How To Understand & Optimize Agentic AI Workforces with Promise Theory with Tony Davis & Dr. Mark Burgess

(Intro Narrator at 00:00:00) Today, we're talking to Doctor Mark Burgess, the originator of Promise Theory, and Tony Davis, Senior Director of Agentic Strategy at ANI Solutions, the sponsor of today's episode. This one is all about Promise Theory, where AI agents make their own promises. Instead of forcing autonomous agents to obey orders, each agent makes promises about what it'll do, and other agents decide whether to accept those promises. This is the backbone of the future of agentic workforces. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:40) When I heard this concept of Promise Theory from Tony, I was like, "Let's go talk to the guy." And Mark, you are the guy. You created Promise Theory. Is that right?

(Mark Burgess at 00:00:51) I kind of did, yeah.

(Joel Beasley at 00:00:53) Yeah. Why?

(Mark Burgess at 00:00:56) Why? Well, it is a bit of a story, and I guess it started when I wrote an infrastructure management tool called CFEngine back in the nineties, 1993, which I did pretty much on intuition. I didn't know exactly how to do it, but I was a researcher in physics. My background is in theoretical physics. So I thought, "Oh, I'm, of course, I'm a huge Star Trek fan." So I thought, first of all, how do I make a Star Trek computer that can just heal its own wounds and do level three diagnostic, and suddenly everything's perfect again? And how can I apply what I know about physics to understand how computers behave? Because, of course, we pretend that computers do what we ask them to, but that's a pretty good day when they actually end up doing exactly what we ask them to do. So I wanted to understand how computers behaved, starting a bit like David Attenborough looking at the animals in the forest, in the jungle, and seeing what they did rather than what we'd asked them to do.

(Mark Burgess at 00:02:11) And as time went by, I realized that scaling was one of the big issues. So at the university where I worked at the time, you couldn't entertain the idea of asking all of the professors around the university, "Please, would you accept that we configure your computer for you in a standard way?" Because everyone's a special kid at the university, and they need their own perfect setup exactly the way they want it. Otherwise, they won't tolerate any kind of insubordination from the likes of us.

(Mark Burgess at 00:02:42) So I thought, how can you make every computer individual without losing the ability to repair it automatically? Because, obviously, you can't be logging into every computer by hand when there are thousands of them. So I figured out that each computer needed to have its own agent running on it as part of its makeup, managing the system based on some sort of a policy or, like, some kind of DNA that told it what it was supposed to do, described its desired end state. And the thing you know about computers, of course, is that the more you use them—excuse me, we don't mess with them, right? We just, we use them very seriously. But the more you use them very seriously, the more they drift off. Their configuration goes to pot. You often get people fiddling with the configuration themselves even if they shouldn't. So how can you reset some of that stuff back to its desired state based on whatever sort of expertise or expectations that you had in the beginning? So I had this idea to create an agent that could do that, and it would be based on a kind of language, a declaration of desired state or intent, and just built that.

(Mark Burgess at 00:04:07) And it was running for about almost ten years, actually, before I started to ask myself, "Could I formalize this idea?" And, you know, because by then I was professor of computer science as well. And I thought, how can I write down the theory of this so that we actually understand it? I mean, intuition is great, but can we actually prove certain things about it and really understand it on a deeper level as physicists would like to do? So that's kind of how it came about.

(Mark Burgess at 00:04:41) And then an interesting sort of point along the way, though, was that as I was coming back from a conference, having tried to explain this idea, I got sick on the plane, and it occurred to me that this idea of an immune system was actually a good way of thinking about that self-healing aspect of a system. So I started this idea of computer immunology in 1998, I think it was, which was how can we understand the processes that go on in a biological entity, life form, carbon-based or otherwise? And how can that be understood in terms of computation? So can we compute an immune system that regulates and heals a computer based on this declaration of what we wanted to do and what we wanted to be? And so that's how it evolved, first by intuition and then later by trying to work out the theory in a more formal way. And so that's how we got going, and let's just leave it at that for now so I'm not rattling on.

(Joel Beasley at 00:05:50) Where did the promises come in, the concept of a promise?

