Agentic AI Is Not a Technology Problem. It’s a Leadership Problem with Michael Fauscette

AI hype is loud, but what’s happening inside real enterprises is much more interesting and much more practical. We talk with Michael Fauscette, founder and CEO of Ariane Research and author of Building the Digital Workforce, to get specific about what “agentic AI” actually means when software can plan, act, and operate as a digital worker. The big shift is simple: generative AI creates content, while agentic AI executes work. That raises new questions about control, trust, and how leaders should roll it out without breaking processes or confidence. 

We dig into the most visible use case today: customer service. When an AI agent handles tier-one triage, solves routine issues fast, and hands off cleanly to a human for edge cases, most people are surprisingly okay with it. The key is guardrails and transparency, plus a rollout plan that earns trust over time rather than assuming instant autonomy. We also unpack why enterprise agentic AI is not the same as a public chatbot. Enterprise systems depend on your private data, retrieval augmented generation (RAG), rules that prevent outdated policy answers, and governance that ensures an agent cannot publish or act on information it shouldn’t. 

Then we go deeper into the context versus memory distinction. Agents can retrieve the right information, but without durable memory, they cannot truly learn from past actions. Michael explains what a high-performing digital workforce looks like as a stack: integrated data, orchestration and governance, specialized agents built for real jobs, and collaboration that enables humans and agents to work together. If you’re a leader wondering why pilots stall, where to start, and how to explain ROI to the board, this conversation gives you a grounded path forward. 

If this helped, subscribe, share it with a teammate, and leave a quick review so more leaders can find the show. What’s the first business process you’d trust an agent to run?

Thanks for tuning in to this episode of Follow The Brand! We hope you enjoyed learning about the latest trends and strategies in Personal Branding, Business and Career Development, Financial Empowerment, Technology Innovation, and Executive Presence. To keep up with the latest insights and updates, visit 5starbdm.com
.

And don’t miss Grant McGaugh’s new book, First Light — a powerful guide to igniting your purpose and building a BRAVE brand that stands out in a changing world. - https://5starbdm.com/brave-masterclass/

See you next time on Follow The Brand!

Transcript

Welcome And Guest Setup

SPEAKER_01 0:02

Welcome back to the Funnel Brand Podcast, where we're talking with leaders who are shaping what's next before it becomes the obvious. And today's conversation is one I've been generally looking forward to because we're going to cut through the noise around AI and get very real about what's actually happening inside enterprises right now. Now, my guest is Michael. I'll let him introduce himself. He is a top 10 agentic AI thought leader, an author of Building the Digital Workforce, and founder and CEO of Ariane Research. And I want to highly recommend his podcast as well. And I'll let him explain exactly what that is all about. Now, Michael, he has spent over three decades inside the engine room of enterprise technology, advising boards, leading analyst teams, scaling startups, and helping organizations understand not just what technology can do, but how it actually changes the way work gets done. So if you're hearing terms like agentic AI, digital workforce, autonomous systems, and wondering what is real and what's premature and what leaders should be doing right now, this episode is for you. So, Michael, welcome to the Funbrand Podcast. And I'd like for you to go ahead and introduce yourself.

Michael’s Career And Research Focus

SPEAKER_00 1:25

Sure. Well, thanks, Grant. That that made me sound really good. I'm kind of impressed with myself at this point. Um, no, thanks. I um yeah, so I I've been working on the my current um current endeavor, I guess, for about four and a half years, uh, area in research. But I um before that, I was uh the chief research officer at a startup called G2, which you may have heard of. It's uh basically a software marketplace and research, uh, crowdsource research. And um before that at IDC, uh for 10 years, I ran the enterprise application group there, uh, research around all the enterprise apps categories. And then I'm a I'm an industry guy. I was at uh PeopleSoft at uh Autodesk, uh did a bunch of startups uh across uh everything from XML to uh to some uh healthcare type uh you know process systems. So pretty pretty broad background. I'm an ex-naval officer, so I uh I got into tech there. I was a uh I was a liberal arts uh musician and photographer that went in the Navy and they went, you'd be a great engineer. And I went, I don't know what you mean, but okay, I guess. And uh and ended up here. And um, and so the the podcast, well, I I guess the Arion uh research, we really I originally focused a lot on customer experience, but of course the world shifted, and I had always had an interest in AI. So we started when generative AI came out, we started really looking at the way that meshed together between enterprise apps and and and AI, generative and then into agentic. Um and the podcast, Disambiguation, I started uh we just started season four actually this week. So uh so it's been around for for a few years, and and we really try to dig into um agentic AI and AI topics and in and look at um you know actual use cases and real-world stories from people that are either building it or using it. Um, and so you know, interesting conversations. It's uh pretty much the same format that you use, just uh a nice casual conversation about whatever is uh is hot at the moment and interesting.

