A Day in the Life of a Forward-Deployed Engineer

Jordan Cab: Foreign.

Brian Gracely: Good evening wherever you are. Welcome back to the Enterprise AI Show. I'm your host, Brian Gracely and today we've got a fun one. We are going to dig into a couple of topics that we cover a lot on the show, but I don't know that we've necessarily sort of dug into the, well, what's it like in reality? And so we're going to do that with both digging into forward deployed engineering as well as harness engineering and, you know, do that with, uh, some folks that are building some new cool technology as well as living this life and sort of, you know, understanding what it really means to implement some of these technologies. So some good stuff coming up right after the break.

Narrator: This episode is brought to you by Outshift, Cisco's incubation engine. AI agents today operate in silos, limiting their potential. Scaling up models isn't enough. We need to scale out like humans did 70,000 years ago. And agents must share knowledge, goals and innovation to evolve. Outshift by Cisco is building the Internet of Cognition, transforming isolated AI systems into orchestrated superintelligence. This open infrastructure enables agents and humans to share intent, context and reasoning. The cognition evolution for agents is here. Learn more@outshift.com that's outshift.com. Today's show is sponsored by sharegate. You're out of time. Copilot needs to be deployed asap, but your tenant really isn't ready for it. Years of data, permissions and users lurk in its shadows, ready to be exposed by AI. Sharegate Protect sees it all so you can find the exposure risks, fix them Fast, and deploy AI with confidence. Microsoft 365 Governance they've got this. Learn more@sharegate.com Protect. Today's show is sponsored by Nasuni. There's a growing gap in AI right now between what's possible in theory and what successfully works at scale inside an enterprise. The difference comes down to unstructured file data. Many AI initiatives struggle because the file data they depend on is scattered, unstructured and disconnected from where and how work actually happens. Nasuni changes that. It brings your unstructured file data into a single secure foundation so AI, both generative and agentic, can access it with the context of governance and performance it needs in production. Bring AI to where your unstructured data lives. See what it takes to activate your data for AI and request a demo@nasuni.com AI

Brian Gracely: We've talked a lot, at least in theory, about a couple of topics that we're going to dive in today. Both Forward Deployed Engineer and Forward Deployed Engineering as well as, you know, AI harness and sort of harness engineering, things that we've talked about somewhat, I feel like we've talked about them somewhat in theory. And today we get a chance to dive in to, you know, talk with somebody who is doing this live, doing this as their everyday day job. And so excited to have Jordan Cab with us, staff Forward Deployed Engineer for Air and D at Mozilla. Jordan, welcome to the show. Great to have you on.

Jordan Cab: Yeah, thanks man. Great to, great to be here. Looking forward to it.

Brian Gracely: Yeah. So, you know, sort of we get, we get a double bonus. You're also back in the Raleigh area. So we always like to, you know, reach out to folks who are, who are near, near Aaron and I and uh, been doing this for a while. So before we dive into kind of what you're doing every day in your day job, as well as some of the cool stuff you're doing at Mozilla, give us a little bit of your background. You've worked in a lot of places, a lot of different roles, but kind of a little bit of your background and then what you're focused on these days.

Jordan Cab: Yeah, yeah, sure. Great. Great to be back in Raleigh. I grew up here, went to school here. Yeah. I've been a software engineer, kind of into computers since I was a little kid, like 12, 13 years old. That's, that's how I got started. I like to joke that I was very much an indoor kid, if you will.

Narrator: Yeah.

