#538 Process Intelligence & AI: Liam O’Neil on Building Smarter, Scalable Organizations

[00:00:00] 

Mehmet: Hello, and welcome back to a episode of the CTO Show with BeMe today. I'm very pleased. Joining me from the uk, Liam O'Neil. He's the MD of BPMD. Liam, you're gonna tell us, [00:01:00] you know, a little bit about you, your background, your journey, and then we are gonna start discussion from there. Like, it's very interesting topics we are gonna discuss with you today.

Uh, just teaser for the audience. We're gonna talk about, you know, process mindset. We're gonna talk about process mining, we're gonna talk about AI of course, and many other, you know, side topics as much as we can do with Liam. So Liam, the floor is yours. 

Liam: It's a pleasure to be on your show. Thank you very much for inviting me to be here and to get to speak to your great audience.

So, uh, my name's Liam and my journey started, uh, coming outta university. I shade over 10 years ago with no idea what I wanted to do other than the fact I wanted to work for a startup. And I ended up falling into consulting and I fell into process consulting, uh, primarily about process modeling. And I've seen over the past 10 years the evolution of the space from being very process model, drawing centric to something that [00:02:00] became very data centric with process intelligence and process mining.

And increasingly is becoming more and more easy to use, democratized and feeding into the whole AI and hyper automation space as well. And over the course of that time, I stayed with the same startup. It's, uh, it's things have transpired such I've taken over the company and I'm never imagined Director of BPMD.

Mehmet: Cool. And thank you again, Liam, for being here with me today. So, uh, you kind of took the first question from me, but uh, you know, if you want to tell us like maybe among all the experiences that you had, like one story that best captures what great process design can transform a business. 

Liam: Definitely, definitely.

I think. A lot of companies have a lot of very clever people. Mm-hmm. And the drawback is the often these very clever people stay in their teams, stay in their [00:03:00] functions, and you can end up with these little silos that are fantastic. Uh, some brilliant ideas, some amazing problem solvers. And one of the most, uh, telling programs I did was an e was with an energy company a few years ago where they had some fantastic people in the finance team, really tech savvy, really innovative.

They had this massive backlog of invoices that were growing every day. They would be mismatched and had to be manually, uh, reviewed and brought and tagged to the right invoice. And they were midway free building, uh, automation bot an RPA bot to help accelerate this matching. And absolute tech leaders in that space.

This was back at the infancy of RPA and they really wanted to innovate, but what I was telling was they were trying to solve the finance problem, but they weren't looking upstream. They weren't looking at what was happening through the logistics process. They weren't looking even further upstream at the procurement side of things.

All those mismatches, they'd worked in their box and come up the best possible solution, but. [00:04:00] Ultimately it wasn't their fault. It was a little bit of an issue upstream, just a data capture issue. And this data capture issue at a point of, uh, a procurement team at raising that PO was causing all of these mismatches, hundreds of thousands of hours of manual effort going in and resolving them and all this investment in this innovation budget and making this really interesting automation tool.

And I think that's what's telling is companies, as I say, very intelligent, but working in silos. Sometimes what we're lacking is the ability to step back, look at the broader process as it runs across teams, across the business, across systems. And sometimes just trying to make that easy little fix upstream that solves a hell of a lot of pain downstream.

Mehmet: Cool. Um, one of the things, Liam, that I hear from people sometimes that, um. Especially transformation leaders. Right. Um, so the word process is cinema to [00:05:00] bureaucracy, right. Uh, bicep sometimes I, I used to think the same. Um, so how, why do you think process has such bad reputation and how we can reframe this, especially in the, you know, AI era.

Liam: Definitely, I've worked with so many process teams where they have this moniker of being bureaucrats, of sitting down with people absorbing hours and hours to refine a model, revise a model, make it clean, but without doing anything useful for the business and. I think it breeds outta this place where a lot of process programs end up sitting under quality teams.

They sit under the head of quality, quality director, and really after transformation, after you've moved ERP, after you've changed your case management system, whatever it is, you have this body of documentation and it moves into the quality teams so they can use it for audits. [00:06:00] And so process becomes a tool for compliance, for audits, for ticking a box.