From Generating Content To Taking Action

SPEAKER_01 3:32

Well, this is very hot, very interesting. I mean, you just you know just recap the 35-year history of what you've been doing. And I know from your vantage point about a jetic AI, here's the here's the deal right now. Like my my first question of the day is this hype or is this real transformation? Can we take an autonomous uh I would call robot for the most part? You know, you talk about robot, uh, um autonomous systems and and things of that nature, and then let it loosen the wild into your business, and you and you're expecting a certain outcome. Is that hype on your world or is this something that really could take shape?

unknown 4:17

Sure.

SPEAKER_00 4:17

I I'll give you the typical and and very well tried and true analyst answer, yes, but so so that I what I mean really is that you know, we saw this evolution from generative. So a thing that is intelligent and creates content. However, with not a lot of guardrails, and so you know, you started to hear things like hallucinations and gee, it made stuff up and blah, blah, whatever. Um, and that that engine's excellent, though, very powerful, but really generated, it generates stuff. I mean, that's what it's for, right? It can generate a video, it can generate a report, a blog post, whatever. And with the correct boundaries, prompt, context, all those kind of caveats, it does a really good job. But what we did over the last three years is really move this from it creates things to it does things. And when it does things, that's a completely different kind of thought process, right? So we have um, you know, business processes we've had for years. We've ended up in a situation where we have a lot of different enterprise applications that do a lot of different things. Frankly, they're not a lot, always not necessarily connected. And so you had breaks in processes. Uh, you you had a lot of human intervention needed. Frankly, a lot of businesses use uh people as the way to integrate those systems instead of systems. And, you know, a lot of a lot of challenging processes, and and we've, you know, we've tried automation, uh, things like robotic process automation, RPA, which is very deterministic. So it you tell it to do these four steps and it'll do those four steps forever until you tell it to do something else. That's good. But unfortunately, in the in the business world, and when you talk about business processes, if you want to start to look at them holistically across the entire organization and across all your data and across all your systems, you can't just create logic that executes every step. That's why humans are involved. If you could create logic that did your whole business, you wouldn't need humans, right? You wouldn't need anybody to think, you'd just do. And so now we've moved to the world where we have these machines, robots. That's a, I think that's an interesting way to think of them. I I think of them as digital workers, but it's it's the same kind of thing, right? It's an it's not embodied, although it can be, and we're starting to see more of those, uh, that has the capability to um to look at a situation, understand it, plan, take action, and learn from that action. At least that's the the way we like to think of agentic, right? Right. It can be autonomous. Now, here's the one thing, though, that that I think a lot of a lot of experts even don't necessarily uh think this this way. I like to think of them like this. If you have a um an agentic system, it's one that has the capability to automate a process fully to the point of agency, to the point of the agent determining its own outcome. So probabilistic, not deterministic. But it doesn't have to execute across that entire thing, right? You could say, do these, do this part of the process, figure this out, and then call Joe and ask him if it's okay to keep going. In other words, you could have a human in the loop. You could have a human on the loop, you could have a human that looks at the processes and goes, uh, no, we can't do that and stops them. Or you could even have probabilistic, you could add deterministic, you could have guardrails that say, do these things inside of these constraints. So, so we have the capability now to automate, and that's real. I mean, we you have companies all over the place doing things that are in fact using agentic AI. A lot of times now, if you call customer service for larger brands, you're starting with an agent, and that agent is, you know, it within the boundaries they've built, autonomous. It interacts with the customer. Um, and it most situations now it interacts with a customer to a certain point. If it can answer the problem, it does. If it can, it passes it to a human. Actually, if you survey people, people are pretty happy with that idea. They they don't like the idea of do you want to just talk to an agent without the capability of talking to a human. But if you say, can we do this in a hybrid way so that you talk to the agent, it solves your problem. If it doesn't, it goes to a human and everybody knows what's going on. That works for me. I don't I don't have an issue with that. And and that's what we see in the surveys. So you are seeing them in the real world. Customer service, uh, you're seeing them in in code development. Of course, that's generative all the way to agentic. And you're starting to see them expand into other use cases. A lot of that, though, is horizontal today and broader. And this year, you're starting to see more and more agents that are built around very special types of things, like a vertical focus or a functional and vertical focus. Like this is a a customer service agent in the healthcare industry. This is a customer service agent in retail. So it has it has context around the industry, the activity, what it's doing, the job, et cetera. So that's, I mean, we we are seeing these in the wild. They do work, uh, but of course, there are other things to think about when you start to go through that uh 100%.

