From Excel to AI: Automating at Enterprise Scale

00:00:01 Mark Smith
Welcome to the power platform show. Thanks for joining me today. I hope today's guests inspires 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 automation to AI and the journey that our guest has gone on. Today's guest is from Toronto and Canada. He's a dynamic leader with deep expertise in supply chain and digital solutions. He founded and LED a team that established an intelligent automation Centre of Excellence focusing on. Robotic automation or R? BA as well as a Microsoft Power platform. You can find links to his bio and socials in the show notes for this episode. Welcome to the show, Joe.

00:00:52 Joe Leo
Well, thank you for having me, Mark. Glad to be here.

00:00:55 Mark Smith
Good to have you on the show. I always like to start with food, family and fun so my guests can know you outside the tech discussion that we're. Going to have.

00:01:03 Joe Leo
Sounds good. So family, food and fun. I've been married 17 years this year. I've been with my wife for 21. She's a high school teacher. Two kids, 10 year old daughter and eight-year old son. My daughter is pretty much a carbon copy of me. She's more artistic though. Give her that. So she loves art and Lego, and she's an aspiring architect. So this summer I said we can look at getting like a 3D printer and getting her to start introducing the more tech and using design tools to build her own structures. Right. And my son, he's a natural athlete. Exact opposite of me, loves all sports, plays all sports really into soccer, European football. And that's the family.

00:01:38 Mark Smith
Epic.

00:01:53 Joe Leo
In terms of food, Italian, Canadian pasta and BBQ, keep it simple. Right. All different kinds of forms of pasta. We must eat it four or five times a week, like if it's not the main dish, it's part of the side, and you can do it so many different ways. So it's always, you know, interesting. You can fill it, you can red sauce. White sauce. Cream sauce. Butter. You know, all kinds of stuff there. And I do BBQ year round even in the winter. Got the set up here that I go year round when it snows. I just shovel a path to the barbecue and keep on grilling, right. In terms of fun.

00:02:27 Mark Smith
I love it. I love it. That's cool, you know? Yeah.

00:02:32 Joe Leo
In terms of fun, I do enjoy music. Classic rock, predominantly 90's rock as well. I play guitar, I sing, I do jam nights, open mic nights. So if I'm ever traveling or whatnot, I always find out where is the local pub that's doing like a jam session or a karaoke and and try to take part if if. If I can. I got a couple spots around the city. You like to go to.You know, sometimes it might go once or twice in a year and there'll be other times you just get in the mood and you'll go every week for 2-3 months, right? So it's just a passion of mine. That I do to unwind at the end of the day. One of the coolest things last year, though, cause like it's an evening thing, right? So last year I actually brought the family out. One night.

00:03:14 Joe Leo
There's a full band set up and I'm up there doing rocking in the free world and just for the moment I just sing love to my son because he's kind of singing along and I'm like, hey, come up on stage. So he comes up and he's singing the chorus and I got all that on video. So that's kind of like a proud moment. Last year, what was like? They know I go, but it's not, you know, typically family friendly after hours, you know. So we're able to bring them out and actually do a set with my. Yeah.With my son. So that was a lot of fun.

00:03:41 Mark Smith
That's so cool. I love it. I love it. Tell me about your first exposure to Microsoft Technology. How did you get involved in tech and particularly, you know, your journey into automation, how did that start? For you.

00:03:57 Joe Leo
Well, if we go to like an origin story, I would say Excel macros.

00:04:01 Mark Smith
Awesome, good, good.

00:04:02 Joe Leo
But in terms of in terms of power platform, it's really within the last couple of years. So I always used you know, excel pretty much runs the world before you know. Power platform if you will. But macros were the way to automate work. You know I'm not a developer, I can do some some simple coding, so I used macros allowed and access databases to to make myself more efficient. And then as time went on and I got introduced to the power platform a couple of years ago through the automation team, it was around 2020, right when COVID hit, you know, we were looking for something that was. Call it little more user friendly with an interface than the traditional RPA tools, and we're looking at a couple of different options. Doing a couple of different proof of concepts and it just so happens that we went with the power platform and it was no turning back after that.

