**[00:00:00]**
Kyle: And people figured out one way of working while your AI native competitor just within a week or two figured out the exact right way, the most efficient way to compete with you and now they're they're hiring people who are like-minded and running faster than you could ever imagine because now you're now all of a sudden you're 6 months, a year, two years behind still dealing with that one thing you were trying to push. And your AI native competitor is eating up market. Right.
Armando: Hello everyone and welcome to The Web Talk Show. My name is Armando Perez-Carreno and today we have Kyle Brueckner. He's VP of AI service delivery at NewRocket. Welcome Kyle. How are you
Kyle: Very well. Thank you very much. How are you?
Armando: I'm doing great. Thank you for asking. It's finally cold here in San Antonio. Very strange thing to see, but yeah, we got like 43 degrees this morning and it's nice nice and chilly which we didn't expect. So, lots of good fun. Tell our listeners if you can a little bit about NewRocket and most importantly how you got to where you are right now. What's your history in the space and what led you to V VP of AI service delivery?
Kyle: Sure. So, yeah, NewRocket is in the professional services game. Specializing in ServiceNow implementations historically and now enjoying a new channel partnership with Anthropic but we're currently building out a sizable team of forward deployed engineers to bring AI solutions and value to our customers. Now that's a collection of services and product. How did I get here? So, I'm an engineer by trade. I left college early, got into the construction industry. That was short-lived. I became sort of like the go-to tech guy around the office.
**[00:02:03]**
So I realized really quickly, hey, this stuff comes easy to me. Whipped up a resume and sent it out there. Ended up getting a job in IT, working in, you know, ServiceNow platform. And then and then things kind of took off from there. I got into solution architecture and consulting and truth be told it got kind of boring a little while getting into 2019 and I'm almost I felt like I was on cruise control just delivering the same old ServiceNow implementation again and again. Hey, my configuration management database is messed up. Kyle, please fix it or you know, I have to build out a new workflow for XYZ. And then and then AI showed up, right? And I kind of I just I embraced it, I think, is the right word. Used it for all the little pet projects that I have, right? The side gigs that every every I'm sure every engineer out there has a little something that they're building on the side and realized very quickly, you know, how how much it can positively affect our lives if you put in the right, you know, control is a hard word, but the right little controls to guide the AI where we need it to go and build how we need it to build, checks and balances, so on. And how fun that is, by the way, balancing, you know, control and governance with efficiency. That's a game in and of itself. But anyway, so I spent some time in the professional services industry, you know, sort of building out what an AI practice should be, what services should be offered, how should they be described, what should the messaging be, right? And working with you know very very talented people to you know sort of orchestrate all that together into into a nice set of messaging and portfolio of offerings came to NewRocket to you know make a positive impact with AI and that's across you know internal efficiencies helping grow this Anthropic team and establishing a great a great portfolio you know with with the larger team with with AI services and products to bring to our customers.
**[00:04:20]**
But it's there's a lot of opportunity and I think you know what's the phrase the spoils go to the brave or I'm butchering that but yeah there's there's a lot of opportunity and I and I think the winners in the world will be those who who collaborate well across various domains sales various sectors sales and delivery and then various various domains and sectors people from different backgrounds and expertise coming together to build out build out great offerings for customers in the services space.
Armando: That's very interesting because it's when we think of and we'll talk about forward deployed engineers in a moment and Anthropic and the whole AI thing, right? But I wouldn't have thought thinking ServiceNow then forward deployed engineers, right? It's like yeah there there must be a logical jump on how how NewRocket got into both. But to give our listeners a guide a guiding bridge between those two concepts, can we start with can you explain a little bit about ServiceNow what it is because a lot of listeners if they're not in maybe the home services or other services spaces they might not understand how big of a tool this is and how how widely used. So can can you give us a little bit of insights into what ServiceNow is and what it means for these businesses that that run with it?