Trust And Humans In The Loop

SPEAKER_01 9:47

And what you just said there. And I want people to understand it's around use case. How are you going to use something that can automate and also clearly kind of mimic human behavior, whether it's uh voice, text, video? And it's almost like I think what turns some people off is like, is it okay that you tell people I am an AI automated, you know, robot? I'm a bot. I am not trying to fake out, fake you out and think you think I'm human. I'm just providing a a service in which I can do it faster, quicker, better than my human counterpart. But as I need that human counterpart, I bring them into the loop. I'm thinking like, remember voicemail 30 years ago, you know. People like, oh, all right, I just want to hit zero and then get to it and operate. I don't want to go through all these prompts. But over time, people like, oh, you know what, this is an easier way of doing it and getting to somebody because I it's just what I call tier one support. I it it doesn't take human intelligence in order to direct me to where I want to go. So if you look at that, like you know what, I just need basic information and I need it, you know, pretty quickly, and that I can trust that the information is is factual and and and and actually real, as opposed to something more complex in what it's trying to accomplish. And at that point, it starts like, hey, you know what? And to your point, can I flag that? Say, all right, this is beyond my capability or what I think I now need a human to come into. And these systems have such programming within them, it should be able to discern. Let me get Mike on the phone for you to answer that more clearly. So I like the use case and customer service, and really in that tier one triage world where it can contact you, it can schedule an appointment, it can make you know very good um you know assumptions about what you need, especially like think about a doctor's appointment, things like that, that it just you just need to get to where you want to go. Maybe it's directions that you need right away. Everybody accepts the fact that GPS using you know your your Google or your ways and it's telling you where to go. No one has a problem with that, and it's very human-like, but you know you're talking to a bot. So my question is is there angst right now or people or misconception that you're seeing right now between AI and we call agentic AI and human adoption?