And that's really useful. For the rest of the business, don't really get anything from that. And so you end up with process models, but being created for the sake of process models, and that's the first thing that we always try and challenge when we work for an organization. If you're doing process for processes sake or just to tick a box, you might as well not do it.

It's gotta be. You've gotta have a really good reason for doing it. What problem are you solving in the business? Passing an audit is one, but you've gotta solve more than one problem. You've gotta do something that when you are asking someone for their time, for their investment, they're getting something back.

They're getting back improvement. They're getting back change in the way of working something that makes their life better. So if you want people to not see processes, bureaucratic, you have to make sure that it's delivering for them in some, in a way that actually impacts and makes their life a little bit better.

Mehmet: Cool. [00:07:00] Um, from your perspective and your experience, uh. How are you seeing the emergence of modern process team look like today? Are you seeing like maybe new roles or scales that emerged that didn't exist? Maybe, I don't know. Before the hype of the AI, at least. 

Liam: Definitely, definitely. I'm actually writing a white paper on this, uh, the moment.

Um, I think there's a lot of structures for process teams. As I say, sometimes they live under quality managers, sometimes they live under it. The best structure for your process team is to not be a process team. 

Mehmet: Mm-hmm. 

Liam: It's not to have a separate. It's not to have a separate group of people who have sat there just looking at models, maybe looking at dashboards.

Instead, what you've gotta do is recognize all you're trying to do with process is change the business, make it better. But you're not the only person, the only team who's responsible [00:08:00] is to make things better. There's this whole cohort of change makers from your customer experience teams who are probably, you know, sitting upstream in marketing and sales to your continuous improvement teams who might sit in operations.

Uh, maybe you've got that sat at a group level and then the whole oversight and delivery teams like PMO and change management. And increasingly obviously you've got data intelligence, data management, and your AI teams as well as a whole product management cycle. And. These are so many different hats you can wear and different types of way of approaching change if you wanna do it effectively.

They all have to be working together. You have to have them sat in the same change maker function, whether it's a dotted line or a solid line that all design structures, that's an open question, but we've gotta be working together. There's only so many people who can actually change the way the business works.

You wanna make sure that those people are all pulling in the same direction. You've not got four or five, six different teams, all clashing heads and just causing this little bit of a traffic jam. [00:09:00] 

Mehmet: Cool. Now, what have you spotted, uh, when it comes, you know, of the kind of major mistakes companies make when trying to scale internal process teams?

Liam: Yeah, yeah. The biggest one is losing sight of the most important piece. 

Mehmet: Mm-hmm. 

Liam: Which is the problem they're trying to solve. Model all the processes, set up every single dashboard, um, you know, create perfect enterprise architecture. All these ways of creating a view of your business, creating visibility.

We're all useless if they're not gonna be used and. When it comes to scaling, it's not about trying to get comprehensive coverage of every single process and every single function. It's about choosing the 2, 3, 4 problems you are gonna be very good at solving. Maybe it's scoping which AI [00:10:00] innovations you're gonna bring in.

Maybe it's. Uh, passing, uh, risk and compliance, uh, getting your risk and compliance registered up to date for a function. Maybe it is something about audit, but what are those handful of services you're gonna offer? You get amazed at those ones, and as you're scaling, you can slowly add in more services, but don't try and do everything at once.

Get very good at a service. Scale that across your organization where the problems of earth to be solved and then grow through that problem, problem led approach, rather than trying to think, you can just. The ocean, 

Mehmet: uh, Liam, um, while doing my, my search and preparation for today's episode, um, I've seen something which attracted my attention.

Um, and I think you, you're the best today to, to explain to me and to the audience about that, which is, um, two terms. I've seen it before, long time away, so maybe I need a refresh for my memory. [00:11:00] Um, so difference between process mining and task mining, and. Uh, what are they, right, how they differentiate from each others and maybe people in, in, in, in your space.