00:04:56 Mark Smith
So give us a sense of a feel for the size of the organization and. The geography it operates it in, maybe the industry operates in, we can't necessarily talk about the name of the org, but what can you tell us about the sizing of the work you did from an automation perspective?

00:05:14 Joe Leo
Well, you have office users and store users like people in the actual retail stores grocery industry. We had over 10,000 licenses, right? So it was it was you know, at that kind of scale if you will. And we were also doing portals at the time. I guess it's pages now. But so we were doing canvas apps, model driven apps, power automate and the portals as well. So probably about 10,000 users is where we were at, but we were potentially scaling too close to 100,000 if we went full. Tilt to end state.

00:05:51 Mark Smith
Wow. Wow. And so this is a brand, obviously that runs coast to coast in Canada.

00:05:56 Joe Leo
Yeah, coast to coast. The idea and the vision at the time was, you know, every employee has an app, whether you're a part time clerk working. The department we a couple of the, you know pilots we were doing at the time was really targeted to like that end user on the floor interacting with the customer. So when you think of the scale of who's in the store and how many people there are, that's kind of what we were looking at was probably over 100,000 users would have been and state think something as simple as. Like this was the classic example, just to give people an idea of where you can go with the technology. Cleaning logs in a retail store, right? Yeah, usually. Once an hour on a clipboard, you're checking the time that you swept the floor. You mark down if it was wet and you put up a a wet floor sign. Imagine doing that in a simple canvas app and then uploading pictures at the same time. So if there was like a slip trip fall incident, or if there was an area that was wet and you mopped it up and put up a sign. You could take the picture to show the state you left it in, which could help with an investigation later if somebody actually. You know, trips or?You could augment it with power automate to send reminders to store managers and department managers. Hey. Yeah, staff supposed to do this every hour. It's been 2 since it's last been checked. You might want to go prompt someone to go check and you can escalate those as required based on the needs of the business. Right? So that was kind of the vision on where we wanted to go and we had a couple pilots in that space.

00:07:18 Mark Smith
I like it when I was living in London 5 or 6 years ago. One of the things that I started defining is like what a real world use cases. That kind of fit well with building an app and clipboard app. One of the the the the metaphors I used was was a clipboard app, right, which is anything that you would have a checklist. That you would go, you know, have a form that you know that had been designed maybe years ago and go through. You would often write your name on it, the date maybe a time, and then check, check, check, check, check what you needed to do right and The thing is, is that it's a very it's a very dumb set of data and that someone needed to eyeball it, and then you to have an app and the first time I you know when I was thinking about it back then was in the airline industry. Same type of thing. They have checks they have. Checks on air bridges. They have checks. On everything, right, there's there's so many checks being done all the time, but now because somebody has logged in, you automatically know exactly who it is that did this. There's no kind of. I can't read the. Name. You have a time stamp, so you know exactly when it happened. If you've enabled, you know any kind of location tracking, you know exactly where the thing happened. And so all of a sudden you're collecting a whole range of data points that your old method just never had. And of course, the beauty of that.

00:08:39 Mark Smith
You can then find patterns you can find. Huh? You know what repeats over and over is the pattern at an individual level as it at a location level is, you know, and all of it. You know, it's what I call digital. Exhaust starts to give you more data to make. With which was nobody specifically put their name, put their location, put the time date that was just inherited as part of using a digital platform, and so give me some examples. That was a really good one of the of the spillage. What other type of use cases we're starting to develop on?