Context Versus Memory In Agents

SPEAKER_00 12:26

Yeah, so I I think if you the the the simple one to think about is this idea of of uh customer service, right? Because it because we've this has evolved greatly over the last 10 years. You know, a few years ago, you dealt with a a bot that was not that intelligent, it was truly deterministic. In other words, you built into it a logic tree. And and you'll even remember when you called in, sometimes it would go, please listen carefully, because our uh, you know, our options have changed. Um, that's because it has a very specific logic tree, right? So it knows if you say this or ask about this thing, I pull up this information and I share it. If you ask this, this is beyond my capability, I escalate this to a human, whatever. That's the old way. If you look at it today, though, if you put an agent in this situation and you build your workflow around it and you put it in a situation where it interacts with your customers, you have the the ability to set the workflow so that it can either answer questions, it can answer questions and do things, it can answer questions and ship things, buy things, you know, it can do a lot, but it is up to the deployment on how much it can do. So maybe when I first roll this out, maybe I don't trust it. And trust is actually really important in the way you use these. I don't necessarily trust it yet to take an action. So maybe I have it answer questions, but then pass it to a human when it has to take action. Well, but maybe over time I test that and I realize it can do this much. So I stretch that a little further. So over time, I build the trust layer in with this, with this agent to do certain things, and I allow it more autonomy as it progresses in its capabilities and we progress in our sophistication and our ability to trust it and understand that it can actually do those things that we want it to do with guardrails, with the right information, with you know, all these other caveats to it. But but you can do that over time. And that applies to almost any business process um with some, you know, with some other uh um capabilities that you need around that. So you'll hear people talk about retrieval augmented generation or RAG, which is you know a technology that sets a database of information, basically. Just think of it like here, here's all the information I have for support. I put it in a container, and the agent can reach out and pull that into the conversation if it needs it. So you uh you know, you want to know, uh, I keep having this issue with my uh computer software. It does this when I do this. How do I solve that? The agent goes, I can solve that. It goes out to the the retrieval, it pulls the information in and it shares that with the customer. So it it solved the problem because it has access, it has context. Context is actually really important. Yeah, uh the downside, the the the part where we're just starting to go to, though, is the second part. So it has context, but it doesn't have memory. Agents are stateless today. So it I like to think of it like this. If you ask me how to make spaghetti carbonara, I'd go to a recipe book, I'd pull the recipe up, and I'd make spaghetti carbonara. An agent can do that. That's context. But if you ask me, what did you have for dinner last night? the agent wouldn't know because it forgets as soon as it does the thing, it moves to the next thing. So that's memory, context and memory. And now we're starting to realize that we can use the agents in places where they retrieve the information, they have the context, and that's really important. And now we're working on systems that can give it the memory, and that lets you take it to more and more advanced states. And you know, I said when I described them, I said it can perceive, it can plan, it can take action, it can learn. Well, without memory, it can't learn because it doesn't know what it did before. But as soon as you build the memory into the system, now it can learn. And the other thing that you'll see this year is you're seeing them broken up into smaller and smaller pieces. So I can build little guardrails in and I can have my agent do these things. It knows it can't get outside of those things, but inside of this thing, it can be fully autonomous. It has context and it now would have memory and it can learn and it can perform in you know, in specific ways inside of that use case, that whether it's vertical or functional or whatever, right?

SPEAKER_01 17:14

You bring up a good point because it appears right now, people utilizing large language models, generative AI, chat GPT, cloud perplexity, Gemini. And it appears to have memory. You would ask it something, and it'd say, like, all right, it seemed like it's pulling data from yesterday, the day before, and it's presenting you with that information. Are you telling me that it really doesn't remember? It's just pulling additional data real time to its, you know, to its to its um application and presenting that information, but it really doesn't remember. I think that's an important point.