They would, might have knowledge, but I like to cover it also in the context of business in general and how they can, uh, how these two, uh, concepts process mining and task mining complement actually each other. And if you can give us like also some, uh, real life example, that would be great Also. 

Liam: Perfect.

Perfect. So I mean, very briefly, the technical view is process mining is looking at event logs. So when key things happen in your process, stringing those together and aggregating them. So you can see trends and task mining is looking at key logs, what you are doing, what an individual user is doing on their desktop.

To put that in a context, it's a bit more understandable. I like to use the analogy of a train station. So process [00:12:00] mining is like you are the, uh, uh, national rail. I'm not quite sure what, what that's called in the US or elsewhere, but in the UK we have national rail overseeing our rail network. And process mining is looking at your rail network and seeing that the, uh, 10 o'clock train from London leaves, uh, every day.

On average, five minutes late from London and gets to Manchester on average, 20 minutes late, and over the course of a year, two years, three years, five years, 10 years, you can see an aggregate. It's always happening. It's always gonna be late. It calls into the station at this time. It leaves at this time.

And maybe what you're seeing is that train from London to Manchester, it stops midway. It stops in Birmingham. And even though it leaves London a bit late, when it leaves Birmingham, it's a lot late. I mean, it catches a pivot as it gets to Manchester. So process mining tells you. Hey, look, looks like there might be a problem at Birmingham.

Something's happening at that station. That's meaning your train's late to leave. That's really what it's for. It's the spotting weather issues. Task [00:13:00] mining on the other hand, because it's looking in more detail. That's like setting up a video camera on the station, on the platform in Birmingham, and it watches the train come in, watches the people get off it, then sees that, oh look, there's a faulty door.

The faulty door, that's not closing and that's taking 10, 15, 20 minutes to be resolved. That's why we're getting the delay. It gives you the detail on the platform. It doesn't tell you anything about the broader journey, whether it's leaving London on time, getting to Manchester at the right time, whether it's, um, doing anything outside of the context of that station, but it gives you really rich information.

So practically where you'd use these in an organization is process Minings typically done across your big end-to-end processes, looking at procure to pay, audited cash, operate to maintain those real core pieces of your business. Buying things, moving things, building things, making things. And it lets you spot trends.

It lets you spot, uh, where you've got [00:14:00] problems. A real example, uh, with a tech company, we work with a tech and. Uh, hardware reseller. 

Mehmet: Mm-hmm. 

Liam: They had looked from quote, to order to cash and they looked across all of Asia Pacific and we're looking at quotes and on average they wanted to get a quote. As in one day, India was doing a day, China was doing a day.

Uh, Singapore was doing a day. Taiwan though was taking a week. 

Mehmet: Mm, 

Liam: and you're never gonna get that from process modeling. You'd have to speak to every single country. You'd have to do a ton of analysis. But process mining tells you, Hey, look, there's a problem here, Singapore, um, Taiwan's pretty slow at quotes.

And then, you know, you have to go on site and do a bit more detailed analysis there. Task mining. On the other hand, you really use that. We've got high volume jobs, your call centers, your support centers, where you've got people doing a similarish task day in, day out. There's a bit more rigor, a bit more regularity.

You need to set that upon their [00:15:00] desktop. Uh uh. A medical and insurance organization I worked with set one up recently to analyze case handling and it was to look at, you know, what's some of the trends, what go, what denotes a good case, a long case, a closed case, what systems they're using, what databases, and there's so much rich information.

There's 25,000 calls a day. You're not ever gonna be able to get to the bottom of that by talking to a handful of people. But by using task mining, you can see exactly how long on average these things are taking, how long it's taken to solve case A versus case B versus case C. And worse, some of those really detailed problems are actually coming to the fore.

Mehmet: Liam. Yeah, this is very detailed and uh, you know, it needs a lot of attention. Out of curiosity, can ai, whether it's a gen AI or any form of AI, help actually in, in, in modeling and mining for the processes? And, uh, are you [00:16:00] currently using it by any form? 

Liam: Definitely, definitely. It helps cross the board task manage.

