AI Needs Adult Supervision: Why Process Still Matters

00:00:07 Mark Smith
Welcome to AI Unfiltered, the show that cuts through the hype and brings you the authentic side of artificial intelligence. I'm your host, Mark Smith, and in each episode, I sit down one-on-one with AI innovators and industry leaders from around the world. Together, we explore real-world AI applications, share practical insights, and discuss how businesses are implementing responsible, ethical, and trustworthy AI. Let's dive into the conversation and see how AI can transform your business today. Welcome to AI Unfiltered. Now, for those listening to this, this is the first time we have released an episode of this. It's replacing the Power Platform show, and full show schedule will start in January next year. But this is a special episode that we're releasing right now. Today, we're joined by two powerhouse minds reshaping how we build and optimize digital services and processes. Please welcome BJ and Jim. to the show today. As always, we're going to make sure in the show notes are any links or any resources that we talk about today. We want to make sure that you have access to those. With that, Jim, why don't we start with you? Food, family, and fun. What do they mean to you?

00:01:30 Jim Sinur
Oh, food, anything that doesn't move. Generally, I'm attracted to Asian food myself. family. I have seven children, 9 grandchildren, and two great-grands.

00:01:45 Mark Smith
Wow.

00:01:46 Jim Sinur
We're all over the map in the US, but a number of kids have moved to Texas. Fun for me is to create music and to create art. And I use AI in both of those. And have had some success. I'm a balloted Grammy nominee for this year.

00:02:12 Mark Smith
That's amazing. That is so incredible. The first thing that caught my attention is seven children. I come from, I have seven siblings, so I know what large families are like. But that is fantastic. So what's the award that you've got coming up?

00:02:29 Jim Sinur
It's for best lyrics for independent rock song.

00:02:33 Mark Smith
Oh, man. That is amazing. Amazing. Well, welcome. BJ, how about you? Food, family, and fun.

00:02:40 BJ Biernatowski
Yes, thanks. Thanks, Mark, for having me again for your show. As far as food, I am being a beta tester of my son's AI recipes. You know, he's doubling the young generation, 14 years old. They're doubling into cooking. And, you know, recently it was an AI mac and cheese that was a dish that we've been all trying to sample, critique, and, it's amazing. Our young generation is using AI to soak up the world. They're looking at the lens through ChatGPT's co-pilots. And it is both, I think, fantastic, but also a bit scary, right? The food, the mac and cheese, turned out amazing. You know, we've used the type of pasta, kata pavi, blend the, rolling pasta I've never heard of. So it pushes me to shop for ingredients. But I think this is in the world where we're going to live in. As far as family time and our family, it is all about college shopping. We've got another offspring. We have two kids and my daughter is a senior high school and she's all about college shopping. So we just came from a 900 mile across the state track, shopping for colleges, looking for colleges, trying to figure out, you know, what is the right decision to make to launch to launch our kiddo into her next chapter. So yeah.

00:04:14 Mark Smith
Amazing.

00:04:15 BJ Biernatowski
On the sideline trying to cheer, coach, provide feedback, but it's not an easy decision to make. It's not a good decision to make.

00:04:24 Mark Smith
The challenge is a lot different in New Zealand. When you're as small a country as us, you don't have the range of colleges, et cetera, to select from. So it's a much easier decision. Just to set the scene for those listening, I was a technical reviewer of a book in the last year called Practical Business Process Modeling and Analysis. We'll provide links to that book in the show notes for this episode. And I'm lucky today to have the authors joining me. And so we're going to discuss their backgrounds and particularly how this book came about. Business process management is something that I came across would have been probably 15 years ago, I was working on Dynamics CRM-based solutions back then with Microsoft. And our organization had an entire division dedicated to this topic, where if from financial institutions to government agencies around business process management and optimization was very much front and center of what we were focused on.

00:05:36 Mark Smith
So Jim, starting with you, can you tell us a bit about process modeling and how that fits into a full digital transformation story for an organization.

