Key takeaways
- Trust is the hard part: people are trusted without being measured, while a robot reports thousands of attempts, so 95% accuracy against a human 77% can still be rejected because it is not right every time.
- Build trust by defining success first, with metrics much like those applied to people, and by starting on low-risk processes that are not regulatory, before stepping up to anything material.
- Pilots fail in production when they ignore the organization’s data flows, people and processes: once users decide it won’t work for them, the crowd is lost. Go live in stages, on a roadmap with quick wins, midterm and long-term goals.
- Use agentic AI to reinvent the business instead of speeding up old processes: start from the board’s risk appetite, look for bottlenecks, keep humans in the loop where judgment and emotion matter, and aim to be second or third, never last.
The fastest change of a career
Mike
Welcome. I’m your host, Mike Parsons, and we are here once again. We’re at the headquarters of FlowX.AI, and we are going to connect some dots. We’re going to figure out what’s happening now and what’s next in this big, fast-moving world of AI. But in particular, we’re going to dive into mission critical applications in the enterprise. This is where scale and complexity start to play a massive piece in this puzzle, and we’re going to try and figure it out. And as always, I’m going to be talking with experts who are really working with the subject matter at hand. Today, I’m joined by Brent from Sutherland Group, and I’m joined by Ian, who is an advisor to FlowX.AI. So, I have two wise men who hopefully make me look all right. And we’re going to try and get to the bottom of how you get that agentic AI advantage, how we get the speed, how we get the quality of AI working at scale. Ian, Brent, welcome.
Ian
Thank you very much.
Brent
Happy to be here.
Mike
So it feels like what a moment in our lifetimes with AI. It feels like just when you think you can relax, no more big announcements this week. Boom, something big, something game-changing seems to happen. Brent, I would love to ask you. You spend so much time working with large enterprises, putting agentic AI agent-based solutions to work. How does this feel? How does this moment feel in your career? Does this feel like a big wave like mobile was, like cloud was? How does it feel for you?
Brent
It’s exhilarating. It’s the fastest moving and the most change that I’ve seen in my career. And it’s exciting to be at a point where I have had the experience I’ve had to be able to bring and apply my you know, experience to this change and help our clients really adapt to it in a way that I don’t know that I could if I hadn’t had the background that I’ve had to date.
Mike
And when you compare this to some of those previous waves, what’s different about this AI moment that we’re having?
Brent
And certainly the speed. Yeah. You know, it felt like. Internet and you know, crypto, and then any of these others in the past, like they were, you know, they kind of built and they but they were fairly stable, and there was a process to sort of then take that into the enterprise. But there’s no break to to adapt, really, it changes. Every week, as you said, something new is coming. And usually there is this lag between a new technology coming and it being adopted and really value being created from it within an enterprise. And so you’d have some lag, some time to adjust and to incorporate it, but it just keeps coming wave after wave after wave. And so I think it’s leaving. Enterprises in a really trepidatious position where, like, when do we get the next wave? When do we get on board and start to ride that wave? Because just waves keep crashing again and again, right? And so that’s part of what our job at Sutherland is to do: to help advise and guide and put our customers on the right wave at the right time to be able to get some value out of that ride.
Mike
Well, I’m glad you’re here because maybe you can help me figure it out. Because Ian, this week alone we saw major new models from all the frontier model makers. I’ve never seen something like that in my career. And I started an internet radio station in 1997 and I look here in 2026 and I have all the waves in between. I’ve just never seen a moment where there’s so much innovation, so much progress. And I think I would love your thoughts on what Brent pointed out, which is the speed of it. How do you reflect on it? And also, how we it seems like we’re all trying to figure it out, but it’s going so fast. Also, I’m keen to know how do you figure it out?
Ian
Yeah. Well, we must be a similar age because I started my career in 1998. And so it’s been 35, 36 years. And what I say to my people is: through your career, you’re going to have to reinvent yourselves many times. New things come along, new topics. Don’t be afraid, but dive in. And as you said, I’m an advisor now. I was at Deloitte until May this year. And I thought that would be my last. Activity. But of course, now I’m reinventing myself as an AI agentic person. So that’s what’s happening to me as an individual. I’m having to get up to speed with it. But when I think back to 1999, the year 2000, I was working in a branch which sat in a cluster of about 30 branches, and I ran the only computer. In that whole organization. And then I became responsible for installing networks where we instead of walking upstairs to see the boss with your three and a half inch floppy, you could actually send it an email. And I think. We’re at a similar point now where the the pace of change then and the technology that was i being introduced had such a fundamental impact on the way that we worked, you started losing sec secretary pools. You were no longer dictating letters to customers, you were typing them yourself very badly. And so the way that we worked changed very quickly. And I think that’s going to be one of the biggest impacts that we see now.