Jordan Cab: But tried to start a couple of companies when I was in middle school, high school, which is not a great time to do it, but definitely tried. Got kind of burnt out right before college. Studied political science, which is kind of a strange, a strange twist. I went and worked in politics for a little bit and found out that I much more enjoyed tech and having a product rather than, uh, working in politics. So, uh, after college, went and worked in fintech, social media. I worked for a medical, uh, device company that made medical devices for veterinarians. I kind of had no focus. Right. I was living in New York at the time. This was 2023. ChatGPT, like 3, I think, had just come out and uh, you know, it was, it, it blew up. The Internet blew up. It was all anybody was talking about, especially in my circles. Right. So my brother called me again this spring. 2023. He had been a consultant at KPMG for a decade, trying to bring machine learning and AI to contracts. And he called me and he was like, hey, you know, I'm going to quit. I'm on the partner track, but I'm going to quit and I'm going to try and start a company and do this on my own using LLMs M. Do you want to join me? And I was like, I don't know, going into business with family can be kind of, you know, sure, yeah. Can be dangerous, can be sketchy. I trust my brother. I was like, you know, I'll build you a demo, I'll build you an mvp because I know you're going to go fundraising and like, that's it. Uh, that'll be the extent of the relationship. Fast forward six months. I was cto. We were on the roadshow, raising money. We raised all of our $2 million seed angel round right here in the triangle. Didn't take a dollar of institutional capital. It was all from angels network, friends, family. So it's really cool. We built a really organic, organic cap table. We had a really cool product, really cool customers. We were in contract analytics, right? So pulling structured data out of unstructured contract portfolios, credit agreements, things like mortgages, things like that.

Brian Gracely: Okay.

Jordan Cab: Made a good run of it. And we shut down in late summer, fall of 2025. So last fall, and then in January, joined Mozilla. Not a company I ever thought I would work for. Not a company that I knew was doing anything with AI, but a longtime friend of mine and now my manager, Chris, Chris Roth at Mozilla, brought me on and joined the team. And then in the spring of this year, we launched Thunderbolt, which is Mozilla. It's underneath the Thunderbird team. So if you go look at the repository, it's Thunderbird, Thunderbolt. But it's still under the Mozilla flag. An open source, local first, privacy first, AI harness across every platform. And on that team, I am a staff, fully deployed engineer, which is very nice. Kind of a new title in the industry, I think. Not new, but become more popular.

Brian Gracely: Yeah, it feels like something over the last couple of years. And this is, I want to dive into that first. I, you know, listening to that story, I feel like we might have to have you back just to talk about sort of the, you know, the realities of starting a company in the AI area. Right. You know, two guys, you know, there's lots of two guys started a company thing, but what does it mean to do it when you've got these AI tools in your hand? You're out in that space and we'll have to have you back on and maybe tell that story as well. But yeah, it feels like forward deployed engineering has become this, this talking point, this buzzword. And I'll be curious as to what exactly you're doing with it. I think a lot of the times we hear it's, you know, companies would like to do things with AI, they sort of have an idea of how to do it, but they're not exactly sure how to integrate it with their business or how to best use the tools around it. And so, you know, forward deploy feels like, you know, it's almost like the Navy SEALs or some, you know, some group that's going to be embedded with the customer. They're going to get to know their business, they're going to, you know, teach them about the technology, they're going to be adapting it as they go. Is that anything like, you know, your world these days?

Jordan Cab: It is. So I think, you know, there's a lot of times I think the role gets conflated with the titles like solutions Engineer, sales Engineer, Sales Architect.

Brian Gracely: Yep.

Jordan Cab: And it's similar in that you are definitely part of the sales process but your job doesn't end once the sale is made. So I know that in some cases, some organizations a uh, solutions engineer is up there until, you know, along with the customer until the day the contract signs. And then you know, they go through an account executive after that to get, to get their bug fixes and things put onto a Sprint calendar. What I honestly think that this role grew out of the full stack engineer like to the way I look at it is if you know, yes, you are embedded in a customer, you're embedded with your sales team. I even do a bit with our marketing team to make sure that everyone is aligned. But when I talk to customers or partners, the conversation in a 15 minute sync can go across the UI, it can go across the backend, the database architecture, architecture, deployment, security, encryption. And then also I get education questions, right. Like how does this work? What does agent mean? Things like that. So I think it's more full stack engineer with like a sales focus and embedded focus rather than like a solutions or sales engineer. And it's very much even less so a consultant. So I like to think of myself like I'm still a full stack engineer with a focus on like you know, helping our customers and sitting with them and hearing, hearing what they want to do. And yeah, there's other, other aspects of that, but yeah, yeah.