00:05:49 Jim Sinur
Yeah, I'd be glad to. It fits in my mind in two particular areas. One is if you have a chance to plan a process, process modeling is great for coming up with a model of what you expect and maybe some planned exceptions. And that's kind of a static process model. And of course, there's also case management, which is, a variety of process that's a little bit looser. In that case, the process model is used in an audit basis. So the model is created from the data to watch what happened on the case, to see and analyze, to see if you can do it better. And these two things are very helpful when you're starting to incorporate AI, you want to do what you can in terms of plan. Now, the process may be smaller. It may be a snippet that an AI agent calls up and says, you know, take this action. Because process is a lot about action. And as AI moves from inside a process to outside a process, it's going to be making more of the decisions. And consequently, the process needs to be modeled. Well, the danger in AI is, let's say you're smart enough to put in constraints and guardrails, and you're smart enough to put in goals, and hopefully, the AI sticks with the goals as it collaborates with other dynamic agents. The idea of a process model is for the action part of it. And then afterwards, what did the AI do? You know, was there a bargaining agent that called in other agents dynamically? Was the customer given the right information? Did you analyze their voice emotions as you were servicing them? Did you listen to their suggestions? All of these things can be modeled. And this process model also integrates with other models. So there are decision modeling, there's basically context modeling, and all of these play a role with the process model, but the process model is a core piece. And if you really want the triple crown of process, and that's increased revenue, decreased cost, and happy customers. That means that from a customer's viewpoint, process modeling is exceptionally good at understanding the journey that a customer takes. And so I see the convergence over time of the customer experience modeling and the process modeling, the context modeling, and the decision modeling in order to do your digital transformation, whether you do it incrementally or whether you do it big bang.

00:08:59 Mark Smith
I just realized as we've jumped into this, is that how do you explain process modeling to a layperson as in, and just, and perhaps if you gave a real life example of where a process might be used and The difference, I suppose, between a workflow and a process, being that, when I think of it, I think of the financial sector, you might have a mortgage that runs over 25 years, and that would be an end-to-end, there'll be interest rate changes, there will be renewal period. It's not that it's, my thing would be a workflow might be a lot shorter, just does a discrete thing. But how do you explain the difference between processes and workflows?

00:09:46 Jim Sinur
Yeah, workflows, I think you've got it right. Workflows are usually shorter. They're generally human to human. Well, there is some interaction with transactions along the way. A process, I believe, has a wider scope. Now, as AI uses process snippets, then again, it kind of shrinks down. These snippets are a lot like automated workflows. They're a lot like, you know, bots. and they're encapsulated. So to me, a process is, the difference is the scope in which it covers, the skills that it needs, the resources that it needs, or workflow. It's generally humans and humans and can transactions.

00:10:35 Mark Smith
Yeah, that's brilliant. That's brilliant. The folks listening to this, a lot of them will have consulting backgrounds, and they might have touched on the edges of process modeling. And they're often called into companies to map maybe business processes to a technology platform. And I feel that there's a whole area, as in, when I first came across this space, I didn't realize how big it was. What was involved, as in from everything from how do you capture a process, how do you diagram it, how do you then monitor it over time? How do you explain that, as in that this is a discrete skill set in its own right, no matter what type of software, AI technology you work with. And it's very much proven. It's not new by any stretch. It's been proven over a long period of time. And so it's an additive for folks in that consulting space. How do you encourage people to approach this? What are the steps that they should take? What should they be looking at to develop their skill sets in the space?

00:11:49 Jim Sinur
Well, just like any other you know, visual representation of sequence of processing. You know, there's precedence involved and there's parallelism involved. If you have a picture of that, it helps humans to understand. When you dig deeper and you start using, you know, standards like BPMN, not only can humans understand it, machines can understand it. And so the idea of the model is for understanding. Initially, it was for communication purposes and to record it as it evolves and use it with business users to say, you know, this is how we think your work flows. Now, in some cases, work doesn't flow. It's more dynamic. I wrote an underwriter workbench for an insurance company. And while there were individual flows, They had to make choices and it was more off of a workbench. Well, I've got a, I'm selling life insurance and I have somebody with high blood pressure. So I'll go down the high blood pressure path. I have a person that has diplopia, which is usually a sign of diabetes because you've got double vision. how do I do that? And this is where we started incorporating AI so we could have the low-level underwriters handle what senior underwriters and doctors did.

00:13:22 Mark Smith
So tell me then, where do you see those kind of, the intelligent processes incorporating AI? How do you explain that? What's the lenses that you look at it from?