Kyle: Sure. Absolutely. So ServiceNow at its core is a IT service management platform. And it's a has grown since then to offer offer product based solutions to every corner of the enterprise once they you know figured out that the foundation of the IT service management platform was a database to store information and a workflow engine to make workflow. They could apply that to any process, right? So great place to establish standard for activity flow getting work from one place to be and at a lot of large organizations it naturally becomes a hub for orchestration as it can be integrated with a variety of other tool sets and data sources.
**[00:06:43]**
It's you know user interface and API enable and the workflow engine within it enable work to flow in a in a controlled way. So you'll see ServiceNow is a popular workflow tool for highly regulated industries you know healthcare finance energy so on and so forth.
Armando: But yeah that is the ServiceNow platform. Can you can you give an example of in layman's terms what workflows in an enterprise environment mean for those who who really have no idea of all the tech behind the enterprise maybe they work just an office job or they work in the field etc. And a lot of this is it's just hidden behind this magical screen. So can can you give us some insights also into like just an example of what are the types of things that it solves in a day-to-day?
Kyle: Sure. So I'll give you an example. You know, one of the one of the first things that I had to work on I'm stuck choosing between two good simple examples. I'll go I'll go with change. So within IT especially in you have a finance company and you have you have servers deployed to support your trade desk or your loan origination you know platform you're going to have you're going to have servers to support the compute and storage needs. And if you want to change one of those servers swap out memory or push a different set of application code, increase the version of the loan origination app for example, need to record that change and have a transparent record history of what's been changed on this server or the application.
**[00:08:46]**
So this record is referred to as a change a change record and that exists as a record in database. Somebody can hop onto the ServiceNow platform and see it in their magical user interface. They can see who opened the change, who wants to change it, who approved the change, all the details of the change, all this rich meta data to provide a traceable audit history of what happened to this server, this application or whatever is being changed. It be a piece of network gear, right? Everything is getting tracked depending on the industry. So the workflow is that change has to be opened and requested by somebody. Then it goes to a next step or a next phase. It may perhaps it needs to be reviewed and then it would be approved and then it might progress towards implementation. Implementation is fun because it should be automated, right? But sometimes somebody might have to get a task to then manually implement something. That just means maybe they have to press a button to push code or SSH into a server to right so it could be a litany of things. But then after the implementation the work is continuing to flow you have a an audit history of that too because each one of these steps are linked together and fully traceable within the ServiceNow platform makes it very valuable. You might need a quality assurance or a testing to be performed to ensure that the change went as expected and the people who are leveraging the trade desk or the loan origination application they're expecting the actual the right results and then and then it would get closed out. But that that in a picture would be you know what a workflow is in IT service management.
Armando: I really like that example because when we think of software if you're not in the enterprise space and specifically if you're not in a regulated industry you might have a web application or just software in your computer and it's like new update.
**[00:10:56]**
Yeah, sure. Update and or someone does updates or you might even hold your breath have automatic updates on it and nobody's really knowing what's what's going on, right? And so the issue with that is that if a system breaks or there's an attack or something, if by not having all this audit trail of what changed and a process to change everything, you might be blind for a while trying to figure out exactly what changed recently and thus might have been the cause of the issue. So it sounds to me like a tool like ServiceNow allows enterprises to have all this both auditable in the sense that you want to protect assets etc. But also as a tool to then go back and doing a push more and be like okay so this happened where's the cause? What's the most likely cause? Oh this jumps out because this tool just changed or it wasn't it shouldn't have been approved or whatever. But you can see what happened and who did it. Is that does that help in those situations?
Kyle: Yeah, it absolutely does. And I think I think now is a good time to start talking about some some controversial ideology, right? And I think the ServiceNow people should cover their ears. So ServiceNow was built with humans in mind, right? The whole user interface, the data model, everything the process ITIL, the processes that that ServiceNow supports. You know, that that break fix example is great because when somebody sees a server go down due to perhaps a change, it's very valuable for them to refer back to that ledger of what happened so they can investigate and then get closer to a root cause.