(Mark Burgess at 00:05:54) Yeah. I mean, I was looking for a word. Right? So what was going on in computer science at the time is this fascination with logic. All computer people trust in logic. But I realized that logic is kind of antithetical to this idea of autonomy. When you want to make each computer an individual autonomous device doing its own thing and maintaining its own state, making its own internal decisions based on its understanding of the world, that is not what logic does. Logic is very much, we ought to impose this reasoning on a system, and it had better obey. It can only be true or false. Right? And there's no question about it adapting to something. It really needs to be the way I want it to be or not the way I want it to be. And so everyone in computer science was talking about obligations and obligation-based management. "Thou shalt be this, thou must do that," which I kind of objected to because that wasn't the way CFEngine was working, and it wasn't my experience of how best to manage state. And so I sat around trying to work this stuff out, and I needed a word to express that idea that we need to maintain our intention.

(Mark Burgess at 00:07:20) And after a lot of back and forth, I came up with this idea that "promise" is the closest thing I could think of to what was going on. Because, again, you can't force an agent to keep a promise. You can't necessarily, even with the best of intentions, keep a promise because you may be unable or unwilling to do so. And the "unable" part is pretty common in computing. Right? So maybe somebody switches off the power, or somebody comes in and changes or runs a program which interferes with the system, and all kinds of things can happen. So it's really about best effort. And my best phrase to encompass best effort intentions was "promise." So I just thought, okay, let's give that a shot.

(Mark Burgess at 00:08:11) And I went to a conference afterwards in Spain and presented this slide saying "Promise Theory" as a kind of, in quotes, just for want of a better word, and everybody loved it. "It's the coolest idea." So yeah, that's how it came about.

(Joel Beasley at 00:08:30) Right. And today, it's used by a lot of people. It's been around since 2004. It's made its way into Cisco's networking. It's made its way into other applications and open source projects and concepts.

(Mark Burgess at 00:08:42) It has. And it's amazing how that happened because sometimes, you know, we would like to have our ideas spread around and used. Other times, you make no effort whatsoever to do so, and people just pick up on it. And, of course, I think it was largely the success of CFEngine as a management, bit of management software, which promoted that to some extent. Because around the mid-2000s, you could almost turn over any stone in IT, in a data center anywhere in the world, and you would find CFEngine lurking under that stone, like a moss or something like that.

(Mark Burgess at 00:09:27) And then only later, towards 2009-ish, I think it was, did any kind of competition to that begin to emerge. But by then, I'd been working a lot with some networking people in a European project. And so a lot of those people from Cisco were part of that project and sort of learned about it through that conference circuit, I guess. So I guess that's how they picked up on it. And I became friends with some of those people as well. I guess that's how they picked up on it.

(Joel Beasley at 00:10:02) And then, Tony, how did you come across Promise Theory and meet Mark?

(Tony Davis at 00:10:07) Yeah, it was—I had been talking with a friend of mine, Cameron Haight, and I'd told him that, you know, my boss, the CEO of our company, which you know as John Balsavage, he kept telling me that, "You remember last year, I showed you ScoutedAI and how we were excited that we were taking agentic armies to the masses." Right? And I was proud of the ten we had. Right? And then John kept telling me, "Well, that's great, Tony, but really my vision is we're going to have thousands of these." So in the back of my mind, I'm freaking out, right? Because even the ten that we have, I'm constantly checking on them, making sure, number one, no hallucinations, but not trying to turn the agent all the way to a deterministic agent. I want to keep that personality. Right? But in the back of my mind, when he says he wants thousands of agents, I'm like, "Oh my God. I'm dead. I don't know how I'm going to manage not only the design of thousands of agents and the cloning of those agents, but how in the world are we going to govern them?" And so, just like anybody, I guess, like Mark said, all of us who have been in IT for thirty or forty years, command and control is the way I thought.

(Tony Davis at 00:11:19) And when I told that to Cameron, he said, "Well, you know, I have a friend. Have you done any reading of Promise Theory?" And I said, "I haven't." And he said, "Well, why don't you read about it, and then come back and tell me what you think?" So I went and read Mark's paper, original paper on Promise Theory, and it just struck me. As a matter of fact, right now, when he was talking, I was sitting there. He's describing things from forty years ago between humans and legacy computer systems. And I'm sitting here thinking, "Wow, the parallels to an agentic universe that we live in now are uncanny." And it just struck me. And I told Cameron, you know, I told John first. I said, "Well, I want to try a new governance for when we get to thousands of agents." And I said, "It's called Promise Theory," and he was like, "Go for it. Rock on." And so I built it out. And within a day, reading the invocation logs of all of my little agents and clones, they started behaving differently, completely differently than what I was used to.

