16:55
I think with this customer it was the kind of oh we've seen a bit, can you show us more? Can you show us more? Can you show us a little bit more? Oh, that looks interesting, oh, that solves a problem. And before you know it, it's not a case of they feel that they need to go out and do a full market comparison.

Jay Wellings: 17:12
They were already heavily invested within the Microsoft ecosystem anyway, for a number of other reasons and it was just like, oh well, this is where we're playing anyway. This just makes sense to kind of invest further into an ecosystem A we're confident with and B we're comfortable with. It's a bit of a no-brainer in those situations. But I think, to get to the crux of your point, it's like, okay, if all things were equal, why would a customer pick Microsoft over any of the aforementioned providers you've mentioned there? And I think it really comes down to that recognition of Microsoft, that confidence of you know Microsoft are going to do right by their customers. They're going to provide a platform that is secure, is going to be scalable. It is going to be here, you know, quite frankly, for the long term.

Jay Wellings: 18:02
There may be detractors of Microsoft, but I think it's fair to say if they put their money and their marketing behind a platform and a product, it's definitely going to be here for at least a little while. We're not living in the world of Google, where it's like one day they might cough and then they'll just close down an entire function or a business because that's not what they feel like doing. And I think Microsoft has spent many, many years building up that trust with the enterprise and with even the SMBs. People know that Microsoft are going to kind of stick with what they say they're going to do, and that does help decision makers when it comes to making those kind of calls on what platform to go with.

Mark Smith: 18:45
What was the first workload that was tackled? What was the kind of? Was it you know? Was it you mentioned Excel? In my experience in the financial sector, there's still a heck of a lot of Excel in use, and even access databases, for that matter, within the financial institutions. What were the kind of first things? Was it more automation focused? Was it more app focused?

Jay Wellings: 19:06
Well, I mean, once we got past the initial stages of you know we can do this and we can do that, and all that looks interesting. One of the first biggest problems that we really solved for the customer was let's get this data out of all of these silos. So it was like how do we centralize our core data sets that solve this particular problem, that are cross unit and cross function, that solve this particular problem, that are cross-unit and cross-function? And it continually causes problems and challenges because, oh, that person's got this version of this Excel spreadsheet and that person's got a copy of it and it's on a shared drive somewhere, but this person didn't upload it and then data gets lost. We're still talking about problems that were occurring 15, 20 years ago and there's literally no reason for that to exist. Were occurring 15, 20 years ago and there's literally no reason for that to exist. So the biggest thing that we did initially was a set of data kind of cleansing into a centralized data schema that made sense to support everything else that we needed to do moving forward, and obviously you don't always get that right on your first. Go right. When you're trying to bring all these disparate data pots in together, you'll your your damn best to get that schema and get that structure and you think you're there and then you put it into practice and then you have the conversations and then it's not quite what it needs to be, but the key thing is, as long as you and the customer having those conversations, you all understand the journey that you're going on and you understand this is not a one-shot deal and everyone agrees that we just iterate and evolve as we go forward. Those kinds of things are absolutely fine. We iterate, we get better and we improve, and that's what we did with that customer. We took all that data, we got it into their version one, we worked with them to figure out what their version two was and now we're in a good place of.

Jay Wellings: 20:51
That is our core data structure. That's how everything exists and the way that we get that data into the system is through this model-driven app. And there are these sets of processes and there are these sets of business rules that ensure the quality, validity and structure of that data is consistent in a way that it never was before. But now the business is dependent on the data being in that format. It's now just an expectation, but two and a half years ago, three years ago, whenever it was, we did this, that was almost a pipe dream for them.

Jay Wellings: 21:23
It was like their day-to-day was. They expected inconsistent data and there was someone's job to reconcile this on a day-to-day basis. They now have a system that does that and for me, that is where software development at its purest form provides the value that it needs to provide to people, which is let's make people's lives easier from taking away the mundane jobs. Take away the uh, you know away that kind of repeat job over and over again, where we can solutionize it once and we never have to worry about it again. And that's a bit of a long-winded way to get the answer you asked for. But yeah, we centralized all that data for them.

Mark Smith: 22:05
So when you talk about centralization of data, are you talking about Microsoft Fabric? Are you talking about Snowflake? Are you talking about Dataverse?