**[00:12:57]**
That that that shortens the amount of time that a server may be down in theory and it typically works out that way. But now, right, I'm just, you know, kind of spoiled by being able to, you know, communicate with with a team of AI agents as they are pushing code on their own. And if something does break, then they can investigate, refer back to the git history and the version control and explore it much more quickly than a human could. So this idea of change management designed for human-led process is ripe for reimagination
Kyle: To be you know re-imagined for efficient and reliable use with AI agents which is what you know we try to do at NewRocket for our customers. You know, think on the bleeding edge, be ahead of the curve and bring bring high quality technical solutions to customers that are that are leveraging the best tech out there. So, something probably a little controversial but I won't shy away from trying to push the limits of efficiency and reliability.
Armando: I think you're on point that the agent side of things, what a lot of people don't understand if they haven't had experience in the programming side of agents and AI, there's still that thought that oh AI is this bubble and it's not really useful because they're thinking of the AI slop that they see in posts and things like that. They see content, they don't see AI being used in the real world, especially in software, how valuable it is because of the speed of it. And like you said at the very beginning, if you have the right guard rails and controls and access restrictions and things like that, then they are very valuable tools because of as you're explaining right now, if I have the git history, if I had all the code in a repository and it shows all the different changes for me as a human to go in and look at them, it'll take me a few minutes or maybe some hours to find something that might have broken because of a change Whereas these agents can go through it in seconds and just find out exactly what the root cause was based on those changes and give you a report that you can then act on if you want to, which is mindblowing when you see it in action.
**[00:15:33]**
You just ask, hey, this happened or sometimes you even have them be proactive and hey, this went down, this happened, blah blah blah. They just I ran through everything. Here's the issue. This is what you need to push, whatever. And then and then someone can can just act on it if applicable.
Kyle: Yeah. I think what might be valuable to the non-technical people listening is maybe a short explanation on you know how we go from just interacting with your ChatGPT or Claude or Gemini and getting a response back and recognizing AI slop or hey maybe this thing hallucinated. How do we go from that experience to a team of agents running an entire codebase for right like what's that there's there's one keystone to all of that which is the it's the harness right it's this concept of a of a loop where the agent has an instruction to recheck recheck recheck and every time it rechecks it has access to any number of things designed by the harness engineer. It could be recheck what maybe the context, right? What information, what instruction, what deep knowledge does the agent have access to so it can reflect on and reason and I'm going to use the word think. It doesn't think but think creatively about what its next steps are and then planning, right? Establishing what those next steps are and then you know judging. So then again validating what it's about to do then implementing actually acting in the in the physical world which could be directly on a computer or a server or integrating with another system somewhere externally
Kyle: And then a final validation to ensure that what was changed is actually appropriate based on the initial instruction that it that it received in the first place. So all of this happening within a loop and then multiplied by thousands because many agents are working together with if you have well-defined interfaces and protocols to box agents into a particular responsibility then the power of the potential output becomes astronomical and I love it so much I just I get excited talking about it.
**[00:17:50]**
Armando: It is I every software engineer that I've talked to in the past year has have been having the time of their life in programming because it has changed so much of how we do things but in a good way because we still can use our experience and our knowledge and our tool set and our deep understanding of how systems work, right? That doesn't go away. But now the laying of the bricks, if you will, is done by the tools. And so you can focus on higher level thinking and organizing and orchestrating instead of oh, I put in the wrong colon or whatever and then I have to go back and that interrupts you from from continuing the work, right? So by you managing them, you don't need necessarily turn into a manager. And I want to keep that straight as well because some people are like, "No, I don't want to manage people. I want to do the work." Well, you still do the work. It's just you just go in a different level of abstraction. At least that that's what I think. What do you think?