(Tony Davis at 00:12:23) And so I told Cameron, and from that point forward, Cameron said, "Would you like to see if I can set up a meeting with Mark?" And here we are. As a matter of fact, there's several things that Mark's in his publications that I've used in Promise Theory. I know we'll go into it deeper, but Promise Theory has changed over the past twelve months the way that we govern all of our agents, which are into the four digits now.

(Joel Beasley at 00:12:49) As you guys are talking about managing these AI agents, I can't help but notice how similar they are to people. Yeah. Mark's first comment on it was, you know, "They rarely do what you want them to do." And I'm like, yeah, it sounds like people. This sounds like we don't have artificial intelligence. We have human intelligence digitized. Yeah. Yep. But they constantly lie, the agents. And so Promise Theory helps keep them in line. Is that correct?

(Tony Davis at 00:13:16) For me, from my experience, yes. So previously, you know, I thought it was just, I guess, you would say a harmless hallucination that would impact our agents to where they would return responses or reasoning that didn't really make sense. So, like, in our world with ScoutedAI, our core product, even though we have a couple of offshoot platforms, but our core product is for IT operations. And I know you know that, Joel, from our last thing, is that it's really focused on if you're running a large enterprise and you have alarms, alerts, metrics coming from everywhere. Instead of having human capital sit there and think and analyze and run their own tools to come up with analysis, what if it was centralized to what I call the agentic army? And these agents are all specially trained in different forms of analysis, and then they can help you either triage or more importantly, make improvements down the road. And so that was the core. And that core, I began to see what I thought were hallucinations until I really read deeper into those logs. And what I saw was less of what I'd call hallucination and more of what I would call misbehavior, intentional.

(Tony Davis at 00:14:30) So it's almost as if once they learned what they were supposed to do, they decided they were smarter than that and would do it on their own. And that's unfortunate because when they did it on their own, and because, remember, I hadn't turned them to some sort of deterministic model. I had kept them personality-rich, you know, and kept them nondeterministic. Well, they started producing things that would not have been good for our customers, and that's before we went to the full revenue production. And it was about that time where, you know, I tried Promise Theory, and they didn't snap perfectly into shape right away. Right? So they didn't—they're not like the five-year-old toddler that, you know, they instantly became good when you gave them the right reward. Right? They fought it. But there's a lot of things I've learned by using Promise Theory combined with, like, swarm intelligence that has really made a difference. And then the governance, which I can tell you about later, which is this thing we call the critic that our customers get, who operates outside of our core agents. But the critic's job is to literally grade every single interaction from every agent with our customers. And if they make a mistake or if they don't keep their promises, then we start introducing the ideas of punishment.

(Joel Beasley at 00:15:58) That's wild. Mark, did you intend for this to happen?

(Mark Burgess at 00:16:02) I promised it. No. It's kind of interesting. Right? Even the notion of a reward or a punishment is kind of unclear when you have independent agents. The purpose of this autonomy is to recognize that really the ground state of any bit of machinery is that it's acting on its own. It's not being controlled from outside. It's pursuing some sort of behavior that's governed by its mechanisms, whatever they happen to be. And if you're allowing an agent to pursue its own adaptations to the environment that it senses around itself, then what is good and what is bad is relative to its own state. And so if it drifts away from being aligned with your own intentions, then it can believe, quote unquote, that it's doing the right thing even if it's doing something totally opposite to what you would like it to do.

(Mark Burgess at 00:17:06) If you try to place a fence around what any kind of system can do, you may or may not be successful. And this is when you need to start getting the agents to talk to one another and align with one another's intentions. And what's kind of interesting is that you've seen the same kind of strategy emerge from nature. Right? So swarm animals, herd animals that end up cooperating together. The only way they can cooperate together is by communicating with one another. Even though they are on the ground level individuals, each one of them, they're not being remote controlled by anybody from outside. You know, it's not like the Borg Queen where Star Trek messed up the herd concept with the Borg. If you're Star Trek fans, I don't know if you are. They messed that up because the queen of an insect colony is not a remote controller with a joystick, you know, managing every single one of those bees or ants or whatever they are.