It's super useful because. A few years. I remember doing it a few years ago in 2016 or so, and you got the data back and it was terrifying. It was absolutely impossible to get anything from it. Pages and pages and pages and pages, and by the time you've found an insight, you've gone partially blind. But now what you've got is some of the AI engines that sit on top and just speed up interpretation and it's not perfect, but it's a lot, lot better than anyone could get to without investing significant time on the modeling side of, on the process mining side of things, the problem with process mining always is.

The so what? 

Mehmet: Mm-hmm. You 

Liam: found a cool insight. You found Taiwan slow. So what? And it's bridging that gap of being able to, A, solve problems, but B, [00:17:00] communicate that there's even a problem. So about, think about, you know, finding a Taiwan one. Obviously they use consultancy to help 'em find that. But then the internal team is, it is difficult to sometimes really get into the data and get those rich insights out, and AI is helping to make it easier to interpret the data, find those issues, and then it's making it easier to think of very interesting solutions to some of those potential problems.

Mehmet: Liam, I'm not sure if I'm asking the question the right way I should ask, but, um, if, if, if I'm starting. To put some, uh, process design from scratch that I've seen people who's saying like, I can give it to ai. So the AI would be doing the process modeling for me. Some people they would say, no, you still, we still need to do it.

And then we bring AI to see if it can find any gap where we can optimize this. Um. Which school you are more, [00:18:00] uh, close to. Give it to the ai, let it figure it out from scratch or No, we need, we need still have the human led process design, uh, and then bring AI to optimize it. How you seeing things? 

Liam: Yeah, so I've seen kind of both sides of a coin practically the hardest bit with any process modeling exercise.

Is you end up, you want what's called a process taxonomy. It's a view, it's a list of all the processes in your business, and it's at a consistent level of decomposition and you wanna understand scope of each of them. The start and end event, if you've got your process taxonomy and RA frameworks like a p qc that you can use.

Uh, SAP has their own best practices. I have a mix of a two that I tend to roll out with clients. Um. Once you've got that taxonomy, you've gotta view what the process is, what the star end event is. Then you can ask AI charge, GPT, some tools even have it built in [00:19:00] to create that first pass draft of the process.

Some really cool tools right now, uh, Signo and Celonis, they get into a space where you can take. SOPs, word documents, emails, text chains, whatever it is, dump that into them and it'll create a first pass of a process for you. You absolutely gotta use those tools to get something on the board. You don't really wanna be going into an interview with a completely blank sheet of paper.

And then you've gotta be refine it with the people to make sure it is accurate. It is solid, 

Mehmet: uh, insightful, I would say, you know, like, and I think it's a. You know, good balance between, um, both, uh, you know, approaches. Uh, I would say, now let's talk about change, right? So change, uh, a lot of people, I can bet like 99.99% of people they don't like change.

[00:20:00] And, you know, we know like many C levels, whether they are CTOs, CEOs, uh, and others, uh, they might have. Underperforming teams, right? And they know, you know what they need to change, but internal momentum, you know, stalls and they get stuck. So why does this happen for in the first place so often? 

Liam: Yeah. Uh, with, so I, I had this a few years ago where I've done a really interesting analysis program with a, uh, it was an infrastructure company.

I built, uh, fiber optic, uh, and, and, uh, utility networks in the uk. We've done this amazing data analysis on how infrastructure was built, come up with some really interesting, uh, suggested programs off the back of it, and then it just hit a brick wall. Because we have a process team, the analysis team, but the improvement definition improvement follow free was done by a different team.

Definition was continuous improvement. Follow free was [00:21:00] PMO. So we hit this brick wall and we, what we'd done is we've worked in our silo. The thing I tell every function not to do, and we'd innovated, done some great things, but then when it came to the continuous improvement team, they needed to follow their own process.

They needed to do their own analysis, come with their own improvements, and do that in their way. And it undid all good work we'd done. They had to retread the same ground, talk to the same people about the same things, and a lot of the really interesting data insights really on earth. Weren't ever really carried through into execution.