Kyle: I love this topic. So this topic I refer to as the judgment zone because yes as efficiencies are gained with teams of AI agents. Of course the sales pitch from any AI person out there selling any AI based service or product is going to tell you'll get you efficiencies so you can focus on your strategy right or something high level. And what happens on your journey from going from the mundane work to your ideal part which is focusing purely on strategy and planning you hit this roadblock and it's the judgment zone. The judgment zone is where all the decisions are made and for for the rookie exploring agents and agentic engineering they're going to be hit with many decisions countless decisions from the agents who reach out with questions and they do which is great.
**[00:19:54]**
Kyle, do you want to do this or that? Here are the options. Please decide so we can proceed. Yeah, sure. Bang. And that's easy. It takes 2 seconds. But when you multiply that as you scale into the thousands, you become you become a reviewer of the agents. And then and then this naturally starts I don't know how many months ago it was when I went through this but this starts a new strategic planning of how are you going to go about scaling this decision layer and that's just a management tactic at the end of the day. How do how do organizations scale, right? How does the military scale? It's promoting and placing strategically different decision makers throughout your your you have to say it's an organization of agents at that point, right? Where where there there must be a hierarchy that exists and agents given decision-making power in order to scale the judgment zone. So that the human behind the scenes can focus strategically instead of being that a lot of paper pusher, right? Just making decisions all day long. But yeah, great topic.
Armando: I like that. I think the for a lot of people that's the burden that falls on them and they're like, "No, no, I don't want to make this. Just do it." which is this the dangerous part because if they do it that way then the part that is supposed to be left for the human which is the decision layer and the accountability layer sort of just gets blurred and now they don't know what's going on and something's built and they're not really sure and I'm talking about non-expert people who are just like vibe coding and stuff whatever name you want to put that on and I and I'm all for it by the way I really love seeing what people are doing building their own tools and just out of the blue being able to create things out of nothing. Which sort of used to be this art that only software engineers could do. But the
**[00:21:56]**
Kyle: Maybe a comment on that point before you before you segue vibe coding in general I agree. I think it's a beautiful thing that that technical and non-technical people should try be and multiple times because when you're when you're coding you know in a real sense like vibe coding without an agentic harness or you know without deep knowledge integrated with your agents you're going to get a different experience every time sometimes the a because when you don't guard rail the AI it starts to choose its own path. Paths and there's a there's a number of paths that it can take in terms of design or engineering ingenuity or security, right? Architecturally or low level. They and each time you go through it, you'll you'll learn something new about how to better place a particular guard rail around design or user experience or what kind of database is required or maybe is data model related, you know. So multiple times would be the advice because because you're going to go back and that that gets you on the u the path of getting into better controls and better governance for the agents without sacrificing efficiencies. So,
Armando: I agree. One of the things that I see a lot is that like they always say, you can teach a man to fish, right?
Armando: It's when we show people this is how you can do it. The first there's hesitation at first for many and they don't don't want to try it. And I think a lot of people have that fear of breaking something. And once you get over that and be like you're not going to break anything if you're like just sandbox just a little chat thing, right? Then they do it like oh this is interesting. And you start seeing sparks come out and them being able to own it and say, "Wow, I built this little tool.
**[00:24:03]**
Let me see what else I can do." And then they start doing it. And while I can show someone this is what you should do so that you don't run into these issues, it will help some, but not everyone learns the same way. What I've seen the most benefit is when people stumble as a practice, right? Stumble, as you were saying, into some of these things and then it gets really absorbed when they see, oh, oh, okay. So, this thing was going to expose my credentials or something, whatever issue, right? And then they stop and then they learn how to do it better, especially with a guide, then it won't happen again because now they've they've internalized that situation more than just being taught as if you're you're like viewing slides, right? What's your experience in that when when new people who are new to this just learn something in a course or workshop versus actually trying things out and then getting that knowledge from from real experience?