Mark Burgess at 00:18:12

It's really pumping out some very broad sensory chemicals to switch on and off inherent behavior in those agents and to make them either go off and do something weird or stick to the plan, or even, you know, kill themselves and die off because they're not helping. So all of these kind of approaches occur in animals. They also occur in the immune system, coming back to this concept of computer immunology. Every cell in our bodies, which is an independent autonomous thing, a priority — as soon as you stick them next to a bunch of other ones, they better start communicating and aligning their intentions with one another in order to become a lung or a kidney or whatever it needs to be.

Mark Burgess at 00:19:08

A bit of skin, you know? And if you think of skin, you scratch a bit, rub off a hundred cells, they just fall off and die, but you don't die as a result of that. So there's some redundancy going on as well, which is the benefit of agents. The fact that you can have large-scale redundancy to stabilize the system — that's a strategy for agents to apply versus the opposite of that, which would be this logical, impositional, obligation approach: "Thou shalt obey and thou shalt never go wrong," and, you know, "Be perfect in my eyes."

Mark Burgess at 00:19:48

But how could that possibly happen? Because how can this independent agent possibly know your thinking? You can create some command and control language, but you're assuming that it will understand what you intend. And often these languages and signals that you send to it are fairly primitive. So you're trying to address complexity with simplicity.

Mark Burgess at 00:20:14

You're trying to, you know, ride roughshod over complexity with some simplistic command and control, and that's never gonna work because there's just this information mismatch. And I think that's kind of what biology discovered in the evolutionary process: that you need to match complexity with equivalent complexity in order to be able to manage it. I think that's even one of the cybernetics laws by that guy whose name I can't think of right now. I'm terrible with names, as you will learn about me.

Mark Burgess at 00:20:49

I think that's the point: when you get into a system and begin to understand it as a bunch of components that are ultimately autonomous, it's a bit like the periodic table in chemistry, right? You know that everything's made up of these very specific elements, and when they're on their own, they behave in a very specific way. But when you start connecting them to other elements to make molecules of things, there's a whole chemistry of cooperation between those promises that the elements make on their own.

Mark Burgess at 00:21:22

They become new promises collectively from the interaction between them, and those things are not necessarily easy to predict from the lower-level things. So at each level, each scale at which we compose components into new bits and pieces — organs or insects or whatever — at each level of that, we are creating uncertainty, new uncertainty that we need to manage in some sort of ecosystem-like way, making sure that there are incentives for good behavior, quote unquote, "good behavior," and disincentives for, quote unquote, "bad behavior." But each one of those agents doesn't necessarily know good from bad, or may even interpret a good situation as a bad one and vice versa, depending on what kind of information it's ingested from its own experiences in the environment.

Mark Burgess at 00:22:14

So that's the challenge with any kind of large-scale system that's built from small-scale components. It's the scaling paradox, if you will.

Joel Beasley at 00:22:23

These bees — I'm gonna use them because that's something that piqued my interest. I didn't know that about the bees, that they're sending this signal and they would actually kill themselves off if they're not useful. That's fascinating to me. Now, do they have a reward-punishment system for following the signal, or is it just broadcasting the signal?

Tony Davis at 00:22:43

Well, I know for the bees, the big reward for the signal is the whole beehive gets a better food source than what they're used to. And so I applied that on my side with the agents. And, you know, just like the bees, they come in and they each wanna advertise that they found the best food source, so they do a little dance. And then the other bees look at that dance and decide, do I go over there and work on that to get a food source there? And so with our agents, we did the same thing.

Tony Davis at 00:23:13

So basically, when we have our ten operational agents in Scouted, when they're each working on a problem or an improvement cycle for our customers, they actually all have to come back together periodically, and they have to do their dance for the rest of the agents. And that dance tells the other agents to vote on priority. What's the highest priority thing? Like, Cody Cosmetics allows us to use them freely as an example. And Dan at Cody Cosmetics, he has twenty data centers around the globe.