Jay Wellings: 22:15
when you come to centralizing that, so for us in this instance, we were using basically Dataverse, with the front end of a model-driven app being their kind of main admin and entry data point, with that model-driven app providing all the kind of essentially business logic and surety of what was going into the data. And then, fundamentally, we used Dataverse as kind of like your SaaS database. And that was what we did and that's what it still is today.

Mark Smith: 22:45
What are the plans around AI now in this mix? Because one of the big things with AI is you've got to have clean data and it looks like you've gone through and cleaned up the data. Therefore, the propensity to use AI is so much higher when you're starting with a clean data set. So are you starting to thresh out use cases now of introducing generative AI into the mix?

Jay Wellings: 23:09
So I mean, I think at the moment we're in a world where the possibilities of AI for the enterprise have not really been fully understood. It would be my suggestion. I think we're in such a world of what is possible and what is new. I think we've just gone through our first wave of awareness. Essentially, everyone's got their hands on a chat, gpt, a co-pilot. They've made it write a poem, they've made it sing a song, they've made it do whatever it needs to do. But now it's come into a brass tacks of okay, how can this help me in my day to day? And I think we're definitely engaging our customers around what's best practice, with where the technology is today and how this can help solve some problems. And I think what we're still finding is there's a lot of stuff that generative AI is going to help with, but it's not just going to replace people overnight. For me, this is just another tool in the toolbox that is going to help with, but it's not just going to replace people overnight. For me, this is just another tool in the toolbox that is going to help people become more effective and efficient of what they're doing in their day-to-day, and we need to get into a world where we're essentially doing what we were doing 20 years ago, which is teaching people how to Google. We're now in a world where we need to teach people how to prompt and over time we're going to be in a situation where things like, you know, copilot 365, copilot insert product name here they're just going to be part and parcel of the products that we've got and they're going to be, you know, as one of my co-workers affectionately calls it, clippy, odd steroids. So it's getting a little guy from Word 95 that used to pop up and we're going to have that in every environment and every system, but it's going to be a lot more contextually aware of what you're talking about. So, instead of the help icon just telling you how to use Microsoft Word, your help icon is going to pop up and it's going to tell you you know, I've observed this in your data Maybe you want to consider doing this and you want to consider doing that and X, y, z, and that's going to be really useful to help certain tasks, but I don't think it's going to be in a place where it's going to completely replace, you know, end-to-end workloads.

Jay Wellings: 25:26
At the moment, that's where things like Copilot Studio are coming into the mix and you, as a developer alongside your clients, can really say it's the next evolution of software development. There's this magic AI engine in the background that can do lots of stuff for us. Let's give that thing some guardrails, let's give that some structure and purpose, let's give that thing some guardrails, let's give that some structure and purpose, and then you can give it a lot more strategic value to a business, rather than throw all my data at chat, gpt and it's going to know everything. Everybody knows about everything.

Jay Wellings: 26:03
Um, and that's kind of my early doors perspective of it as a technical person. I think, from a non-technical perspective, there is a bit of a perception of this is a you know silver bullet and it will fix all problems for all things, and and I think it's our job as technology people to educate the wider audience about where absolutely there is stuff that AI and co-pilots and LLMs can really help us out, but also where those limitations are Kind of the classic software engineering problem of managing expectations, and I think that's where we are at the moment with those pieces. I'd be interested to hear your thoughts on it, though, mark.

Mark Smith: 26:48
Well, the reason I ask is I've just finished a presentation internally for Microsoft and I was doing research on AI adoption in the financial sector In 2023, $20.6 billion in the finance sector was spent on AI investments, right? So it's one of the biggest growing sectors that are adopting AI, and that's why I wanted to say you know how much and we're not just talking about chatbots, right. We're talking about using AI to understand fraud detection. We're talking about using AI to understand security threats and profiles, using AI around hyper automation of things, you know. So, yeah, I mean, you're in a sector that's massively adopting AI and really, for kind of all those non like, the workloads you're discussing are what I call productivity workloads, which are, I don't think, ai. The productivity story is not the main story with AI. It's. It's so many of the other components, particularly around the benefits and security and and really, you know, going to that point of you've heard the concept of agents that are coming out and what that will mean, which would be like taking a role and fully training them on, let's say, 20 years of data about what that role does, and then says go on, tell me where we could improve this role, tell me whether you can prove this in function. Tell me where we could take costs out of the process. Tell me where we could, you know, let's say, take a 20-step process and reduce it to 15, because, rather than manual looks up of data sets, you can now automate that type of functionality. So it's not just AI, it's AI with automation. It's AI with data sets that might not be just internal data sets, right?