So the same way across P two P, the pain is when you hand over between functions in transformation, the pain is when you're handing over between those changemaking teams. It comes back to what I was saying before. If you've got this changemaker function, you've got your PMO, you change management, BPM, data intelligence, enterprise architecture, and 1,000,001 other ones working together in blended teams.

When you can start right from the inception of projects and carry that all the way through to delivering. [00:22:00] 

Mehmet: So I think that it take, it takes some, some time to build, you know, this kind of balance. And, uh, between performance, accepting the change and accepting that someone is coming from outside. How much communication is also important, Liam, in all, in all this.

Liam: Oh, definitely, definitely. The, so the people element is massive and I think that's often overlooked, especially when you get into more data and tech centric transformation programs you have, I. Right at the outset, you should have a lot of engagement with the team, with stakeholders to understand landscape and real business problems.

And sometimes that's looked over, especially in a lot of the, uh, AI programs I'm seeing now is edicts from leadership, but then not sufficient engagement to really scope that business challenge at the front end. Um. So if you've got a change maker team, you need that people management capability in the [00:23:00] team.

You need advocacy in the business as well. Clear people who are gonna own the change in the area. Clear accountability built into the business and for some really interesting change management tools on the marketplace. Now, digital adoption platforms are something that's really exciting. Tools like, uh, WalkMe, which is.

Trying to use tech to solve a people problem, but doing quite a good job of it. 

Mehmet: Yeah. Um, another one, which might look like similar but. I would differentiate it, which is, you know, trying to bring something new, like innovation. Right. Uh, and, you know, while preparing and checking your, you know, your, uh, your profile, and I've seen you talk a lot about the tension between innovation and experience, uh, and you call it like tech versus old hats.

Um, from your point of view, what's the right balance?

Liam: I [00:24:00] think it's, it's nuanced, isn't it? It always does depend. I always remember I was doing some work with a, an upstream paint company and I was in the how we do sales and, uh, organizational partnerships and we were talking about the sales process. They were playing about getting a new CRM and a new ERP and I'll speaking to this, um, old Danish guy.

Uh, he was near in retirement, really experienced one of the best salesmen, and I was trying to understand his sales process, trying to understand why what he did was so effective. And he was telling me, you know, when he gets to quotes, you know, he is already built the relationship, he knows the person, and then he is just writing his quote down on napkin.

I said, oh, okay. I've not got that in my list of systems. Is is that a new one? Is that one you use? And just pulled the pack of tissues out of his pocket, pulled one out and said, I write it on the back of one of these and handed it over. And he was signing eight, nine figure deals. [00:25:00] Over a drink, over a coffee, on a scrap of paper.

So that's something that's never gonna be automated. You know, that, that traditional industry, the sales cycle, that relationship based piece, you can't solve that with tech. You can't solve that with new tools. You have to rely on experience. But similarly, a lot of the, uh, even the same program, a lot of the CRM piece, a lot of the ERP piece, the transactional, the processes for a routine.

There's no need to carry on doing that in the legacy old way of working, where it's a routine process where it's not strategic. Adopt best practice. Take what SAP, what Salesforce, what viewers are doing. Use them to use the platforms. Use the prebuilt automations. Use the easy best practice approaches. So if you don't build this unnecessary tech debt and you do stay ahead of a curve, you do keep on using the latest and greatest features of these big organizations continually roll out.

Mehmet: Is there a boundary that you can define Liam, when, for example, you say, okay, [00:26:00] this much innovation is okay, but if we. You know, where is this? That's a limit. Like, okay, like, uh, things gonna break down. I don't know. Maybe people will start to leave the company. Maybe things would start to go, uh, south as we call, like go wrong.

What, what's, what's, like, what, what's the, what are the boundaries I would call between innovation and keeping things on, on track? 

Liam: So, a really good tool for this, um, what we do with a lot of our clients is we use, uh, we build that capability model. Uh, it's tools like Lean X, it's an enterprise architecture tool, but it's fantastic for this.