SPEAKER_00 17:59

Well, so think of it this way. If I'm talking about publicly available chatbots, they have the capability to have context, and those context windows get bigger and bigger. So, for instance, let's say um I in fact I just did an exercise this morning using my uh using using one of the well, claw, using Anthropics Claude, right? So I had uploaded a copy of an ebook that I published a few weeks ago. It's a readiness assessment for agentic AI. It's you know, it's useful as a bunch of questionnaires and score sheets and that sort of thing. Well, I wanted to build a little executive level one that I could put on my website so you could come to it and you can fill it out yourself and see kind of what's you know, what's my state? How good am I today at this? Do I have all the things I need? So I uploaded the ebook into its context window so it knows the ebook now. And then I gave it a very detailed prompt of what I wanted. I want to use it for this, I want it to have these things, I want it to take about 10 minutes, I want to be targeted towards this level, you know, very long prompt. So I prompted it correctly and I gave it the context to answer that question. It did it accurately, did it quickly, did exactly what I wanted, actually. And I may edit two or three things in there and add a couple things, whatever. But in general, save me a week's worth of work, maybe two weeks' worth of work, to be honest. That's the way those public chatbots work. In the enterprise, it's a different thing. And this is where I think people get a little bit mixed up with this because we talk about Claude, ChatGPT, Gemini, all those things a lot. They're great. I use them all the time, right? They're really good. But what it doesn't do is necessarily solve your enterprise problem. Now, the language model underneath that could be a key component to building an agentic system, but it's not the agentic system. That system is a platform that sits on top of the language model and it has access to your business information. That's that context that I said, that retrieval augmented generation. Basically, I like to think of it like you put a book into the retrieval augmented RAG system. It it shreds the book into pieces into a database, and then it gives the agent the capability to reach into that database and pull those pieces out when it needs it. That's good. Could be bad though, because it only uses you prompt it or it's prompted by the interaction and it reaches in and it gets the thing that fits the prompt as good as it can. So that's where you have these kind of concerns, right? So I ask it a question about a policy, uh HR policy, and I wanted it to use the 2024 policy, but I didn't say that. And it goes into the database, but the one that matches its keywords is the 2022 policy. So it pulls that out, it answers your question with great authority and goes on. That's the danger, right? So it needs the guardrail, it needs that deterministic piece that says use the policy that is the latest date every time. Okay. Now it's an enterprise system because it won't give you inaccurate information because it goes 2024 is later than 2022. When you use the 2024 policy, answers your question. Good. So that's that's the difference. Enterprise has more guardrails, has more control, has more specifics, has more context, has more memory, and we're and we're building the system more and more to include more memory so that it will, in fact, have the capability to go. I did this two days ago and it didn't work. I won't do that again. Yeah. And it goes and does it a different way, and it learns from that that worked. I'm gonna do that the next time. That's what an enterprise system is.

Why Pilots Stall And How To Scale

SPEAKER_01 21:36

This is interesting because what you that's a very good distinction when it comes to um enterprise applications. And now it's not using public, it's public and private, right? So you're talking about private data and understanding what is that data consists of, you know, your data warehouse, whatever you have there is a good data because it it does not know if it's good or bad. Like you said, the It just knows its data, right? It's data, it's going to pull that information and it's going to perform the uh activity that you have prescribed it to do. Uh, so we've got to really make sure that you bring up a really good, good point that I want our audience to lean in on that agentich they are is you do utilizing your data. It's not just going out to the uh internet scraping the information and bringing it back. Um, that's the LLM portion of it, but then agentic is more inside your enterprise, and it's it's getting that information out. This is you know, that's a huge, huge differentiator because I think there are some misunderstanding about that. And you talk about you're you know deploying a digital workforce. How good is the data that you're going to be utilizing that the agentic application is going to uh pull from and actually be available so that you get into these use cases around agentic AI and look at the education, the credibility around it, and then model assessment. Now, you've developed these readiness assessments, and you've got some you know, maturity markers. I think it's really I'm really talking about maturity. And I my question is why do so many organizations stall after these pilots, even with great technology?