00:13:35 Jim Sinur
It's going in three phases in general. Obviously, there's going to be crossover. But the first level of AI intelligence is to have bots, agents, or intelligent decisions reporting to the process. So the process is managing the Uberflow, as somebody might describe it. And it becomes a support, whether it's an automation piece or whether you're creating dynamic data and doing some late binding to say, well, you know, if this data is over X, I want to go this way. Well, so let's use the idea of underwriting. Let's say you're with an insurance company and any policy that's under $100,000, it's going to be jet underwritten, you know, straight through processing. The rest of it requires humans. you're in inflationary times. And do you want to put that 100,000 in and embed it in your system? No, you want to late bind it and make it happen based on doing maybe some machine learning and do some late binding so that number varies for jet on rate. That's probably the simplest example I can use to explain it.

00:14:55 Mark Smith
And then the other phases.

00:14:59 Jim Sinur
The second phase is where you have resources assisted by AI. And that example of underwriting, doing diplopia and figuring out the 53 reasons for it. It may be, you know, you're drinking wine and that's how you're getting it. So the assistance of humans. And then the AI goes outside of the process. So it starts making decisions. It starts brokering. So, life insurance is not an example of that because it's pretty static. But let's say you get into the notion of, let's say you have a drone inspecting oil pipelines. And that drone is an agent and it's making decisions. It's making decisions who to send out based on where they're based and where they're available and what kind of crews are available to plug the leak. or you could have, a pollution issue. there's an example of where the, smarts are outside the process. Another example is farming. We've seen this particular example in Australia. And we see it a lot in the southwest of US where I am, I'm in Arizona. You have, you know, watering machines that create these like kind of like crop circles. And it used to be somebody had the press button, make that happen. Now they automate it and it's automated. Well, that's okay if the automation is right, because you really don't know whether that soil needs it or not. So then they started throwing out pills, bots out in the field that would sense the temperature, sense the soil moisture levels, know what kind of crop was there and would send out the proper mix of water and fertilizer. So there's an example. And by the way, they took it further. They incorporated with water. water runoff models from, you know, from local mountains and weather forecasts. Hey, we were in a severe drought, so we're gonna see what we can do. And then it led to creating specific kinds of plants that would work best in that scenario. But the bottom line was 40% increase in yield. And when we've got a lot of hungry people, And there's an example where the smarts is going outside the process and the process is evolving.

00:17:40 Mark Smith
How much, from the traditional background of BPM, how much is AI changing the game as in that you're seeing in organizations and businesses? Because the example you gave there and the crops is a very sophisticated and powerful use case, right? I've seen it in the mining sector in Australia as well, where the analysis is literally having per day multi-million dollar impacts on their business by incorporating AI into their processes. How much are you seeing AI really taking the whole business process arena to the next level?

00:18:23 Jim Sinur
Well, what I see it on mostly is speed. So how fast can you react to changing conditions? Back in the old days, things didn't change a lot. In today's world, things are changing. So you can see the leading edge of this thinking, and it's not a topic I'm real happy about. I wish the world was at peace, but in war kinds of situations, the governments are using, you know, dynamic processes, dynamic feedback, real-time configurations, swarming weapon systems, and deciding what weapon to use in what situation, depending on the threat level. So there's a, it's combining with situational analysis. So now, you've got AI, helping you with the situational analysis because it can pull the data so fast. Situational analysis and SWATs, it took months for companies to do it. They can do it in less than an hour. And if you're part of the government that's involved with war, they're doing it in nanoseconds.

00:19:27 Mark Smith
Yeah, phenomenal, phenomenal. BJ, tell me, what are the pillars of success when it comes to DTX? What's been your view? I know you've recently been to Microsoft. What have you learned there?

00:19:41 BJ Biernatowski
Yep. Before I start talking about the success, I think one of the things I want to highlight is, I don't know if your listeners know, but Jim has been one of the people analysts at Gartner who actually defined as the space of BPM.

00:19:56 Mark Smith
Wow.

00:19:56 BJ Biernatowski
Not sure what it was, but he has coined the term. So I think we're very lucky to be able to connect, the practice of business process management with all of the excitement and the new wave of AI. And I think it was super, super great for us to be able to have a thought leader like him kind of set the stage for a lot of our chapters in the book and fast forward into the future because I believe that Jim has a both this forward-looking visionary, you know, I see into the five, 6, 7, not sure the areas, but he also brings the wealth of experience from, hey, we've seen it enough, right? For us, it is AI, but Jim has probably seen a lot of other transformations, right?

00:20:42 Mark Smith
And what's amazing in that is still be forward thinking, rather than, because you get a lot of scenarios where people just want to defend the ground that they've got in the past. And so yeah, it's massively refreshing to see that adoption of the new and going, how can the two work together and advance more? It's awesome.