Brian Gracely: And that, yeah. So I think the, the part of it that I think resonates with what we've heard a lot is the idea is it is very much some, some variant of full stack engineer. Right. Like you're, you're going to be living in the stack, you're going to be close to it, you're figuring out like, you know, what can you reuse, what has to be new and then, and then the other part of it, like you said, is wearer of many hats because, you know, the people that you're working with, you know, even though we've been talking about AI and doing stuff with AI for, you know, three and a half years or however long it's been, that's still a very short learning cycle for, you know, for a lot of people. And so it's everything from like, what does this do, how can you make it work? How can we educate internal teams, how do we, how do we sell teams on the concept of what we're trying to do? So that makes total sense of, you know, kind of the way you explain it.

Jordan Cab: Yeah. And I got to remind myself a lot of time that yeah, we live in a bubble. Right. Like my team, Brian, very much as well. Like, we're obsessed with this stuff. We're always on. We want to learn what the bleeding edges, we want to push these tools as far as they can go. But you know, I got on a call with a, with uh, a partner. I think there were 15 or 16 stakeholders on the call. And they ranged from, you know, the technical decision makers all the way down to people who would be using this tool every day. And it's not a technical product. So I had to remind myself like, I live in this bubble. I live and breathe this stuff every day. A lot of the people that I talk to, their exposure to AI might be, you know, ChatGPT or Claude here and there to like format documents or ask a quick one off question. So yeah, I think part of the role is backing up and like taking, taking the temperature, your stakeholders and be like, do these people need education or do they need implementation?

Brian Gracely: Yeah, absolutely. Uh, yeah, no, I couldn't agree more. I think, uh, it's something we've talked about a lot on this show is you, you can, you can, you can gain a lot of, of skills very quickly if you sort of completely immerse yourself in it. But, but the, you know, the gap between the, the person that's really immersed in it, you know, they're living with it every day, they're really curious and the ones who, you know, just are, you know, they're, they're sort of around the fringe of it or it's not their, their day to day, that even the gap between those can become really big. And then, you know, the amount of new technology, the amount of new things that are happening across, you know, lots of tools is so broad it feels like it's nearly impossible to keep up. So, yeah, I can, I can imagine from day to day, the people you're talking to are, you know, all over the spectrum in terms of, you know, what they want, what they need, what they think might be possible.

Jordan Cab: Yeah. And one of the interesting things on that note is I have recently talked to a lot of, like, local technology leaders within city governments, town governments, like Morrisville, Cary, Raleigh, Wake county, and I'm just curious to learn, like, what, what are people doing? Like, how, how's the everyday, you know, state government employees, state agency employees, city agency employee, like waterworks or public utilities, how are they using it? Right, right. And it's one of the things that I've learned is, especially in government, and this is probably true in some large organ private organizations, but there's this, like, there's a tension between being actually allowed to use something like Claude and ChatGPT and having the opportunity to learn about it because, you know, it's almost like those two things run orthogonal. It's like I can't put my, you know, Excel sheet from work into Claude because it's banned at work, so I can't learn how to use it or what it's capable of. To then go ask, you know, my boss or my boss's boss, like, hey, can we get a license for this? So on the public sector world, on the public sector side, it's very, uh, it's, it's kind of stressed in two opposite directions.

Brian Gracely: Yeah, yeah, it's, you know, it reminds me back to, you know, when we, we were doing this and we were mostly talking about cloud computing in the early days, it was, it was the same sort of thing, right. There was this, this bubble and this perception of like what Silicon Valley looked like. And then, you know, kind of the further you got away from Silicon Valley, you know, you went through these things that you're like, oh, I can't believe somebody doesn't have access to the tool or they haven't been using it, or it's not a part of their strat. But yeah, what you're highlighting is very much a reality. Let's stick around in that space. So as you're thinking about this from like, uh, a full stack engineer perspective, like what, you know, you came into this with a set of sort of technical background and skills and you're doing stuff day to day. Which of the things that were sort of pre AI are you still using a lot of? What are some of the new things that, that you've had to do from a, you know, an FTE perspective that's, you know, wasn't. You weren't doing two years ago or something like that?