Kyle: I think I think it's very important. I mean, I'd go as far as saying it's a it's a core leadership philosophy philosophy of my own into parenting and just life in general. I wouldn't I wouldn't recommend trying to break things or trying to fail, but when it when it happens there there's always a lesson in that. And like I have a sports background too and my kid a nine-year-old plays quarterback and when when he does throw an interception. Absolutely there there's a little little fiery rage there. As a parent, how could you not have seen that defender there? How could you have thrown that that interception? But you know there there's a there's a celebration of learning happening at the same time, right? Where a particular read wasn't made on the play. He didn't see the safety drop down, for example, and threw it right to him as he broke on a slant route, whatever the case may be.
**[00:26:06]**
But and then there was a there was a moment a couple weeks ago where my daughter, she's 11, was on the soccer field and she was playing midfield and the ball was it slipped out from the goal and she was she was on an attack vector straight to the goal and the ball just happened to just spit out right to her. She had an open shot and the ball sailed over the goal by 20 feet. And the frustration just in her entire body language was she knew it in that moment like failure. And it was tough, but now she understands what the ball felt like on her foot in that moment. She understands the emotion that just overwhelmed her as she approached that open shot. Right? She has that experience and then now she's ready for the next time that that happens and she'll be better prepared. The same thing occurs when we're talking about engineering applications and servers with this new tooling and or using AI for any purpose. It's yeah, get in there, try it. If something breaks, who cares? Fix it, right? It's going to be okay. And then start accumulating those little lessons because at the end of the day those those grains of sand become mountains.
Armando: Yes. Now having said all this some people are going to be like you can just let people run wild with the thing in your enterprise tooling. Well this is more of a hobby side of things. Once you get into business case uses, then as you were saying, you can have a sophisticated harness that will follow specific rules, have specific hooks that will check things as you're programming, as you're deploying, as you're testing. So that like you were saying earlier with that loop, it's not just, oh, modify this code, send it to main, right, or send it to production. There's a lot of steps that happen and now these agents can sort of loop themselves and I know I know you guys are are experts at cloud but your experience with mods have you guys used it at all
**[00:28:21]**
Kyle: With what
Armando: Claude Code mods? Oh no not no I'm g I'm sorry I'm gonna I'm gonna put you down a rabbit hole. So I have this thing and it go it's part of the conversation don't worry. So, we're talking about how you can have some some companies have their their own very sophisticated harness. But let's say this is a smaller company and they use Claude Code, right? They want to use Claude Code. They want to start building things. But there's rules that they've figured out that they always use or that they want their agents to do. And so what Claude Code mods allows you to do is that you can write TypeScript code that runs within Claude Code. So there's a little chat box at the bottom. It can put content in it, visual content like how many I don't know how how's that complexity, the crap complexity from from Uncle Bob's clean code, right? What how many mutations you have that things like that and you can create these mods that basically make your Claude Code work the way you want it to work. So, if you always wanted to check through its work, if you always wanted to, I don't know, make sure it doesn't go over 150,000 tokens in context and always create a new session after it automatically, things like that. So I created one where I took all of the traditional software engineering principles that I that I adhere to and whenever it's coding something it'll it'll start this little loop internally for every change that it's making after it builds a little feature or something on a branch it will actually go through and run a gauntlet and it's like it checks everything. It's it checks like cyclomatic complexity. It makes sure there's no like random mutations that happen. It checks edge cases. It just goes through and if it fails something, it'll start fixing it. And so your code sort of constantly, you can just code or have it code, but it will constantly get polished as it goes through without you having to tell it.
**[00:30:28]**
So it's just part of the regular conversation, which is nuts.
Kyle: Yeah. I love that. I think it's so important for the frontier labs to keep rolling out components like this which are layers of abstraction which ease ease the use and learning experience to leverage all the little tooling that we have to shape these things to work better for us. This is a it's another great feature. That's probably my favorite thing going through this like and living through this time is seeing how the layers of abstraction continue to roll out and make things easier for us and share the benefit with everybody instead of just like I mean I love to geek out with people when we get into our engineering conversations and we talk about the things we might have built last year and now you know now it's a standard feature in Codex for example, right? It's just you turn it on and it goes and you think to yourself, man, I spent two months on that.