SPEAKER_00 23:18

Well, so the part of the there's there's a couple issues there. One is that it's not a technology problem. Like, you know, I back in the 90s, I was working at PeopleSoft doing implementations of large enterprise systems. And guess what the biggest problem was with adopting an old line enterprise system? It was the people, not the technology. The technology did whatever you told it to do. The people, though, did what they knew how to do. I I give you an example. I was working in one implementation in the in the mid-90s and uh was was leading the project, and we went and we talked to this person about changing their business process, and they and they, you know, I said, what do you do today? And they took me through what they did today. And was basically a uh a you know a simple mainframe kind of based system where they filled out these forms and did this, filled in this field, blah, blah, blah, whatever, hit in. And I said, Okay, so we're gonna design it around this. That's what that's all you do, right? And then they went, they looked at me and they went, Well, no, that's not all I do. So I actually also put it over here in this system. So they put it in two systems. Okay, well, that's okay. And then they they and I go, okay, cool, then we're gonna model that in this. And then they go, no, no, no, we also actually put it in this paper ledger. Like, why? Well, because the system fails sometimes, huh? So my point is the change was harder than the system. The system could do everything they wanted, both of those systems could, and the new one we had would do the same thing, but this person didn't trust it. So they did three or four backups to the thing because they didn't trust it. And the same thing's true with the systems today. As you roll these out, it's more of a change management issue than anything. And there's also there's so much hype today around ooh, the agents are coming to take your job. And so there's a fear factor there too, right? So if I'm an employee and you're showing me this agent that can do my job just like I do, the first thing I think is, oh, you're gonna replace me. What am I gonna do? But if you came into this and you said, look, we're building a team, this is a part of your team, this is a digital worker, you're you're you're a worker, we're building a hybrid workforce. Here's how this all functions. Nobody's being losing their job, not the digital agent nor the human agent. We need the both, and here's how that happens, you'd have much better adoption. So that's that's the first thing that fails a lot is the it's just the change management that you need to make it work. Then the the second thing that we have to worry about is we we've sort of approached this in what I like to think of as God mode. We build this big agent, it's gonna do everything. That's not really the reasonable way to do this, right? What I really need in a business is I need an agent that does a thing. It's just like hiring a person. You give them a job description, you have them do those things. I don't hire one person to do marketing, sales, and finance. That'd be silly, right? They don't know how to do that. Same thing's true with the agent. You might try to give it all that information, but it's not ever really going to be good at any of those things because it's just too broad. But if I broke that up, I could, and especially if I broke it up by job function and by industry, I could give that agent really specific information that it could work from. I could also build guardrails into it so that it worked inside of the box I wanted it to be in. It's autumn, you know, it automates and it can have agency inside its box. It can't get outside the box. Now you mentioned data. Data is absolutely key, but there's a second part of data that you have to think about too when you're doing this. And that is I, well, first of all, I got to fix the fact that most of my data is all fragmented and not very good quality. So that's but that's also an agentic problem, and I can have some automation to help me with that and you know, whatever. But the other part of that is I need that data to be inside the, you know, I need it to be the relevant piece inside of that function that I need for it to do. And I need it to have access to that data. So it has to have tools to have access. We're using things like um MCP, which is like an integration standard. Uh, we also want that agent to be able to work with other agents, so now it has to have a collaboration framework to work inside of it as well. So it has to have we A to A is one standard, but there's several. So the agents have to work together. And then I also want it to work with humans, so I have to have a collaboration system that works with the humans to the agents, the agent to the human, the agent to the agent, and human to human, I guess, too, right? So all those things, I need all of those pieces in place for it to function correctly. So I guess that sounds complex. It's not really as complex as it sounds, but it's specific. Just like implementing any system would be specific, it has certain things you need to do. And when we first started into this, we built platforms to build agents and we let the agents, we let the humans in the process control kind of level of trust and how much it could do and that sort of thing. But it didn't really, the those early systems that were agentic didn't have a good strong set of guardrails. And so you saw, like I'll give you good examples. They build a an agent platform they rolled out a couple years ago. It's evolved. Well, at Dreamforce this year, all of a sudden they announce something that is a scripting language that lets you put guardrails around it. And why do I want that? Well, data. So if I don't have specific guardrails, the agent does, it has autonomy, it can do what it wants. Let's say it looks at this and it goes, I want to do some marketing. You know, I bet if I pull Grant's profile up, oh look, Grant told me these four things last week. That'd be a great case study to put out into the public. It builds it, it's beautiful, it puts it out into the public. Except that's private information it didn't have the uh authority to do. I want it to have guardrails so that it can't do that, right? That's a thing. I don't want it to publish your information without permission. Simple. So I need the capability to do both. I need autonomy, and I need the ability to limit autonomy in the right situation.