00:21:02 BJ Biernatowski
Absolutely. As far as the pillars of DTX, right? So I've worked with, aside from Microsoft and a few other very, very large organizations, healthcare, retail, technology. And over time, you start noticing very similar trends, right? Topics like change management, importance of low-code, having access to inexpensive cloud ecosystems, empowerment, connecting process, and business architecture. quality data, right? We carve out seven pillars. It is a framework that we believe, and we're putting it out there based on the experience with other transformations, will continue to operate in the same fashion. Whether it's, you know, original workflow automation or whether it was RPA or intelligent BPM or now AI, this circle of progress continues to move. And if you don't have these foundational concepts, pillars in place, you're going to be struggling, right? You need those established. Whether you do it top down or you do it inside team, grassroots, up to you. We spoke a little bit about the types of transformations and how these waves usually happen. But, you know, they are foundational. And, you know, our, the experience of from these transformational programs was you got to think about them up front or else you're going to hit, you know, a curb. Yeah. Because you're going to start going fast. You're going to start going fast. And if the culture is not conducive to support all of these changes, having so many different changes, technology, again, change management, changing how people do the work, right? throwing in a lot of new concepts, feeding it with concepts like BPM, Lean Tech Sigma. That's a lot of stuff to throw at folks, right? And how do you connect all of these? You need some stability, right? And those pillars, they do create that.

00:23:07 Mark Smith
 Yeah. So if people were to pick up your book, Practical Business Process Modeling and Analysis, would that give them the, you know, let's say they're working in the AI space, they're working on helping organizations in their digital journey. Will this kind of give them the foundations that they need right through to kind of where they can go further in their study of this topic?

00:23:35 BJ Biernatowski
I believe so. However, our book is not per se about AI. It is about preparing yourself for the wave of massive AI transformation. And I believe that this whole process thinking BPM and connecting BPM and continuous integration and transformation with AI is one of those missing blocks or concepts that is a, you know, just, it's a corner no one's looking at, right? So if you're forgetting about that, how are you going to be able to verify that your AI is not lying? You're going to spin up all of these agents that are then starting to choreography work and how do you know they're doing what is supposed to happen, right? How do you check them? Who's checking all of these bots?

00:24:25 Mark Smith
And we're seeing in this in the news, right, where people are, business are handing off too much control to AI. and almost leaving the fundamentals. I mean, right now, Deloitte in the last 48 hours have been in all the papers because they're rolling back half $1,000,000 worth of work because they had AI write a report for a government agency. And it looks like there was no validation, no checking, and a lot of the errors were because none of that kind of work was being done. And I feel that when I read the book, it really allows you to make sure that you get your foundations right so that when you bring AI to the mix, it's from this very rock solid foundation of understanding so that you don't hand over the thought, the rigor that's, I suppose, that it's happened over the last 15 plus years in this space, is not thrown out because let's make AI do it all. Let's let it do all the automation, let it do all that process. And I feel that, although you say, you know, it's not specifically about AI, I found it a very good leverage point for making sure that foundation was robust As we go in and change business processes, et cetera, at scale, we should not forget the work that was done here to get us to this base level and just expect AI to take over. What are your thoughts, Jim?

00:25:50 Jim Sinur
I think it's important to understand that AI needs adult supervision.

00:25:56 Mark Smith
Yeah.

00:25:57 Jim Sinur
So, you know, the process is the adult supervision for the business. So let me give you an example in my field. I'm creating music. I could catalog 20 songs. But nobody wants to watch an old guy play a guitar on music videos. I wouldn't, frankly. And so I used AI to create all of my music videos. So I would give it a script and I'd come back and I'd go like, What? This doesn't look right. And then you got to give it a new script. You got to give it better description. You got to start talking to the AI, supervising it. And even after that, there's some scenes that I had to keep where somebody had seven fingers, you know? Yeah, yeah. So, you know, you need adult supervision for this. And business needs to understand that it's the customer experience, which is a form of process, and processes give you those boundaries and direction for AI. Now, once AI goes out, then you've got to include goals and boundaries.

00:27:13 Mark Smith
Yeah. You know, what strikes me right now is that it's no longer business as usual for a lot of organizations, right? They are, it's not that AI is just a tool set that we can bolt on to whatever we're doing. It's really re-engineering in many times the entire process to do anything. And so we are seeing, particularly in the Microsoft space, a lot of organizations going, hang on a sec, it's not just take what we have and bolt AI onto it. Let's roll back and go, now we have this tool set, how would we re-engineer our organization going forward with that in mind? And that's once again why I felt that this book was so timely at this point, is that a lot of organizations are embarking, or if not started to, this journey of large-scale re-engineering of how their businesses run in the context of this tool that has a similarity to electricity and that it's going to affect everything and everything that we do. But both of you, who wants to take it? As you look at the future, what are you seeing as the impact in this space inside organizations?