00:09:14 Joe Leo
Oh, there, there's. There were so many and you just triggered my memory on on one because, like, checklist was like a whole category on its own that we would talk about. That was really, you know, if anyone listening, it says, you know, what's a good place to start with canvas apps checklists. You know, hands down a great way to get your hands on the tool and and and learn with it and get data that you didn't have. Before we had a fresh program, OK, it was a, you know, once a week throughout the course of the week, you go through Department, department store manager has a checklist. There's over 200 questions on the checklist and they would basically rate the departments in the store on how they were performing that week. So you imagine. 200 store managers. They're doing this every day of the week, you know 200 so. Questions and the method that was in place before we automated it was all in Excel. So I have 200 Excel files being emailed to a central person whose job was to aggregate them together and to yet another file to create some pivot charts that would then go out to leadership on the, you know, the Monday or whatnot. Wash, rinse, repeat. So we turned that one into a canvas app and then the beauty of it was twofold. We layered in power. The eye right and then that was the vision to say, imagine a dashboard at the back of the store where you got a power BI report that can now give you real time metrics on how the store is performing. You know in their region, in their banner, you know, get a little bit of healthy competition going all in real time. And you had the ability to. This was the other part that was, you know, observed and doing it that wasn't intended going. Those questions can change, right? So something can happen that would make a certain category or a certain area be more important, or one that didn't exist. You know, COVID is a classic example. Sure, there are items on that checklist that before COVID didn't exist, and after they do so, if you think of the old way, if you have to add in some new items, new some new categories. It's all manual, whereas in the cameras app you literally just update your list. Refresh and promote the app right? And it's all real time. We could put toggle switches on there that could say during seasons if there are certain things in a season. Toggle a section on or off right? So those that that was a really good one about incorporating power BI and giving real time metrics and extra there's a data and just to see the engagement from the from the supporting teams. Going crazy on power BI and splicing the data, they were getting insights to things they didn't have before because they just had an aggravated view, whereas now we're giving them individual answers to every single question so they can start to see new patterns that they didn't have before. So it was really, really good adoption on that one. And then there was. We were discussing this one before the the COVID tracking app. That was a very specific use case. You know, one of, but that is the one that kind of launches down the power platform to say, OK, there's something here. It was basically COVID hits, everyone shuts down, but you have to identify incidents where there was. The COVID exposure right and and people on a leave as a result and you had to do all that tracking and there?

00:12:21 Joe Leo
Regulations involved with. With that and everything else and certain rules you had to adhere to. So the beauty of that one was in a two week period we were able to meet with the business, get the requirements, design the app, develop it, test it, do user training and then do a mass deploy to 1000 users in one shot. In a two week period. That was our first app. Wow. So you know, when you think about what you can do with the technology, you know, never building an app before supported with Microsoft. You know, we had some resources available to to help train us and help guide us through that process. But to do all that work in two weeks, whereas prior to COVID, it would have taken. Two weeks just to do a kick off meeting. Right. So that was the one that really launched us to say we've got something here and then we started coupling it with with the our PA tools, right. And really where we got to some exciting stuff was.

00:13:05 Mark Smith
Yeah, yeah.

00:13:17 Joe Leo
Is integrating the power platform with the traditional RPA tools. So then we created a model driven app. This one took a lot of time, but it was pretty slick. We created a model driven app that we called like a control control station, right, like Mission Control. If you will. And we had a process that required manual work setting up data in the system, and there's various steps in setting up this data and different use cases that would apply. So then what we did is we said, OK, we're going to, we're going to create bots that will do the work to to create the data in the system. But depending on the scenario, we may want someone to manually do the work. And there are different parameters, if you will, combinations of parameters that would would dictate the sequence of which things need to be set up. So what we did is we created a model driven app that would track the progress of setting up the data. You know whether it be a new item, new store or what have you. And then through that app, they could trigger whether it would be automated or manual. So when they would do the initial upload, they would set up all the data to say. Here's what needs to go in the system. Here's the steps that would be automated. Here's the steps that would be manual, and then they would use that model driven app to then kick off the RPA tools or to notify someone manually to go do the work and you can manage it all end to end centrally, right? And that replaced the whole slew of what used to be. Emails and follow-ups and meetings and chats and you could all do it centrally through the model driven app as your command centre and then either executing people to upload data manually or triggering an RP a bot to do it automatically. Right. That one really took off and and it took about a year to to iron out and get going, but that was one of the more complicated ones that we had we had developed.