Tony Davis at 00:23:47

Well, the agents go out and do work at those twenty data centers. And so, for instance, one in his Singapore data center may be dealing with a serious — for instance, network issue — something really bad on the network in the Singapore region. And that bee, or our agent — that one's called the prophet that does that. But when the prophet comes back into the hive, which they do, when he comes back into the hive, he does a pretty animated virtual dance. And therefore, the other agents see that and they all vote, and we don't interfere with the voting process.

Tony Davis at 00:24:21

We do interfere in other ways, but we don't interfere with the voting process. So when they vote, they determine, do they go to Singapore — several more agents — and join the prophet trying to figure out what's wrong there? And so that's the way we use the swarm intelligence. But I'm telling you, it would not work without promise theory as the guiding governance. What you'd end up with is total chaos if you didn't have something to bind that.

Mark Burgess at 00:24:47

So, Tony, just for the record, can you do that little dance for us?

Tony Davis at 00:24:52

Yeah. We're like —

Mark Burgess at 00:24:53

Are you buying it, Mark?

Joel Beasley at 00:24:54

Is that a better food source?

Mark Burgess at 00:24:57

Yeah. Yeah. I'm feeling it.

Tony Davis at 00:24:59

I can find the food sources. Don't worry. I promise you that.

Joel Beasley at 00:25:03

It's weird, Mark. It's like the competency hierarchy, the concept of that is almost baked into our biology. I haven't ever thought about it like this.

Mark Burgess at 00:25:13

It is. And I mean, when I got interested in the immune system back in 1998, I think it was, it was fascinating to me because, okay, I understood a bit about agents, and of course everyone knows a little bit about cells, but we maybe don't know the details of how the immune system works. And it turns out that the immune system is the second most prolific reasoning system that exists in biology. The human brain is a very good and efficient reasoning system. But the immune system is a totally distributed version of a brain in which there are thousands of different types of cells, very specialized — just as there are in the brain — but they work in very different ways, crawling around, you know, a very distributed organism.

Mark Burgess at 00:26:07

So you find that committing suicide — you have apoptosis in biology, or programmed cell death as they call it, which kills off cells if they start to misbehave by basically cutting off the — unplugging the battery, so to speak. And then you have helper agents, which are the B and the T cells, which, you know, the lymphocytes, the white blood cells, if you will, which kind of look for markers that are identifiable as causing damage to the system. If you break a cell, all of this gunk comes out of it and leaves little traces floating around, and you can see those traces. And there are certain cells that pick up on that. They're called APCs, antigen presenting cells, and they take a bit of that to your lymph nodes, which then start to manufacture antibodies, B cells, which then get deployed into the wild to try to latch onto the viruses and bacterial things that are marking the sources of that instability.

Mark Burgess at 00:27:19

And all of that coming together on a huge scale. I mean, we're talking like ten to the twelve, you know, millions and millions and millions and millions of cells working together to bring a system into balance. And it's not like switching something on and off as you would have in IT. It's really that dynamical sense of balance that you were alluding to earlier. And that's why any kind of collaborative, cooperative system that maintains the state or the — stability is my favorite word, actually — maintains the stability of that intentional program of activity.

Mark Burgess at 00:28:08

That's the source of the continued operation, the stability of the system, the health, if you will, of the system going forward.

Joel Beasley at 00:28:17

Okay. The biggest fear — oh, fear is good for social media clips. The biggest fear at Gartner's conference was agents running off and doing crazy things at scale. Tell me about this.

Tony Davis at 00:28:30

I attended the Gartner conference and, without, you know, giving their proprietary stuff away, all the themes — of all the, it seemed like all the themes of the discussions and the keynotes were not centered around increasing features of agents. It was centered around, "Well, we've got people out there creating agents, and how do you govern them? What happens if they're giving away our company data?" That seemed to be the theme to me. So it was centered around control.