00:28:29 BJ Biernatowski
I think you hit the nail, Mark. I think Microsoft has been an example of a company that very aggressively empowered the transformation internally, right? I think they went, the company went the fastest out of any company that I've ever seen, right? And it also created a situation where you had a lot of people who might have not been prepared to fully drive it because they haven't been exposed to business architecture, process architecture, right? There's these different camps. And one of the reasons we wrote the book was to bring all of these sometimes competing camps. You have people from tooling, Dynamics and Copilot, right? They're thinking they're doing their own transformation. You have process architects, you have enterprise architects, you have people in operations. Everybody's talking about the business process, but are they truly talking about the same business process, right? How do you bring the same They equalize the playing field so that metaphor of the business process means the same thing. Sure, they could be using different tools. Maybe you're going to be doing process advisor to do the mining to accelerate discovery. but it is still part of your continuous improvement, right? Or maybe you're going to be utilizing Copilot to spin up these automated workflow. Hopefully, you're going to know what they're going to be going after, right? But again, the metaphor of a business process, what is a business process? My experience talking to a lot of professionals, there's a lot of talk, not a lot of understanding, and the depth to truly, truly know where these efforts are going. I don't know. What do you think, Jim?

00:30:09 Jim Sinur
Well, I just finished a video on AI successes. And the reason I did that is right now, all we're hearing is failure here, failure there. This didn't work. That didn't work. Why didn't it work? And I really got a bit irritated and volunteered to do something on successes because I had 15 or 20 that I knew about, two or three I was personally involved with. And what you find is you have to, as a company, decide what kind of risk you're willing to take. Do you want, is your culture the kind of culture that likes to break things and be innovative? And I worked for two really large companies. One was one that wanted to incrementally get better because they believed that automation and AI would save them money. and allow them revenue opportunities and keep the customer happy. Another one I worked for was like, let's break it, let's start it, let's throw what we can against the wall and see what sticks. They had a unit that just did advanced technologies. And so knowing what your company is willing to put up with and who has the control, what I find in most organizations is the accountants' and the lawyers' control. Yeah. And because of that, you know, they're looking to get bottom line stuff right away. And that's a short-term strategy. The nice thing about processes, you can do that and you can grow and get it better and bigger and more context over time. Now, my suggestion to organizations is that if you're saving money on your processes, take a certain amount of that and use it as an investment mechanism so that you can try. And maybe you don't like throwing 20 things at the wall. You throw one of them that you think is important, not what somebody said to you out on the golf course as an executive.

00:32:27 Mark Smith
Yeah, I like that. I like that. That's about all we've got time for. Is there any If people want to buy the book, is there a special code or anything that they should use in using that? We do.

00:32:40 BJ Biernatowski
We do have a special code for the listeners of your great show, Mark. It's going to be included, I think, in the podcast itself, direct link to Pact website. Pact is an amazing, amazing publisher. They publish a lot of books about Microsoft technologies, a lot of other technologies. So we're really grateful to be supported. by their team. And I think the coupon provides 20% discount on the book we sell in Amazon. So there's a Kindle edition. For those of us who like Kindles, I'm one of them who love Kindle. I'm A Kindle generation. That is my fun activity. I didn't mention exploring the books and never ending access to magazines. I am a magazines generation.

00:33:27 Mark Smith
Nice.

00:33:28 BJ Biernatowski
So yep, we do have that.

00:33:30 Mark Smith
Excellent.

00:33:30 Mark Smith
Well, thank you, gentlemen. It's been so good to have you on. And Jim, your history in this space is absolutely phenomenal. So thanks for taking the time. Thank you.

00:33:43 BJ Biernatowski
Thanks, Mark.

00:33:43 BJ Biernatowski
Thank you, Jim.

00:33:45 Mark Smith
You've been listening to AI Unfiltered with me, Mark Smith. If you enjoyed this episode and want to share a little kindness, please leave a review. To learn more or connect with today's guest, check out the show notes. Thank you for tuning in. I'll see you next time, where we'll continue to uncover AI's true potential, one conversation at a time.