00:15:09 Mark Smith
That's that's. That's phenomenal. Right. As as in just. As you're talking, I'm like thinking ohh, this is where some of the AI story comes out nowadays. Around taking all that you know, e-mail chains that go round and round and and and all these kind of manual things that often have to happen inside business process inside an organization that can be automated and you are doing it. There with power automate and and and and ARPA tools and things like that. Which obviously has a massive impact in the repetitive mundane work for a lot of people, right? Allows them to focus on other things. Tell me, what are you seeing in the AI and BPM integration space and what future career opportunities you're potentially seeing around that space where you know we've had power automate particularly let's take that as a tool and we could talk about blue prism. We could talk about. Now, what's the other one that's quite common UI path right in in market. And when I look at where power automate is going with, it's now integration to AI and the ability to take dumb data and make it intelligent. What I mean by that is let's say ingesting an.

00:16:07 Joe Leo
UI path.

00:16:25 Mark Smith
E-mail which is, you know text. In using AI to understand what the text is and then perhaps update another system with that data without any human interaction. In other words, it knows it can because it can read it, it can understand the context of it, it can understand its application. You know any data around where it came from, how it how it got to be in the. In box at Inception Point so. So I'm seeing this convergence of automation and AI and the ability to create much more intelligent systems. You know error handling built in. By default telemetry data captured as part of that rather than the old way. You know an e-mail arrives, the person that is the interceptor. It's gonna copy and paste a bunch of data out of that, maybe into other systems. And then workflows are triggered off from that point. Bringing me back to my question, AI business process management integration and what are the career opportunities now you're seeing for folks to really up the game so much more with AI?

00:17:32 Joe Leo
I'm I'm really not sure how to answer that question to tell you the truth, because I feel like there's still a lot of unknowns, right? There's definitely something in that space. Like you can see it, right. And the way you're talking about it there, right. The Devils in the details. And I haven't had enough. Exposure myself to to, to to make a statement, if you will. On on predicting where the job opportunities are going to be, there's definitely something in that space and and time will tell what it morphs into, but what drives AI is data right? And and we looked we, you know we looked at so many different BPM tools and process mining tools and all that and and you know they all promised the world. And and there there are some really good use cases, but it's always devils in the details. So I think it's an area worth exploring. And I think as we start to get some use cases coming out of it, there will be lessons learned, which will then turn into, OK, where the job opportunities. So I think you know where we're at right now is more of an exploratory phase of. This is an area that has a lot of potential. Not quite sure where it makes sense and where it doesn't make sense. And then out of that we'll kind of come into the the skill sets and the jobs required to fully integrate them, right, like one of the examples. We'll pull from the RPA world is we were always told, you know, plugins with Outlook, right? You can, you can get a plug in for Outlook and you can ingest data from an e-mail and you can trigger a bot that way and use the data in the e-mail to to to trigger your automation and collect your data.

00:18:59 Mark Smith
Yep.

00:19:12 Joe Leo
It's not that simple and and even when you build it, the ongoing maintenance is is is not worth the effort, or at least for us it wasn't. So that is actually one of the reasons we started using the canvas apps because what we were saying was instead of using an inbox and an e-mail to trigger an RP A automation. We will insert into the process a canvas app to cleanse data right up front before it gets through the automation. We had a lot of success with that. We ended up with quicker solutions, better quality data and less maintenance required after the fact. So that was my experience with automation. And BPM, if you will. And I'm not quite sure what the lessons will be or what what have been so far when you augment it with AI. So I guess if I were to sum all that up based on the experience from the RPA world. You really need to have business knowledge, right? Those exceptions like for the AI engine to work, you got to have good input data, and if you're talking about business process BPM within a company, well that's proprietary to that company. You might not have enough data to drive the engine and it may not be good quality. So you know you can get some general things where there's shared across industry, but the nuances of a specific company on what they do. Is going to be specific to them and you're going to need people within that company who knows the nuances of our four walls in order to do error handling and exception handling and help define where those parameters are would be my guess.