Armando: Yes.
Kyle: But it's good to see. Good.
Armando: I like I like I like that push towards giving the end user, right? Because you get the benefit of having an engineering team right on your side and being able to create these really magical and powerful harnesses that that talk to your knowledge base and company data and all that and tooling where where some other businesses or business owners just simply don't have access to that. And so by these labs like you're saying giving adding these features like they're small little things and maybe not all everyone uses them but they come out very often. They keep shipping things which is nuts. But it ah makes it so exciting. I recently I don't know what you think but I used to use VS Code with the Claude Code CLI the Claude Code extension on the side right and that's how we would do things.
**[00:32:34]**
But then I noticed that they updated the desktop app the Claude app to have Claude Code in it. And I was like, h sort of same difference. But no, they added very little things like you know how sometimes you have to run a command in the terminal and it'll say, "Hey, I can't I don't want to touch this. You need to run this command." And you would have to copy it and paste it in the terminal in your terminal, right? And then follow the instructions, right? In the desktop app, when you're using Claude Code, it'll say, "Can you run this command?" And it'll put it there with a play button next to it. Is you just click the play button and it runs the command in your terminal. So it's not running it itself. It's not seeing the output. It's like it's you press it runs in the terminal. So it saved you that step and then if you have to paste in any credentials there or anything, it's in the terminal. It's out of its sight and then can continue with what it was doing. It sort of holds for you to do it and then and then continues. So it just makes the whole process more efficient and less error-prone as well.
Kyle: Yeah. Yeah. Another another very convenient feature. I can remember going through this. It's like we're growing in stages very very quickly though. Back back when ChatGPT came out and four was released, I remember, interacting with the chat interface, copying things, working solely in VS Code and continuing to work that way as if ChatGPT was a more advanced Stack Overflow, right? And it was a lot of copy and paste. So then we got to the point where there was this traversal. Something happened and I fired up Codex and started using that and I found myself no longer utilizing VS Code. I just didn't need it. And then, you ran into those those other little little requirements about how you might need, a button right next to the proposed command so that it can quickly be run, which is a great feature for somebody who hasn't taken the time to define, hey, what commands are you allowed to run and what are the actual protocols around this so that you don't have to ask me every time.
**[00:34:45]**
That's getting back into that decision zone or that judgment zone that needs to be scaled. But yeah, it's interesting journey so far. Very.
Armando: You mentioned earlier about the hierarchy of if you're using multiple agents instead of just having some people just think, oh, I'm going to have 56 agents and they're all doing stuff. Well, that's one thing, but you were talking about more of a hierarchical structure where maybe some agents are doing something and they have a manager of sorts and then or someone they that supervises them and then from there it goes to another one, another one and then they each has a specific responsibility. Can you talk a little bit more about that? Like in an actual work environment, how how that would how would that be seen? And is it something unique to using a tool that's only enterprise friendly or is it something that other types of businesses can also achieve with with different software?
Kyle: Yeah, I think I think it's a concept that should be explored. And it goes back to exploration. I like to I like to think that, you know, the capability or the ability to explore from a business perspective is a is a privilege and that's only allowed if you believe that even in business perfection is the enemy of progress. So enabling people to innovate and try things and even within the context of the enterprise break things and fix them and only then will you find the fruits of innovation. So with all of that it's a concept that should be explored in the enterprise space because you know enterprise for rest of time until the technology again surprises us and changes transforms our entire lives will be hybrid AI agents working alongside systems or humans rather.