Can we trust it?
Mike
So this speed is forcing a rate of reinvention, of change, this acceleration is really it seems so apparent in this moment, in this week, we see all of these frontier models releasing brand new models, and it’s like all the benchmarks are updating at top of the benchmark today, top of the benchmark tomorrow. It’s it’s crazy. Now for all of that technology speed and reinvention, the next topic I really want to get your insights on is this idea of trust. And I think the funny thing is Brent, that for all of this technology, a very human question of can we trust this is coming up. What do you see in the enterprise when folks are trying to get their heads around how to bring agents into a workflow? It’s not just answers to a question now, it’s coming into the workflow, it’s maybe making decisions. But there’s a very human moment of can we trust this? How do we trust this? What do you see?
Brent
I absolutely see that. A lot of concern, some fear, and a sense that AI is a black box to most people and they don’t understand how it gets to the answer that it’s gotten to, whether it will get to that same answer again and again in a repeated, predictable way. And then, you know, how does that impact me as a person, as a human working in this in my job, right? And how will I play with that? How will it be an extension of me or how will it replace me? Or, you know, so. There’s a lot of concern of like, what is my role with it? How do I interact with it? How do I reinvent myself to leverage it versus will it replace me ultimately?
Mike
So, what’s interesting is you touched upon this topic that’s come up in a lot of the conversations we’ve had the last few days. And the parallel I seem to be seeing is that. When an individual is just asking a question of AI, if it’s correct 90% of the time, yes, that’s okay.
Ian
But as soon as it becomes a robot, it has to be right all the time. Because as humans, we ascribe a different level of trust to an individual because you don’t see them failing that often. You don’t have such a sample size. They’re always doing it right. You don’t see the times when it’s done wrong. But when you are reporting how much a robot system succeeds or fails, you’ve got thousands of attempts. And so you can say, yeah, it’s ninety-five percent accurate and your humans are only seventy seven percent accurate. And they’ll go, yeah, but I’m sorry, I’m not buying it because it’s not right.
Mike
Now, this has to be one of the biggest topics right now because what I’m definitely seeing as a pattern is that with banking, insurance, and logistics, you know, AI is getting out of just giving answers, it’s getting in the action. It’s coming in and it’s revealing this trust question. So, this really brings us to this moment, Brent. Like, what can executives do? To build that trust because what they’re all concerned about is that when you put AI into something enterprise that is at scale, and scale means complexity, so the size of the mistake can be big. So, this is a trust moment. How do we build the trust?
Brent
By being very clear on what success looks like, what are the metrics that define success for how we’re going to apply AI in our workforce? I mean, they’re not unlike the metrics we apply to s for success for the humans. But it’s understanding clearly what those things are and designing and applying our AI with those metrics in mind so that we can clearly measure the outcomes and that we can see how the outcomes relate to our expectations and that we can. Determine how what is the variance from that? And then how do we adjust from that point?
What the enterprises that earn trust do
Mike
Your colleagues told me about an example where, with claims for a large insurer, they were able to get this massive increase in speed on low-level decision-making in the orbit. The compression was 80x. That’s how much quicker things could go. So, what do you see in great enterprises that are starting to figure out AI. What do you see as examples or moments where they are getting the trust with AI and getting the results? What are the characteristics of those enterprises using AI well?
Brent
There’s a curiosity, there is an adaptability and enthusiasm for servicing the customer, ultimately getting it right and doing it in a way that creates value for the company, for the consumers of the company’s products. So if you have that in your mind and in your heart, this is just another tool to help you get there and get there in a way that is more effective and creates more value for them.
Mike
So this is interesting, Ian, because what this really starts to touch upon is some themes where companies need to really think about full systems. It’s not just the model, right? It’s about the data that goes in. It’s about the policies that are processed. What do you see from your vantage point when an organization or enterprise is embodying that curiosity, that openness, that desire to get it right, what sort of things are you seeing that those that are getting the trust and the value out of AI, what are they doing?