00:20:51 Mark Smith
Yeah, yeah. Any other lessons learned from automation at scale like this?

00:20:58 Joe Leo
Nothing's as easy as as it seems. Right. The Devil really is in the details, and it's it's good to get excited up front and celebrate some quick wins. But you do have to think about the deltas, where things change and downtime and maintenance, right. Everyone is excited to build things, but then you tend to forget that you need maintenance. You know you go out and you buy the car and you're all excited. I got my new car and then you got to start getting the repairs. Done right and you have a maintenance. Yeah. Schedule it was the same thing with our PA similar experience with the with with the power platform, and I'm going to assume it'll be the same in the AI space. So you can't just build, deploy and forget. There is ongoing maintenance. An example I'll use is websites, right? We had a couple that would go in and.

00:21:44 Mark Smith
Yeah.

00:21:47 Joe Leo
And upload data into a website to submit information on a website.  It was a three day process that would run once a month. It's one of the first bots we built. And on day one, the bots working fine on Day 2 it goes down. You can do all the planning you want up to you could you could have all the regression testing and you can say I have a fully automated regression tool set that will run t -, 1, the day before the bot runs, just to make sure nothing's changed. But it's a three day process on on day 2. The bot goes down. Why does it go down? Because there's elements in the website that. Why did those elements change? Because the host of that website is trying to do anti spam attack stuff. So to make sure their site doesn't get attacked by bots like malicious bots, they change elements on the site which affects our bots which are not malicious. And then we have to go in and perform maintenance. Mid process. Yes, that's the stuff you don't think about. So you have to start layering into your development cycle maintenance time. And I have no idea what that's going to look like in a in in a in an AI augmented world, generative AI and all the rest. Of. That, but there's going to be an element of that. So so that would be a lesson learned. I would say to be. Aware of is. Think about after X is delivered, what is that maintenance schedule? What is required and how much effort are you putting in to keep it running? It might be quick to build and deploy, but then you find that the maintenance is not worth the effort and we took a couple of bots offline as a result of that. Didn't really experience it with the power apps. You know, power apps are really the very agile I could be on a call with with with a business partner and they're asking me to do a change on an app. They have an idea. They have an enhancement. It's 1/2 hour call while I'm talking to them, I'm messaging a developer in the background. Can you update this? Can you change that? By the time we're done, the call, you know, I already have the deployed and test and say, OK, you can go and test it. And they're like, what do you mean I can test it? It's that it's that quick, it's that easy.

00:23:50 Mark Smith
Yeah, yeah. That's cool. That's cool. I love it. Just as we we we draw to an end here. We're almost at time. I want to switch gears slightly and just tell me about the market conditions that you're seeing across Canada and a couple of things that I know we have discussed and I just saw you, you know, a couple of weeks ago when I was in Vancouver at the Canadian Power Platform Conference, Tell Me an importance around personal branding. At work, what are you seeing in this fear at the moment?

00:24:17 Joe Leo
Well, it's a tough market out there right now. Right. You're seeing a lot of companies laying off people. Unemployment is going up, and if you're looking at the LinkedIn is the number one place to go look for job postings, right? You go on the jobs section of LinkedIn and 15 minutes after it's posted, you see over. 100 applicants. Right. And you got that metric in front of you, it makes you feel bad. It's like it just got posted. There's already over 100 applicants.

00:24:43 Mark Smith
Yeah.