You build out a capability model and you get your capabilities to about level three most places. You don't need to reinvent the wheel, just use a template of best practice one, and then what you want to be doing is heat mapping it. It used to be 80 20. Or it's 80% routine, 20% strategic. What [00:27:00] we've been doing more recently, and the strategic is obviously where you invest, it's where you make the, you know, the big moves, the big money.

It's where you try and differentiate yourself from the market and the routines where you just need to be good enough. There's been a bit of a shift recently. It's gone from 80 20 to 75, 25, 70 5% routine. Just adopt industry. Best practice. You don't need to reinvent the wheel. 20 innovate. It's what sets you apart in the market.

Be amazing, be spectacular. Be unique. 5% innovation. It's not what you're competing on today. It's what you're gonna be competing on tomorrow. For example, I work with Phillips domestic appliances, blenders and earth fryers and, and all that kind of, uh, goods manufacturing. And this structure was obviously the design, the r and d, the build all super quality.

That was what they compete on. But tomorrow we're considering about moving to that product as a service model. How do you not just make [00:28:00] the, uh, initial point of sale, your only point of sale? How do you keep it. Recurring revenue, provide the services, stay valuable, stay engaged long term. So that moved to the product as a service structure.

That's where they needed to innovate. That's where they really needed to push the boat out and do something that wasn't for today. It was for tomorrow. 

Mehmet: Cool. Now I want to shift gears and bring, you know, um, the discussion to what the, this podcast name is all about CTOs, right? So, um. From, from your experience, like for CTOs specifically, uh, and maybe like even technology department leaders like that can be in, in both contexts.

Engineering department, CTOs, or like maybe a CIO? Probably, I believe like there could be some common ground here and about spotting signals, uh, that indicate that. Their [00:29:00] process or transformation is on teams are underperforming, like any early signs and like what kind of actions they can take to, to, to respond, uh, uh, for, for these signals that they spot.

Liam: Definitely, definitely. Um, a lot of the work I'm doing now is with companies who are, um, moving ERP system, uh, S-A-P-E-C is being discontinued. A lot of those companies are moving on to S four hna, the 2027 cutoff date, so it's very aggressive, but as part of that move, as part of buying S four hna, a lot of companies get RISE bundle, and that comes with some synovial licenses.

Whether you move in SAP or moving Oracle or moving anything, you'll have this capability. Visibility to model processes, understand what you're doing. Be a transformation team, but it's not there to just, um, sit in on sessions with the si, but instead to properly engage with business and [00:30:00] try and shape what that transformation looks like.

And if you do have a BPM team, if you do have a transformation team. Who aren't proactively trying to shape that discussion shape the way of working tomorrow, slim down unnecessary requirements. And then when it comes to rollout, get really engaged in supporting that rollout, then what you're missing, a big opportunity to accelerate that ERP, that system change program cut out unnecessary overruns, cut out unnecessary customizations.

So you should really see that transformation team using the tools they have available, actively engaged in those big transformation programs. At the start and as it follows through, and then it shouldn't die at the end. It needs to carry on scaling to the rest of the business. It's not just about ERP systems, it's about making the entire business effective, functional, and using the best possible tech.

Mehmet: And I think, uh, Liam, it's, it's a continuous thing. It's not like just a one time, uh, task [00:31:00] or project. Like it's something that they should be always, you know, continuously trying to, to, to optimize and, and organize. Uh, uh, I believe now if we want to, if, if like. Let's say as a CTO if I come to you, okay?

Right. Uh, and I ask you how to get more value from process, uh, where you would start to tell me to look at starting tomorrow morning. 

Liam: I would say four things. Firstly, what are the problems your business needs to solve? Is it process improvement? Is it fixing broken invoicing? Is it, uh, improving sales revenue?

What are the problems? Start with the problems. One problems. See if process can help. If it is that, say your software product is being pushed into a market and it's got the [00:32:00] wrong ICP process isn't the answer that strategy, it's product development. But if it is that you're not, you're struggling to fulfill orders If it is that you are slow to follow upon quotes on sales, whatever that may be.