Ian
Okay, I just want to go back one second and say that I think a lot of organizations have a fear of missing out. And so they want to step into this world of agentic or AI. And the most sensible thing that you can do, in order to create trust, you have to have repeatability and familiarity. It has to be around, and you need to take the fear away by rolling your sleeves up and getting stuff done. And it’s very sensible to use proof of concept on. Processes that are low risk to the organization because you don’t want your brand to suffer if something goes wrong. So, what I’ve seen is that organizations use the technology to start with on things that are certainly not regulatory based, and their decisions that, if they go wrong, are not going to affect the reputation or the P&L or balance sheet hugely. And once you’ve got a few people in the organization comfortable with that, then maybe you can step it up to something that’s a bit more material. Now, the other thing that I see in large organizations is that at board level, they’re not too keen on people with pointy hats. So for many years, you had a bunker of bright people building models and doing tech stuff, and they would just let them get on with it. But more recently, both in the US and the UK, the regulators have said, I’m sorry. You don’t know who’s flying the plane. You’ve got a bunch of graduates down there doing stuff and you have no idea what it is. You now need to take responsibility. The governance responsibility has sort of found its way up to board level, and that has required them to start to understand the technology, understand how do you govern it. And it’s at that point where both those things meet that you start to get the biggest benefits from the use of the technology.
From pilot to production, without losing the crowd
Mike
Well, let’s dig into some of those because you mentioned proof of concept. The famous pilot that looked so shiny, then it goes into production and things get a bit wonky. I’m very curious. When a company has a good pilot, what advice do you give to them? Like, they now want to take this into production. What’s some of the thinking that you challenge them with to get it right?
Ian
Gosh, that’s a tough question because it’s so specific to the situation. A successful pilot. Is normally a go decision of the go-no-go. And of course, depending on how well the pilot was structured, you might be able to translate straight into a go-live environment to a larger, bigger process. But where it breaks down is that if the proof of concept didn’t take into account the organization’s data flows or the people or the existing processes, and it was only a sort of microcosm. Then, as soon as you try and go live, all of the users say, well, this isn’t going to work for me. And as soon as you’ve done that, you’ve lost the crowd. Yes. And you might as well go home, even if the model’s fantastic. Because losing the the clients at the large-scale implementation is the worst thing that can happen to any technology system, even if it’s brilliant. And I’ve seen it over and over again: if people don’t use it or it has a bad reputation, you can’t come back from that.
Mike
Brent, you have obviously had the chance to work on some projects that have gone into production and succeeded. And I think what Ian touched upon is. When people are in the sandbox and they only partially think through the use case and then they put it into production and it’s found wanting. So that’s obviously something that can be very challenging, particularly at this moment. Everyone’s trying to figure out AI, what are you seeing in those that get pilot to production right?
Brent
It’s a great question. And I think, you know, going live doesn’t necessarily mean big bang, all of a sudden I’m live everywhere with everything, every product. You can scale, and I think you want to, right? You want to start small, start in lower risk areas and start to, you know. Observe and manage and scale as appropriate, right? So you don’t need to launch for every customer, for every product. You want to launch in a place that you have some control and some consistency and visibility and then scale from there as you’ve had success or scale back, right? I mean, it shouldn’t be a binary thing.
Mike
If an executive is feeling like they’re falling behind a little bit, what advice would you give for them? Would you say do lots of small, low-risk pilot proof of concepts? What practical advice would you give them if they’re feeling like we’re falling behind, which is something we’ve been talking about? It feels like everyone is going at 100 miles an hour. What advice would you give them so that they can get these experiments, learnings, get on the tools and start mastering it?
Brent
We like to work with our customers to lay out a road map, right? And any good roadmap has short, mid, and long-term objectives. And so some of the things you just discussed, you know, some of the POCs and quick wins, you want to go after those so that you can have results and learn quickly and adapt and pivot. But you also have to be looking toward the midterm and the long term as well. So you’re laying the groundwork and the foundation for that as you go. So it helps, I think, to have that longer-term roadmap that has interim steps. And has immediate quick wins that you’re really trying to learn from and get value from, and that help you reset as you go.