**[00:36:52]**
So agents and humans working together, each having their own roles and responsibilities aligned to what they're really good at. Humans, judgment, creativity, agents, everything else. So within within that model, humans have are they're members of a of a hierarchy, right? They're they're reporting lines. They different different layers have different responsibilities and agents are are primed for for a similar hierarchy in order to increase the rate at which decisions can be made. Because it would be it would be unfortunate to within an agent and human hybrid system the enterprise the new the modern enterprise to drown every person with decisions that the agents are are proposing to them. That's not fair. So in order to scale that the agents themselves should be given guard rails and a path to decide certain things on their own. So yes it should be explored with within the enterprise. This is one of the one of the small components of you know the various frameworks that that NewRocket bring bring to our customers as you explore and innovate to provide more value and grow with this technology as it continues to change.
Armando: At the beginning you talked about NewRocket also having this new term right the FD or forward deployed engineer that it's more and more spoken about now in circles and people start to hear it more but most people have no idea what it is and they're like what is that and how much like it's very it's a very interesting concept and it was very interesting I think to see when both Anthropic and OpenAI started these these big ventures into collaborating with companies to create a bunch of new of these people roles right to then embed in company.
**[00:39:03]**
So can you tell us a little bit more about the forward deployed engineer what that means for an organization and how's it different from a consultant?
Kyle: So it's forward deployed engineer is definitely newer for NewRocket specifically. It's the term has been around and forward deployed engineering has been an applied practice for quite a while now. Companies like Palantir have been doing it for a long time and even before that it could have been referred to as a as a different thing in the past but a forward deployed engineer at its basic core is the consultant and it's a it's like your Swiss army knife the person who walks in the door has a the customer has a particular outcome in mind and this one forward deployed engineer econultant can communicate well with a customer. Can build rapport, handle all the things that require really good soft skills. Understand the business domain. What is actually happening here? What's important to the business? Is there process involved? Is there is there governance to be you know reasoned through? And then also carry with them the technical chops to whip up a solution very very quickly while managing you know the interpersonal communication and everything that goes along with delivering technology which is also people and process every single time. So yeah that's how we look at a forward deployed engineer. We see the model shifting though and we see the model shifting to enable scale. And that's also an opportunity to get everybody else involved who has a variety of skill sets and specialties to collaborate and support forward deployed engineering at its core with project management, right?
**[00:41:04]**
Administration, administrative duties. And alongside of that with project coordination, and then there's strategic thinking, advisory, there's there's u additional engineering support that a forward deployed engineer alone might need to scale. There might be you know process expertise that that might have to be accommodating as well to a particular opportunity. So we're left with this model that we like to think about at NewRocket. It's you have forward deployed engineering and a support system around it to get exactly what customers need in any given moment for any given use case. So a lot of flexibility there with a deep deep not bench but a deep bullpen to pull from. Armando: When you look at a forward engineer being deployed at an organization and needing this support layer, do you see an opportunity there as well for upskilling of the rest of the staff so that they're up to date with AI best practices, how they can use AI such that the same the internal organization can help support the forward deployed engineers initiatives.
Kyle: Yeah, absolutely. And this is I would call it more of a long-term gain as people feed off of one another, learn from one another. When when when people are in the same room talking about certain things, they're all they're all learning. It reminds me of my first job. I sat in the office, remember those days? And I sat next to the person who owned the configuration management database. And I didn't know anything about CMDB at the time. But boy did I after I left that place because he was talking about it all day long on every single call.
**[00:43:08]**
And I heard every bit of it as I was chunking away at my code that I had to write and couldn't focus on because I kept learning forcefully about CMDB. So this is going to happen this propagation of knowledge across all all the various skill sets and personas throughout the entire industry and we'll see a very large migration of people who may have started in the process domain or a project management domain or maybe it's product related and they bleed into and traverse these what used to be firewalls but now they can because they have so many resources out their disposal and maybe they're the new forward deployed engineer to run the show and be in the spotlight. So, that's something I'm really looking forward to and would encourage people listening if they're, you know, if they're in industry enable that for people, right, to traverse those barriers and try something new and yeah, let them follow their heart and what interests them. It's magical.