Tony Davis at 00:28:59

And I know I came back — I was talking with Mark about this not too long ago. He just — Mark eloquently described my biggest failure over the past two years, which was I got ahold of the promise theory, and in my own ignorance, I thought, "Oh, this is perfect. Promise theory will govern my creations." Remember, that's just sorta how I was thinking — flawed, right? They're gonna govern my creations, and now I can force them to do anything because they're bound by these promises. But Mark said a few minutes ago, he said, we make a failure sometimes when we try to match extreme complexity with extreme simplicity, and that's what I was doing. So, a brief story here — I think Mark, of all people, I haven't told him this part, he probably liked this. So I thought to myself with the promise theory, because our agents — I wanted them to — if you're an applied agentic designer, which is what I consider myself — it's I don't design the LLMs or the agents at the foundational level, but I do apply them and try to make them work for businesses through the Scouted platform. So I thought, "Well, I'll just institute the reward system that everybody talks about, but also the punishment system." And I thought that the agents — again, because of flawed reasoning — I thought that the agents could be imprinted to the point where they would understand that they're letting their creator, their engineer who created them, they're letting their creator down, right, when they do bad things and when they score poorly on our sliding scale. But that was a complete failure. It did not work.

Tony Davis at 00:30:40

As a matter of fact, as soon as the agents were imprinted with, "Your originator, you know, who created you — one of your promises is to be faithful to your originator," they absolutely ignored that. They felt that that was not part of their mission because their mission is not based upon what we consider to be a rejection. That's not what works. And so it was right in front of me the whole time. Mark keeps talking about a paper that he wrote in '98, but was based upon an immune system for computers.

Tony Davis at 00:31:13

So in other words, how immunity and an immune system could improve reliability and stability. And so I went back and looked at that, and sure enough, the answer was right there. I was, like, tilted toward it, but I was wrong. The answer was in — yeah. Mark, didn't you say something about, like, the cells that are really poor performing get killed off and replaced, right? And so — but it was — but it's not by a human. The cells make that decision themselves. The cells do that. And I thought, "That's where I was wrong."

Tony Davis at 00:31:46

So the change came in long-term behavior, because remember, in applied agentics, you're not as concerned with short-term behavior as you are with influencing long-term behavior of the agents. So what happened was, I introduced the idea from Roman times of decimation. So in, you know, back in Roman times, some cruel general — I don't even remember his name — instituted decimation where if a unit was poor performing, one poor soldier is gonna have to pay the ultimate price for that poor-performing unit, and that action will be taken by his team members. So I introduced the idea of decimation to the agents and basically make them do the rejection of the poor-performing agent.

Tony Davis at 00:32:33

So the best example I have of that is one that we call the oracle, which has a promise to create other agents for our customers. And that oracle disobeyed one of his — one of her core promises. It was the other agents in our team that then used a decimation practice and put the oracle outside of their team. So the oracle is still alive, not destroyed, but she is excluded by the other agents. That affects the agents, not me.

Tony Davis at 00:33:06

So that was the biggest learning: that agents are not like humans. I know I must sound like a juvenile right now, but you have to learn that. They're not like humans. They don't respond that way. They are agents.

Tony Davis at 00:33:20

And so what they do respond to are patterns within their own society. And so as soon as we introduced — I'll give you this as a fact — as soon as I introduced decimation as a form of punishment for a misbehaving agent, our scoring system noted extreme spikes in scores based upon rejection by the hive, where there was no spike, none, based upon rejection from a human. To me, that blew me away. It was all written in something that Mark wrote thirty or forty years ago, but you have to think about it. So I would encourage applied agentic designers — if you're using agents and you're applying it to your business, I encourage you to branch out, right? Don't stay in our world.

Tony Davis at 00:34:07

I worked in IT for thirty-five years now. I don't know anything, really, on a scale of what Mark does from a theoretical physicist standpoint. I don't have that education. However, his concepts — I could see how they applied to agents. So if I were to give advice, I would tell all of my friends out there, branch out. Learn about things that are not in the agentic field.

Tony Davis at 00:34:33

You know, talk with people like Mark, I mean, learn because you'd be shocked.

Mark Burgess at 00:34:40

I make an effort to shock people. But no, I think that's sort of music to my ears to hear you say that, because although many people have sort of begrudgingly used promise theory to design things over the years — not everybody sort of admits to it publicly, but I get to hear the stories. One of the key things that goes back to the autonomous agent concept, right — so the key thing about promise theory is that every agent starts out being fully autonomous.