There is a process element about that. It's about bringing people, systems, data all together to get something done. So start with the problem that you're gonna solve. Get really clear on that. Once you're clear on that, you need to make sure you've got the right content in place, right models, right dashboards, the right information to be able to support solving that problem.

And then the next two pieces are probably the most difficult ones. One is governance. Getting the right people to put their hands up and take accountability for change. Getting process owners in place so that when you are not only defining a process and signing off a dashboard, but when you're trying to affect change into it, you have a single point of call, you have someone who can approve those changes.

You have that delegation of authority from the head of function into the process owner level. So you [00:33:00] don't have everything going through 18,000 reviews. Instead, it's a process on, it, signs it off, you can move quickly. And then finally it's about building that culture. So people are excited to engage in processing change and transformation instead of people seeing it as a big, scary thing.

It's a busy day for you if you're doing all that tomorrow. 

Mehmet: Absolutely. Absolutely. Um, if. You and me, we, you know, we want to draw kind of, uh, guesses because I'm not saying like, uh, trends, I'm not saying where things are going because, you know, AI is changing things very fast, but, right. If, if you want to imagine how this discipline would.

Uh, would evolve with AI automation, probably like maybe other things would come. How do you think it would look like? I've not asked you in five years. I would say maybe in in the future, because [00:34:00] honestly I stopped asking 10 years, five years. I can't put time on things. But of course, like you must have a vision and understanding of where this practice is going, Liam?

Liam: Definitely, definitely. And indeed. Let's say immediate future, say the next year. 

Mehmet: Yeah, yeah, yeah. 

Liam: The, the big demand I'm seeing is it's really easy to, so with AI I see some similarities to when RPA came, and I'm not saying it's the same level of technology as RPA, it's far more disruptive, significantly more.

Some of the same trends and and challenges come up with RPA. It was a case of when you got it in, there was a few obvious problems to solve and those problems tended to get solved and it didn't scale out too well from the, and with ai, I work with companies like Lego, like s and others, and there's a few obvious [00:35:00] places to bring AI in.

Customer service handling, um, maybe transferring a drawing to card, some really obvious ones, but. A lot of the momentum post those obvious ones, uh, is lost because what you need is a really nice view of where there's real business problems, where there's gonna be a business case of value behind solving those with AI and making sure that you're not just throwing AI into something that's broken.

So for all of our clients, one, the really emergent services now is to look across an end-to-end process. Heat map weather might be opportunities to explore a gen agentic or generative ai. Make sure that the process is ready for that. So any of those things that are broken, the system, the underlying systems are data are put into a good place and then it's lens and use an opportunity to surface where you should be doing it, and then at the backend, using process to help communicate about change to the rest of the business to [00:36:00] ensure that they adopt it effectively.

If we're talking even longer term, then, you know, maybe AI takes my job and who needs consultants? AI will do it for you, 

Mehmet: Liam. You know, uh, I think today is the coincidence days for me because I don't know why, um, in the morning. I was like, uh, you know, in the traffic. Um, and I dunno where this, this, this idea or thought came to my mind.

I said, Hey, like, what happened to these RPA people? Because you just mentioned RPA, right? Which is, uh, uh, robotic process automation. And I said, correct me if I'm wrong, you know? Um, do you think. It was great. It's still great technology. I, I want everyone to understand, like when I mentioned something and I'm not expert, also Liam, you are the expert, but.

Do you think that the way it needed [00:37:00] the interaction with whoever want to use the technology made it a little bit complex versus AI making things much easy because it's a chat bot and everyone knows how to interact with the chat bot. Uh, and honestly, I didn't prepare for this question, but because you mentioned RPA, so your, just your quick opinion on, on this, um, like, was it like the chat.

And, you know, speaking or maybe using straight English language, straightforward English language with the model that might help you in, in, in doing things. 

Liam: Yeah. Yeah. I, so, so I, I got really on VRPA, um, hype Wagon when it first came out. And, um, what the downfall was is it was very fragile. It didn't have a brain.