Ian
Can I dive in there? I think that long-term thing is really important. As I said at the moment, people have fear of missing out, so they want to use the technology and there’s a danger of the tail wagging the dog. So let’s just go ahead and do some AI stuff, please, get on with it. But that typically results in people reengineering the existing processes and systems. And what I see is that that is a mistake. Why not have a look at the business now? What is the long-term vision for this business? Do you need to actually to succeed in your business, do you need to change some of the processes, all of the processes, some of the products? And now let’s use the new technology to quickly move to that area, not just use it to make the existing processes quicker or more efficient.
A brief for the board
Mike
This is a great moment. Let’s do a little bit of brainstorming together. So, what I’ve learned from doing this video series is. Something that you touched on at the beginning, Brent, which is it seems like speed and acceleration is the kind of theme that is really associated with AI. So this is very interesting for anybody at enterprise scale. They can get a lot of things moving faster, which could become sort of an advantage in your marketplace here, particularly if you are at the early adopter stage. If you get onto it early, you might achieve market share advantage, faster times to market with new products and services. So, in you kind of challenged us there. What would be, as an exercise, a good challenging objective? Would it be something like: how can our business be a market leader through being agentic within three years? Is it this kind of thinking? Help me kind of frame this. Imagine we’re going to give an executive a really challenging long-term mission based in this theme of speed and acceleration.
Ian
I think first of all you have to take it back to what is the risk appetite of the organization. But before you step into any of this, can the board actually articulate what it is that they’re willing to do? How far are they willing to go? Once you’ve understood that, then you can move into, okay, guys, where are we going as an organization? What do we want to achieve? And which technologies are going to help us get there? It doesn’t always have to be agentic. Of course, that has a place, but there probably are some other things that we need to look at. And now let’s all put it. Put it all together in one vision rather than just attack it in small pieces. So I think that’s the high point, the high level that you should be looking at at board level before you press the button.
Mike
So, Brent, if you were sitting with a board right now, they’ve brought you in as the technology expert. And they’re like, okay, how do we win with AI? What are some of the challenges you’d put to them?
Brent
I would challenge them to think. I mean, Ian mentioned it a little bit, I would challenge them to think about what are parts of our business that we can reinvent. Like, this is a time and place where we have the opportunity to operate in a different way. And rather than just. Patching AI into parts of our business, as Ian said, make them slightly faster. We have the opportunity to reinvent big portions of how we operate or how we go to market or how we serve our customers by thinking from a long-term perspective on how we can use AI to do that and reinvent how we operate. And there’s so much up. Opportunity to move faster and to do more with that technology.
Mike
And my build here, because we’ve got to build this little agenda for the board, I would say and look at clues might be in manual processes. Highly repetitive, similar pattern processes. I think those two areas, whether you’re in banking insurance or logistics, I think with that sort of brief that we’ve built, I think there’s so much value to unlock.
Brent
Absolutely. And any place that you see. Bottlenecks historically and your ability to service your customer or get products built. Like where there are bottlenecks, there are opportunities ultimately.
Mike
Oh, that’s interesting. Yeah. So where’s the team blocked? Yeah. Let’s go look for those.
Ian
Yeah. So the blockage in my career has always been data. So customers will tell you the data is fantastic, there’s no problem with the data, and they send it to you. And first of all, it’s the wrong data. And then when the right data arrives, it’s a disaster. There are gaps, there’s miskeying. And so, in most of the companies that are going to be using agentic AI valuably, they will be relying upon old data systems. Now, I think part of the beauty of the latest technology is that it can manage poor data quite well. It can deal with it much better than we’ve been able to in the past. So that’s another advantage of it. But data is still going to be a limiting factor.
Data, policies and what people actually do
Mike
Now, let’s jump onto this because this is where I totally agree. I have never seen records on an enterprise that are as good as they thought they were. You know, it’s the classic moment. You finally get, oh, that’s missing, and that’s a duplicate. Now, this is a great governance opportunity. It’s a great design decision because if you build handbooks and ontology for your models, for your AI, where you say, we know it’s not perfect. And I think this is where the big clue is. We know it’s not perfect. So here’s. The scenario planning A, B, C and D. So you have all of those fallbacks, that redundancy. Because once the model knows that it might not always be perfect, if it’s not this, then it’s that. And it goes through those permutations. That sort of thinking, where you accept that the data could be rough around the edges, that could be something. The key thing I would love to your thoughts on is what gets often matched with the data goes into an agentic process is the policies that govern the flow of that data and guide the decisions. What’s been your experience when you introduce new technologies? The data wasn’t so great, but then something like AI often reveals the inadequacy of the policies with all those edge cases that AI is very good at finding. They’re not dealt with in the policies.