Mark Smith: 28:44
Sainsbury, for example, in the UK recently in their AI pilot trialed using weather data on deciding what meat got placed in the supermarket. And so, if it was going to be a sunny day tomorrow, they put in meats that people generally eat in sunny weather more barbecues, sausages, steaks, things like that. They didn't put the roast dinner type things out, and what happened is they had a market jump in sales of meat because when the shopper went past, it was the stuff that they. The weather told them, hey, let's do that for today type thing. The weather told them, hey, let's do that for today type thing. And so this is where I'm seeing you know AI really in.

Mark Smith: 29:25
You know very unique use cases outside of the productivity story or the chatbot story, cause I think that's just the starting area of things, but I think that you know one of the actually banking use cases I had is at a checkout teller has somebody come up and said what's the fixed term rate for the next three months, year or three year? And when that teller goes to look in their system for that data, it has all the fixed term rate for years back, right, because in their SharePoint libraries and stuff it's got every year that they had the fixed term changes, et cetera. And the problem with that just a typical search is that you don't know if this is you know the context of now. Is that one? The AI needs to know that I'm a service agent and I've got a customer in front of me, so timeliness is important.

Mark Smith: 30:21
Obviously, the query is in regards to today's date, so it should automatically inherit that information, right? And if I've already loaded the customer on screen, you can look at it in context of this customer. So when you give me the answer, the probability of it being 99% or better, the fixed rate as of today for this amount of money is going to be highly accurate, right? So that's a very you know. Now, in that bank scenario I just gave you what that agent was doing. That teller was picking up a phone calling an internal contact center desk which was costing the business $6 million a year to run, which was costing the business $6 million a year to run and to ask that question, for them to go and do the search to find that information and make sure it was the most recent piece of information. Right Now you can see AI totally changes the game in that scenario.

Jay Wellings: 31:15
No, no, absolutely. And I mean for me it's coming back to the whole simplistic concept of this is another tool in our tool bag to help solve problems for our customers and it's kind of demystifying the what is AI? I mean, even within the conversation. I'm coming at it from one perspective, you're coming from in another perspective and it's really kind of identifying where the value prop is for the customer by leveraging ai and with this extra tool in our tool bag it's, I think when I had a conversation with a fellow at one of the recent microsoft conferences I went to, for me this is like the next technological jump. So we've kind of gone from, you know, back in the day where computers were literal kind of bits of card and we stamped out the holes and we put that in, and then we went from that to kind of you know, binary machines. We went for that to higher level languages like c++, and then we've gone from that to java and we've gone to the web, and now ai is just another one of those tools that we're just going to add to the stack and it's going to be the responsibility of the developers and the technological people that are involved in the process of building these software and solutions to understand where best appropriate to align AI and where best appropriate to use all of the other technologies that have come before.

Jay Wellings: 32:40
And it's important that we view this as well as not just in place of everything else that's kind have come before. And it's important that we view this as as well as not just in place of everything else that's kind of come before, because 100% AI is only as good as the data. So if we start ignoring data you know practices and everything that we've been building up for the last 20, 30, 40 years around quality data database schemas. You know data cleansing all of that good stuff we threw all that out the window because the AI is going to figure. Quality data database schemas. You know data cleansing all of that good stuff If we throw all that out the window because the AI is going to figure it out, well, the AI is only going to be as good as the data is trained on and then eventually the AI starts failing itself. So at the moment that's the way I see it it's another tool in the toolbox and that's.

Mark Smith: 33:17
You know, we've got to be excited to kind of embrace that and see all the options it brings for us. Definitely, I like it. Jay, thanks so much for coming on the show. Thank you very much, hey. Thanks for listening. I'm your host business application MVP, Mark Smith, otherwise known as the NZ365 guy. If there's a guest you'd like to see on the show, please message me on LinkedIn. If you want to be a supporter of the show, please message me on LinkedIn. If you want to be a supporter of the show, please check out buymeacoffeecom. Forward slash NZ365guy. Stay safe out there and shoot for the stars.