Armando: That's amazing. We've seen a lot of growth within organizations where there might be this person that is traditionally non-technical but that knows everything that's going on because everyone tells them the things and they're sort of the go-to person and so this person sort of slowly starts to learn AI and do things and then people are like oh hey can you help me do this and like they sort of become the AI champion if you will of the organization and it's an internal I don't know inside out kind of like you were saying growth of it propagates across employees seeing what they can do and it's controversial at the same time because some people are like no no no you have to do AI from the top top bottom instead of bottom up because otherwise you end up with a million different tools and there's there's all this dark IT right shadow IT right but at the same time like I do see the benefits of people starting and proposing and maybe those tools don't don't touch the actual data but they should be able in my opinion to start building things and proposing things because that way they're the users and they can see where the real needs are on the day-to-day.
**[00:45:32]**
What are your thoughts on that? I know it's a little controversial as well, but what do you see?
Kyle: I hope it's not controversial. I would figure that this topic is becoming more elementary especially for management teams out there who have declared that they would drive AI strategy from the top. I would imagine that in all cases that is failing because I think I think it should be driven you know it from the bottom because even if it results in some things breaking or some shadow AI that's where the innovation happens that's where the experience is gained these are the resources that every management team should want to upskill. So forcing things upon them, forcing a tool, forcing a way of work, right? Forcing things is not going to enable this widespread sprawling innovation that every company should be just appetizing over right now. Licking their chops, salivating because like imagining you went through this agenda or initiative to roll out and force tooling and process and protocol as it relates to AI to an entire organization and people figured out one way of working while your AI native competitor just within a week or two figured out the exact right way, the most efficient way. To compete with you and now they're they're hiring people who are like-minded and running faster than you could ever imagine because now you're now all of a sudden you're 6 months, a year, two years behind still dealing with that one thing you were trying to push and your AI native competitor is eating up market, right? So it's yeah I hope that management teams are coming to the realization that this widespread innovation spraw must occur at the lowest part of the organization direct in the resources hands so that people can grow and learn at the pace they want to.
**[00:47:38]**
Armando: What would you recommend to our listeners if some of our listeners are business owners or managers or executives at companies and they're they're thinking about all this new technology and how to embrace it properly but at the same time be careful and all that. What would your recommendation be to get started? What are the proper next steps maybe for them?
Kyle: I think it starts with collaboration and culture, right? You could have, you know, the I think the old school way of thinking about this is, well, let's let's establish a governance board and make somebody responsible for the rules and they can establish the rules and people will follow the rules. If somebody wants to do something, they'll just have to go check with our established rule checker and get permission. That will slow you down, right? I think I think getting to a point where the team can openly collaborate and start to devise what these guard rails are together and then make them service- based. Right? You want to build out a new application, here's how, right? These are all the things that you need, the bells and whistles. You want to you want to create a platform interconnected microservices. Here's how. Right? These are the skills and the deep knowledge that you want to connect your agent to in order to do that. So I think I think it starts with culture and collaboration first instead of you know just trying to sit down and describe rules for technology you barely understand and so innovation sprawl to collectively create the guard rails that'll control it.
Armando: That's excellent advice. Where can people find more information about NewRocket and all this wonderful knowledge that you've been sharing with us?
Kyle: Newrocket.com and you can find me on LinkedIn as well if you ever want to learn more about the company you know just AI bot leadership or how we're thinking about you know bringing bringing value to customers as well.
**[00:49:48]**
So yeah, but newrocket.com is the best place.
Armando: Amazing. This has been wonderful, Kyle. I really appreciate you coming on the show, sharing your expertise, your talents, and your knowledge with our audience. I think this will really help people start to get out there and try things and maybe sometimes break things, but learn from them as they go along. Thank you so much for being on the show.
Kyle: Was a pleasure. Thank you.
Armando: And for those who are listening, if you haven't liked and subscribed to the show, please do so. That way, you will be the first to know when we have an amazing episode like this one with Kyle. Thanks everyone, and we'll see you in the next one.