Ian
No. So. We have this amazing luxury now of being able to sweep through huge amounts of information and identify the holes or the attachment points or the overlaps. And that that is that is incredibly valuable. Because if I go back before AI, the scenario that that you just described led to organizations that were unable to function properly. You had a lot of dysfunctionality, where even if it was so there’s also this gap between what’s written down and what people do.
Mike
That’s interesting. That’s the culture [unclear].
Ian
Yeah, yeah. So even though we have that policy, that’s not what I do. I do this thing. And organizations typically build up this is what we do around here policy rather than what’s written down. And I think we’re moving to a point now where. The tools will be able to see what’s going on, recommend slightly different courses of action, and improve ways of working. Because it’s now, like going back to what I said earlier, who’s flying the plane? Whoever has said this is what we want to do as an organization now becomes actually how it’s done.
Mike
What an interesting picture that we’re building because if we are thinking about deploying agents in the enterprise, bringing it into decision making, it’s no longer just giving us an answer or a summary over here. What we’re seeing is it’s so much more than the prompt and the model, it’s the harness. It’s the governance. And it’s not just the model, it’s the data on one side, the policies on the other. It’s a much bigger picture. I’m interested, Brent, do you think a lot of enterprises underestimate the effort with putting agents and AI into the kind of beating heart of the business.
Brent
I think they most of them underestimate the effort.
Mike
Why is that? Why is that?
Brent
I think that it’s a lack of awareness and understanding of how this technology fundamentally works, right? Like there’s a certain amount of. You know, and it’s built up just in our culture. There’s something magical about AI and we don’t really understand how it works. And it, I mean, it is math at the end of the day, and it is predictable and can be observed, and there are elements that we do. Fundamentally not understand, I suppose, under the hood. But as it comes to applications in the enterprise, we do have control of what policies we give it and what data we give to it. And I just don’t think people executives understand fully the level of control that we can have the you know with the harness, with the ontology, the level of our ability to define the the criteria and the parameters under with with which it works. We have control over that it with a modern day harness to to to guide how the agentic world operates within our business.
Sovereign AI and the data advantage
Mike
Okay, this gets us to a very hot topic: sovereign AI, where enterprises now can have their data and policies sitting nicely in their walled garden, locked under lock and key. What has definitely developed over recent times is you don’t have to only deal with two frontier labs. There are plenty of other local and open models that you can design and customize, train specifically on your data and bring into the enterprise under this. Umbrella concept of sovereign AI. How big is the opportunity with sovereign AI in your mind, Brent?
Brent
It’s huge. And I think to your what you were just saying, like you don’t want to depend on the frontier models exclusively. And they’re not designed to do the best with what you need them to do. They’re designed to do well with a lot of different things. And in an enterprise, you have very specific problems that you want to solve. And so you want to apply AI solutions to that that are tuned to solve the problems that you need them to solve. And those are relatively predictable problems. The frontier models are designed to be able to be good and great at a wide variety of things, but I don’t need that. For most of the problems that I’m solving.
Mike
Well, it’s interesting. You bring up basically general intelligence has to know my holiday recommendations, my recipes, and how to look at the P&L. Well, that’s a couple of big verticals, big use cases.
Brent
Overkill for what I might need, and also overcost, right?
Mike
Like, well, there’s the cost, and then there’s the next thing: well, do I? Am I comfortable sharing my data with them? And you shouldn’t be. Yeah, because they’re training on that data. So, this is really interesting. So, what this notion towards sovereign AI is bringing AI under control, which I think really pairs nicely with governance, these themes of thinking through your policies, thinking through your data in order to take control of it. What I wonder. Ian, is when you look at this idea of bringing AI in house, really this idea of sovereign AI, when you look at that, what is the opportunity? And what are some of the outcomes that an enterprise might enjoy if it decides, you know what? We’re bringing this in, we’re going to take control of this.
Ian
Let me think about opportunity, but whilst I do that, I just want to pick up on what you were saying. When you’re in-house, you don’t have to worry about your IP walking away. You don’t have to worry about your customers’ data potentially going back in or outside the firewall. And so so those are huge benefits of being inside. Advantages, of course, are that you can train your own large language model to understand precisely the activities that you do as an organization. And it’s specific, not a generalist. And at that point, your business becomes the driver. Rather than whatever the data it was that it was trained on. So you have potential to go deeper, become faster, have it be more reactive specifically to you as an organization and the complexities that your industry or your specific business has.