What A Digital Workforce Looks Like

SPEAKER_01 29:36

I think this is uh I'm glad you're you're touching on these points. I want people to really understand that you you start to move from different ways of thinking from you know applications to workflow and workflow automation or workflow-centric, you know, applications that are really critical and then understanding the difference and how you got to start thinking about these things because you've talked about agentic AI operations or AI options. I think you're talking about now, uh, and how all that, you know, has to look. What does it look like from an autonomy world, the governance uh section, and and bring this in just like you said, an employee and setting it to work, but then setting it into the team. I think there is too much buzz around, you know, it's replacing everybody like no, but I think personally, if you're gonna be replaced by it, a machine, you gotta really look at what is it that you're doing. Um uh, and and that's very, very important. But as we move forward and as you start thinking three to five years from now, here's my question for you. What does a high-performing digital workforce actually look like inside an enterprise?

SPEAKER_00 30:53

Well, I think I if I take this kind of in in building blocks, I think that businesses start to realize and build out and understand they have to automate a high-quality integrated data layer inside their business that captures everything they need to run their business. So all that data consolidated together, not fragmented like it is today. So the CRM data and the financial data, it's all in the same data layer, right? All those things are together. That's the first thing. The next thing is I need inside of that, I need a agentic orchestration and governance layer. So I need a an agentic platform that has governance built into it, and it has access to the data, and it has the capability to build and deploy agents that are based on each specific thing I want it to do. So that that data layer, agentic platform with orchestration and governance. And then atop that, I need those agents. Um, and I need them to have access through the system. So I'm not throwing away my enterprise systems, but I'm taking them back a layer. So the agents start to become my way of interacting with those systems. And then above that, I have a collaboration layer. So I have a way for humans and agents to work together, whether it's agent to agent, agent to human, human to agent, you know, human to human. So, you know, think of it like data, uh, enterprise apps that do things that are behind the scenes, agents in platforms, the agents themselves and how you interact with them, and a way for you to work with them and your people and your agents and all those things together. That's the whole system. And inside of that, it has context. So it has all the data that I have in my business, and I've solved the memory, the stateless problem becomes different, right? It becomes a system that remembers everything that it does. That's a part of governance. That's also a part of autonomy. The more it can remember and learn, the better that autonomous system gets. I have all those things incorporated into that platform so that my whole business becomes this workforce platform that enables my humans and my digital workers to do what I want them to do inside of the boundaries of those definitions that I put into the system.

Boardroom Value Revenue And Speed

SPEAKER_01 33:21

So, with that said, and then the reason why you would do this is because you want to have a better outcome. You should be more profitable, you should be a better business, you should be going from the horse and buggy to the horseless carriage, uh, so to speak. If you were explaining this to uh a board of directors and the and the C-suite, like, okay, I hit you talked a lot of technical terms. Why would I want to invest in this? And why should I invest in this? Because it seems the world definitely is on this AI train. But help me understand what I gain by deploying agentic AI.

SPEAKER_00 34:01

Well, the the thing about it, I think you have to put it back into business context. And I I can I love to talk about the the stuff we can do. It's exciting. I I love it. You know, I'm a nerd, I can't help it. But the truth is, what should it do? Well, it should help you do things more efficiently, quicker, with more accuracy. Uh, and so what does that mean? It means I could um perhaps build my revenue process so that it generates more revenue because it follows up faster, it interacts more effectively, it brings humans in when it needs to. It, you know, in other words, my sales development reps become agents, but my salespeople use those agents to build out their pipeline and then they close the deal because I've got a collaborative effort. I increase revenue. I go back into the back office and I go, look, instead of closing the books in two weeks, I can close the books in in an hour and a half. Okay. It didn't eliminate the agents nor the humans, and you need both, but now I can do it in a fraction of the time so that those humans can go do other things that add more value to my business. And I have a more accurate auditable outcome from my monthly close or my annual close or whatever. So, in other words, I can do those processes more accurately, faster, better, which helps me increase my margins, or I can go to market, market correctly, all those kinds of things in the go-to market end of the business and increase revenue because of that. Then there's the last thing, your competitors are doing it. So, in the end, if I'm on the board, I've got to think, uh, okay, well, I don't, you know, I could be a leader, I can be a fast follower, but I certainly don't want to be a laggard because I'm gonna miss the opportunity to increase market share, to increase revenue, to build a more effective, better business without having that hybrid workforce. Because the hybrid workforce is just simply better at executing my business. And the more I can build out those capabilities across the business, the more effective it is, the better it is, the more revenue, the better margins, et cetera.