It was, you know, click here, click here, do this interface, do this button, whatever that is. But it's always follow a pretty strict script. [00:38:00] If you had a really solid rules based engine on your ui, your user interface, your system stayed very static. It could do a good job, it really could. But when things changed, when it tripped up just a little bit.

Things could change very quickly. It would go downhill, very fast project go for data migration. It wasn't really something you want day in, day out, running a core business just because of quite a high risk profile. And I think AI's got a little bit more resilience because, um, with, uh, agent ai, with generative ai, it's not as rigid.

Now you still wanna put really solid rails on it. Obviously if you've got guarded information, don't give that to certain models. Uh, you can't just say, don't read this, it's gonna read it anyway. So there's still things you have to do. There's still precautions you have to take risky decisions. Still need a human in the loop.

But say I enter. Um, I'm gonna be [00:39:00] reductive here, but say you enter a field, let's say you put, it's a delegated po. Great. 

Mehmet: Mm-hmm. 

Liam: Okay. But then someone does delegated purchase order instead of po and in a field, RPA might trip up over that because it's not in the script. Whereas some of these AI agents, they can do a little bit of the passing themselves, a little bit of thinking.

Yeah, 

Mehmet: that's, 

Liam: don't wanna give them terrible data because it can break, can break much the same, but they're more resilient. RP was, that's gonna make reduce the risk profile. I think that's gonna lead to much broader adoption. 

Mehmet: Yeah. Uh, makes sense. I would, I would say, Liam, that, that makes, makes a lot of sense to me.

Yeah. Because the way you describe it and, you know, interacting and the name was robotic, right. So, so, so when you think about robots in the old sense, like it, yeah. You just need to instruct it. You need to teach it what to do. So you actually need to do the work. Right. And then it'll just try to, to, to [00:40:00] do it after you, so.

Uh, LLMs and, uh, like AI is trying to simplify this. So probably this is why, um, uh, it, it, it took like, uh. Better, I would say, um, adoption, um, as we're coming to an end, uh, Liam, like, uh, final words you want to share, maybe final, uh, advice you want to share and of course where people can get in touch and, uh, maybe learn about what you do or maybe if they want to reach out to you for any consultation or any help that they might need.

Liam: Yeah, I think close and thoughts is if you're a CTO, if you're a technology leader, the process tool is a way of being able to get what your message you wanna take out to the business and get the message from a business back to you effectively processes. Superb lens, being able to translate. System into people and people back into system.

So if you have a [00:41:00] process team, even if you're not in your remit, start to reach out, start to build that relationship, bring them into your Changemaking network. And if anyone wants to hear more from me, um, the best place to connect is to follow me on LinkedIn and I post pretty regular content about process, Signo, Celonis, enterprise architecture, and everything else that goes process nodes, absolutely love.

So, uh, please do reach out. 

Mehmet: Great, and I'll make sure that, uh, your LinkedIn profile, uh, is shared in the show notes. And of course, if they are watching this on YouTube, they can find a description. Lim, like really, uh, very interesting topic. Um. It's, it's, uh, eyeopening I would say also. So thank you for sharing your insights here today.

Thank you for sharing your experience with us. And yeah, as you mentioned, if someone is interested, so follow Liam on on LinkedIn. Reach out to him if you have more questions. And this is how I end my episode. This is for the audience. If you just discovered us. [00:42:00] Thank you for passing by. I hope you enjoyed if you did, so give me a favor, subscribe and share it with your friends and colleagues and if you're one of the people who keeps coming again and again, thank you for your loyalty, for your support, for all, what you do for the podcast guys, like really you making me, uh, you know, so proud of, of all what we did together because this cannot happen without you, the audience.

Uh, I was just checking, you know, a couple of minutes ago 'cause I received an email and actually you. From long time we didn't see, you know, the podcast trending in multiple countries at the same time. So as I'm speaking, and this will be aired probably in one week or two weeks from now. So the podcast is trending in five countries at the same time.

So really appreciate the help. Really appreciate that you're listening to us from multiple places in the world. So I really appreciate it. And as I say, always, I promise you always the best and I promise you to be again in a new episode very soon. Thank you. [00:43:00] Bye-bye.