00:24:43 Joe Leo
So my observation is whether or not there's less jobs available. There's more people chasing the same jobs and they're good quality people. That are more than qualified, so it's it's a tough market out there and in terms of branding, that is the question everyone says you got to have a brand. What's your brand? I get, I get asked that question all the time, Joe, what's your brand? I don't know. I know what I'm good at articulating. It is not something I'm good at. But you got to have a brand and you know. Talking to people to get feedback on on, on how to how to sum up your brand. If you will is is, is worth is worth doing because how you perceive yourself and how you perceive the value you add might be different than someone else. But it's not just it's no longer just a resume, right? You don't apply to a job and and just say here's my resume. Here's my experience. It's it's what stands out about you. That is your brand. So find a way to easily articulate in a few minutes to someone why they should invest in you. Right. And investment could be either a job. Mentorship 15 minutes of their time to network. All of those things are part of your brand and you need to be able to articulate to someone and say you know, I'm applying to a job. Here's why I'm a candidate. You know, I'm in the interview. Here's why you should hire me. Hey, someone's reference me to for networking. Can I have 15 minutes of your time? Well, why should I give you 15 minutes your time? That's all part of your branding. Right. And it's very important to keep the conversations going because. Over 100 applicants. What's going to make you stand out?

00:26:13 Mark Smith 
Yeah, it's interesting. I think the whole world of job application like you said with a CV, I just.

00:26:22 Mark Smith
I don't know it it scales well as when you say that there are fewer jobs and higher number of qualified people going for those jobs and the fact that we are now in a global market more than ever. You know my customers that I am working with this specific week from New Zealand. Are located in New York and Canada. And I can do that because of creating a not just creating a brand because I think a brand is you have a brand, no matter whether you can articulate it or not. But it's like Coca-Cola, Coca-Cola spend a high degree of their money constantly telling people about their brand and I think. The brand needs to be known and it needs to be out there in the market so that people you are known for something but getting knowing takes a lot of work.

00:27:18 Joe Leo
It does.

00:27:19 Mark Smith
And it takes time. It takes time. And and but I think it pays phenomenal dividends and opens up massive amount of doors. My experience it it absolutely has. So that's why I I don't know. I don't know how the job market was going to be in five years as in from a you know will we even have a concept of a CV or anything like that. I just feel it's such an archaic way. Of engaging these days for jobs you know and like. Heck, I a year ago I left a big company where every time we advertised I would get all those. DVD's and then I've got to choose five of them. And how do I choose five out of all those? You know, I would often take the person's name and Google them and see if they appeared like if there was any external towels about this person.

00:28:08 Joe Leo
See, it's interesting and and similar exercises where I was before and you know different approaches, but you know. I was at my company for almost 20 years. I had a brand right? Don't know how to to to brand myself in the global market, which is part of the process here. Get your name out there. You said Google Joe see where he is. Hopefully the link for the podcast comes up. You get to know Joe. But I think it was on another podcast I was listening that I think it was you were on talking about similar. You know with with Google employees, you're working at Google and then all of a sudden they start doing layoffs and you find yourself saying, I've never had to make a resume. What am I going to do? I find myself in that space too, where it's like I didn't plan for being in a in a global economy where I'm trying to get my name out there on a global scale because I was inside my 4 walls and. Life was good and then you find yourself out here saying, well, what do you do? And you know, I see this guy, mark. He seems like a decent guy. I sent him a note on LinkedIn and here I am on his podcast. Right it it. It was. It was throwing a a Hail Mary pass. Right. I said to my wife. You go. I was going to send this guy an e-mail. I I saw him on a Microsoft tutorial type thing and seemed like a decent guy. Mark, I didn't even Google you. I just. I just added you on LinkedIn and said I'll send a message and then I was like, oh wow, he's pretty good guy.

00:29:30 Mark Smith
Yeah. So good. So good, Joe, thank you so much for coming on and sharing your story. I'm I'm looking forward to publishing this is. It's been great to get to know you and I cannot believe being on the other side of the world, we've even met in person in the short space of time as well. So it's been fantastic to meet you. And thanks again for coming on the show. 

00:29:53 Joe Leo
Yeah. Well, thank you for having me. It's been a pleasure.

00:29:56 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.