Practical Next Steps And Where To Find Michael

SPEAKER_01 36:13

I think you are absolutely accurate in your statement and understanding first, we got to get our our mindset uh around this and how to deploy. Where does it go? Because it doesn't just go everywhere, to your point. And you do need an expert. Don't think you can just take your, I don't believe, you can take your current IT team, what say, because this is not really a technical problem. There's technical components to it, but you gotta look at everything that's involved and then start looking at the low-hanging fruit. What can we automate today that is a friction point in our business that's going to help number one, our customers to do things better, how how they're gonna come to us as opposed to my competition. So there's a lot of mindset you got to look at, and then evaluate your skill set to your point. You gotta evaluate what do I have in house today, and and and what can I do with them, and then what can I do to bring something in? And those are those tool sets. What are the tool sets that are gonna be there? Because right now you hear out there in the world, like there's a lot of AI projects that have failed. But this is just the beginning. You gotta start looking at, and I always look at time, it's cyclical, right? I brought up voicemail. I remember voicemail first came out. People did not like it, they really didn't. You know, now it's an archaic uh uh tool. Nobody likes voice, you know, it's like getting the voicemail is just not not not not uh they don't look at it as being innovative at all, right? However, if you start getting a a bot that can interact with you and then now can seamlessly put together exactly what you want to do to get interaction done, whether it's a scheduling appointment, sending out emails, doing some of the blocking and tackling that takes a lot of the tedious world. And to your point, speed everything up to a point where you can slow everything down and start doing the things that really create business impact. I like that. So, Michael, I want to thank you again for being on the show. And if you could leave us just what one more thing, what would you like to bring is number one, how to contact you? Number one, tell us about our Ariane research and how to bring you in because you sound like an expert that can actually help me run my business.

SPEAKER_00 38:29

Sure. Well, I mean, you mentioned the building the digital workforce book, and that's really the context of how we've laid out how we look at this, how you build the strategy, how you go through these things. It's a good asset, and and it's something you can find on Amazon. Um, as far as the business itself, we we work with a lot of companies. We work with the providers, so we understand the systems, and then we also work with companies that are you know using those systems or deploying those systems. So, you know, across that, it is very useful to have outside people who understand the way the business works and can learn how your processes work and then tie that together with a strategy that says, here's how you build the digital workforce piece, here's your human workforce, and here's how the hybrid workforce goes forward together so that it can, you know, so that it can be um more effective for your business. Arianresearch.com, of course. Uh and uh then I'd also offer the disambiguation podcast, if you check that out. We we talked to a lot of really interesting people. I already have a dozen or so interviews set up for this year's season that we're starting this week to record. Going to be some really interesting conversations, everything from uh, you know, consumer kind of issues around uh dealing with grief. That's a fun topic that we have you know worked out as a companion, all the way down to the business piece of how do I, you know, how do I solve this kind of problem with this sort of agent system? Uh what are the uh limitations today? What's it gonna look like in the future? You know, all those kind of detailed things. So we we go across the gamut on the on the uh podcast as well. So good stuff. Really um, Grant, thanks for for having us too. I uh I really appreciate the opportunity. As you can tell, I'm really passionate about this stuff.

SPEAKER_01 40:16

It's it's oh no, and that's what we love. We love to tune into the passion, especially of our guests, because that is what gets everyone uh very excited and they understand how do we go forward in this world and know the starts and stops before you take on some of these major challenges uh that you may or may not uh find the outcomes that you're looking for. I want to thank you again for being on the show. I encourage your entire audience to tune in to all the episodes of Follow Brand. They can do so at five star BDM. That is the number five. That is Star S T A R, B for Brand, B for Development, and for Masters.com. I want to thank you again for being on the show and have a glorious 2026.