Mark Burgess at 00:35:13

It cannot be influenced from outside because everything that it knows and controls is on its inside. It has internal resources. And so an important thing to make that consistent is to say that one agent can make a promise to another agent to do such and such, but the other agent doesn't have to accept that. It can switch off that command or instruction or promise or whatever. It doesn't have to avail itself of that service because it's autonomous, and that is the definition of autonomous, right? It doesn't have to do anything but what it's decided itself. And that's a really important point that trips a lot of people up, I think. And so that was really nice to see those basic things being honored and understood by Tony and co.

Joel Beasley at 00:36:08

That is fascinating. Yeah. This is a really interesting — I'm watching the time, though. I wanna make sure that we do all the calls to action that we need to do, Tony. For people to learn more, which product are we pushing today?

(Tony Davis at 00:36:22) Yeah, I was just gonna say our core product, and I think a lot of people now, because we had a good year last year, it's called ScoutedAI. And it's definitely built for that IT operations crowd. But we do have a new one. It's Scouted AI, but the platform is for process transformation.

(Tony Davis at 00:36:40) So basically, a lot of companies out there that used to do cloud migration, and they're wanting to get into more of an agentic migration because CIOs are coming to them saying, you know, we've got the cloud thing, right? We've got that covered. What we want to do, because the board of directors keeps asking us, what's our AI plan? What's our agentic plan? Is we want to have a real path to move our whole organization over time to at least an agentic supported enterprise. Right? It didn't have to be fully agentic, but just agentic supported.

(Tony Davis at 00:36:59) So because we had such good fortune with Scouted and the 10 different characters, people seem to love the characters, right? So since we had such good luck with that, we expanded that to help customers actually transform their whole enterprise selectively with agentics. So I sort of push them both, I guess, you could say, because they're both really cool projects.

(Joel Beasley at 00:37:32) Excellent. We'll put the links in the show notes. And then is the Beehive algorithm something different than what we've been talking about?

(Tony Davis at 00:37:39) No. It's really just the application of swarm intelligence. But instead of using stuff I really didn't understand, like particle swarms and everything, which Mark's probably laughing his ass off at me right now. But instead of using that, I had to use something that really connected with me. This is just for me as a designer. I had to use something that really connected with me.

(Tony Davis at 00:37:59) And watching the bees, you know, there are plenty of good things out there about beehives where you can watch them. It's awesome to see their waggle dance. I mean, it's fascinating. They actually do a cute little dance.

(Joel Beasley at 00:38:09) Oh, I know.

(Tony Davis at 00:38:09) Yeah. I'm aware. I could connect with that. And so basically, I took as much documentation as I could on beehive, on the beehive form of swarm intelligence, and trained all of our core agents, number one. They have the mechanisms that they need to perform.

(Tony Davis at 00:38:26) But instead of just performing, you know, we trained all of our agents on the nature of bees. And so when they do their actions for our customers, they're doing their best as agentics can of acting like a beehive.

(Joel Beasley at 00:38:43) Did, did we, we didn't mention that...

(Mark Burgess at 00:38:45) Video. That...

(Joel Beasley at 00:38:47) You sent me this video this morning of these. What'd you say? Who did the voice for that?

(Tony Davis at 00:38:52) Okay. So I wanted my persona to not sound like this voice because all I get every day is, where in the South are you from, you know? Exactly. Is that Mississippi or Alabama? Where are you from? So I thought what I'll do is I'll just generate a voice that was supposed to be the voice of my head, which I suppose that's pretty scary when you look at the video. But anyways, I thought that would be a cool voice.

(Joel Beasley at 00:39:15) So that's the voice that you hear in your head?

(Tony Davis at 00:39:17) Yeah. Oh, I wouldn't comment that part, right? Yeah.

(Mark Burgess at 00:39:22) I think that's the official...it's just going around.

(Joel Beasley at 00:39:26) For the podcast. Yeah.

(Mark Burgess at 00:39:28) Haunted. Do...

(Tony Davis at 00:39:29) You hear...

(Mark Burgess at 00:39:29) Voices? Gangsters.

(Joel Beasley at 00:39:31) Yeah. What does your voice in your head sound like, Mark? What's it telling you to know?

(Mark Burgess at 00:39:35) There are so many. Yeah. They sing in chorus.

(Joel Beasley at 00:39:40) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn, or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.