Mike
So with that customization, I want to return to our boardroom brainstorm format. We’re going to help executives pitch this move towards agents and agentic ways of working. So it seems to me, if I connect a few dots, one thing we touched upon is they want to be in control of things like their customer data and they don’t want it going off-prem. Okay. And customer data is one of the, and material data, this is some of the key intellectual property that they own and control from which they build their marketplace advantage, customer advantage, unit economics, and so forth. Yeah. So if data is often at the heart, might not be the only, but it is one of the key drivers of growth of profit. Of margin within these large enterprises, then if I do this logic and we were presenting the big picture value of AI, I would say if data is the holy grail at the heart, the beating heart of the proprietary advantage, agentic thinking, AI thinking, large language models, and those harnesses around it are purpose built to take advantage, to process said data. So are we seeing that the winners across category and do you think it could be those that adopted agentic, like those that adopted perhaps cloud or that perhaps adopted mobile, particularly towards customer experience. Are we seeing a move like this? The stakes are: if we become agentic first, we will be first in the market as a result because we get to utilize our data beyond that of our competitors.
Ian
Yes, I do. Actually, I saw an interesting article on LinkedIn just this week about an organization, a big bank, that was saying, we don’t want to be first, we want to be second. First is too dangerous. Second is the sweet spot. And I agreed with that. There’s one thing that we haven’t mentioned here, which is the organization is changing. They’re looking at the data governance is changing. What about the customer? So that the world of the customer is also changing. Their expectations are changing. Their lives are changing. And so if you’re in-house and you’re seeing the change in the customer’s behavior in your organization, now you can respond to that. And it goes back again to your long-term planning. There’s no point in long-term planning against the current customer base and its expectations. You have to have some sort of prediction of where that’s going as well.
Humans in the loop, and checking the return
Mike
Now, what I see is a win here. We’re talking about customer expectations. And hey, they’re all the customers are changing, their expectations are changing. It’s the classic thing. If everybody is using Amazon to buy things, when they go to other retailers, they expect buying things to be as easy as Amazon. The de facto standard gets changed. So levels of customer expectations change. This, Brent, sounds to me like the perfect opportunity to think about when we can put humans in the loop, particularly in those moments of judgment or a high-emotion moment like a claims [unclear] in insurance. This seems like. That as we’re reinventing the business, we can redesign the customer experience and find new ways for humans to be in the loop, maybe to have more quality time with their customers because they have all of those savings and time and speed advantages from AI. What do you see here with this new definition of humans being in the loop?
Brent
Well, a couple of things. I would think we think you talk about trust and you talk about also differentiators. The more identified and AI driven the rest of the world becomes, the human touch and the human in the loop become a differentiator, frankly, in your customer experience. Because while yes, the standard gets raised in certain cases, like the Amazon example you gave, they also get diluted in terms of the level of connection and the level of personalization. Where that’s where a human can make a difference, right? Can make a difference in that customer experience beyond just sort of the generic, you know, AI-tuned interactions that they have. So, you know, where there is that opportunity or where it is critical, because you also talked about the, you know, the. Like where to target AI, right? And there are places to not target AI where you don’t want it making critical decisions or where there’s emotion driven, and you do want to insert a human in that. And that can make the difference for your customer and your brand. So those are that human in the loop can. I believe strongly can be a huge differentiator for a customer in their interaction with your company.
Mike
What’s so curious about what started with a technical conversation like AI agentic coming into the picture here is that we’re seeing that it’s not only forcing like a re-evaluation of the tech stack, but it’s the customer experience. It’s in fact, it’s I might be overselling it, but it feels like it’s a redesign of business itself.
Brent
I think that’s accurate. I mean, to the long view and to the point of understanding, like, how do I want to transform my business to operate five years from now when this is a whole different world and my customers are operating and have a whole set of different expectations? And yes, I don’t want to be first either. And I want to be second or third, but I don’t want to be last because I’ll be out of business if I’m not engaged and part of that journey.
Mike
And that’s almost like it’s carrot and stick. Hey, move faster, get the AI advantage, or potentially be out of business. I mean, to be honest, it feels like you’ve got to make a deliberate choice. What are you thinking?
Ian
Yeah. I see my own journey as a customer over the last many years. I like the ability to not have to speak to somebody to start with. If I want to carry out a task with a bank or ordering a product, I’m very happy to sit there and do it by myself. When things go wrong, the switch flips, and I really want to speak to somebody that can just help me sort it out. And I think. Where we are now is that agentic is going to help organizations get through all of the clutter around a process very quickly and present you to a human being. So it will be like when I was working in banks in the early nineties, you went into a branch to speak to somebody who could help you. And then there were too many customers and too few people in the branch. So it went to call centers and you felt that you could never speak to somebody. But I think now we have, because of technology, we have the ability to. Filter out all of the issues and present you to a human being again who instantly knows all the things that have gone on and what the problem is and can solve it for you. So I think we’re coming back round to wanting that special interaction and being able to have it because technology has managed to get through all of the volume of activity that was required in order to get you to that, because the human resources. Is a limited resource, but technology isn’t. Technology can get through whatever volume it needs to, but you only got 10 people in the call center. So somehow filter out all the other rubbish.
Mike
So it’s almost you paint this picture of opportunity that we have to keep in mind that when we’re trying to figure out how to govern this agentic pilot project. The stakes here are more than the pilot itself. What’s really at hand here is: are we going to reimagine our business, reinvent it in order to take advantage of acceleration and speed, which is now on tap? It’s there for all, but only a few will figure it out. Like, do we want to figure out this advantage?
Ian
Agreed. Yeah, absolutely. Which probably goes, you mentioned ROI earlier, which we haven’t touched on. So maybe that’s something to dig into. Initially, the assessment of ROI for any project is what will kill it or make it survive that initial let’s go, let’s not go. But what I’ve seen certainly is that. ROI in itself is not something that is checked post-implementation. What’s checked post-implementation is the reputation, the noise that’s coming back from the people, and the cost of a system is only ever questioned at a renewal stage down the line. And so, if you do have an ROI capability in your tools, which of course FlowX.AI do, then that is immensely valuable when you’re looking at: should we do more of this or should we renew this particular agent?
Mike
Well, I mean, it seems to be so pertinent because if you look at what agents can do, they can grind through that manual paperwork, those manual processes far better than humans, therefore affording more opportunity to deploy your people. To more customer facing moments, quality, trust building, maybe that’s an advantage. What are your thoughts, Brent, when we when we think about ROI? You’re already seeing projects enjoying great ROI on their investments. What’s next in what’s happening here, and what can we expect for those that are going to win in this agentic race?
Brent
Well, first off, I would just, to Ian’s point, I’ve seen the same even before agentic and AI as it relates to any kind of technology investment. You have a business case. You get the enthusi you know, you get the board approval to move forward and you do all the hard work to launch something and then you sort of move on to the next thing and nobody ever remembers to go back and decide, hey, was that actually worth the investment that we made? And did we learn anything from it so that we don’t make the same mistakes? We did we had the same problems previously. And I think you know, what where I’ve seen and found success in that is, as I said earlier, you know. Identifying clearly what are the definitions of success and the metrics that we’re going to measure, and instrumenting whatever that solution is to be able to see the outcomes and measure and report on them, and creating the governance, the financial governance around managing those outcomes. And then as feedback loops into future investments. And so we’re learning from what we did. And also having the nerve to stop if we’re not getting the result. Just the fallacy, the sunk cost fallacy is something that we’ve invested so many millions in this, we should just keep going. Like, no, you probably shouldn’t. Like, this is the time to cut it and move to something else.
Mike
Well, I think that’s a perfect way to bring this conversation to an end because what you’ve really done has brought us back to the fundamental idea of investing and returns for infrastructure that is all about accelerating the enterprise and making sure the numbers add up. So I think we’ve unpacked a lot of ideas. So I want to say thank you to you, Brent, and thank you to you, Ian. That’s a pleasure. And I want to say thank you to you as well, all our viewers. I hope you’ve enjoyed this great discussion with two seasoned experts where what we did together is we unpacked that speed is on the table for those that truly want to win, those that want to reinvent and reimagine their business. This is far more than taking AI from answers to action. This is about taking action across all dimensions of your business so that you can lead your industry, you can lead your market, and most importantly, you can delight your customers so all the numbers add up. And the end of the day. Okay, that’s it for now. We’ll see you next time.