Jordan Cab: Good question. I mean I only started doing, I don't know if the question is like which technical things am I using but also like from a skill set. I was doing a lot of selling when I was running my company and I tell my boss a lot. It's nice to have customers to actually sell to because as a founder you're going out and doing cold emailing. I will say one, one place that I'm starting to get a lot sharper on is actually building out. Like we, when I ran my company, we were cowboys about implementation. Like you tell us where, you tell us when, we'll meet you there and we'll do it right. And working for somebody like Mozilla, uh, you know, the risk is way higher. You're going to do it right and you're going to do it right the first time. So I think like something that I wasn't really doing and now I'm doing a lot of is thinking through like this is new technology that we're going to deploy. There are risks inherent to like using non deterministic software just inherently and these are customers who their brands value are very valuable. How are we going to deploy it and get it right the first time? So I, as an fd, I think that's one of the things that I've, I've really been trying to sharpen like in my toolbox is, is the planning because like planning on implementation, planning on deployment. Because I am, I'm still kind of like a cowboy when it comes to that stuff. I'm like, oh, we'll just set you up plc, let's build you a sandbox environment. Here's a login. Go for it. Like try and break it.

Brian Gracely: Yeah.

Jordan Cab: As far as technologies.

Brian Gracely: Yep.

Jordan Cab: I, I would say my day to day is still the same. Like it's just augmented and I think that's where I'm comfortable right now. Like I, I don't like passing a task off to clogged code or something and like letting it run for four hours, coming back and like you know, merging a PR or asking for review, it's still like a very senior developer who sits next to me. And that's uh, really the primary way My workflow has changed. Aside from like speeding up research, which is really nice.

Brian Gracely: Okay, okay. You're not, they didn't, they didn't drop you into like an office space that's just nothing but vibe coders that, you know, I've never done this stuff before

Jordan Cab: and no, I, I remember when we launched. So we open sourced Thunderbolt in April or May and we were on the front page of Hyper News for like a couple hours and I remember one of the top people have like a love hate relationship with Mozilla and one of the top comments was like, nobody asked for this five coded piece of bullshit or whatever. And I was like, oh man. Like, go look at the, go look at the commits. I promise it's not that. Yeah, but.

Brian Gracely: Yeah, yeah, well, and that's, you know, that, that's just sort of a byproduct of not so much Mozilla as just the industry right now is, is people aren't ex. You know, what stuff has, has gone through all that sort of normal engineering rigor and thought process and what was, you know, you know, put together in two hours because somebody saw a thread on Reddit and they were like, oh, I could do that. I could just put that together really quickly. So yeah, I, it's, it is, it is interesting though that, you know, the thing that you're evolving around is really like structure and thinking about it. And I don't mean that in like, oh, okay, you know, you went from startup, um, to other stuff and so things get slow. I mean, the reality with AI is like, you know, you're dealing with people's data, so there's, there's gotta be a lot of thought about like, where does it go? Should it go in this model? You know, what are we combining it with? You know, GPU usage and tokens aren't cheap. So, you know, if you're not getting stuff right, like, you can experiment a lot but you know, at some point, you know, you're going to experiment your way out of the number of tokens you have. So, yeah, it makes sense that that evolution is sort of naturally happening as you're working on bigger things.

Jordan Cab: Yeah. And uh, our partner, Deepset, based in au, who is, they're kind of a channel partner, distribution partner for us, we're working really closely with them to be the front end for their pipelines. A lot of these customers in the EU are under a tremendous amount of scrutiny compared to the customers in the U.S. right. Oh yeah, there's the AI app, uh, there's extensive data Security, data governance rules, which makes Thunderbolt a natural fit given that we're pretty sovereign.

Brian Gracely: Yeah, no, yeah, sovereignty, especially when you get in the EU and to a certain extent in apac, it's, it's the first conversation as opposed to being like, oh yeah, you know, kind of put it at the end and so forth. So talk to me about, about Thunderbolt and talk to me about sort of the work you're doing with harnesses. So kind of give me a sense of like the scope of what problem it's trying to solve and sort of how it works and you know, how people are starting to think about implementing it.

Jordan Cab: Yep. Yeah. So Thunderbolt runs on any device where you can currently use AI. Uh, so if you're browser, desktop, phone, iOS, Android, Linux, anywhere you want to use it, you can do that. It's also local first. So what we're about to release right now is a way to use it completely without a backend. So it's completely standalone and it will sync to all of your devices. So if you were to sync the interface, you'd be like, oh, this is another, you know, Open Web UI or this is another Libre Chat or chatgpt style interface. And from the face it is, you have your niceties, you know, add your own MCP servers, add uh, your own models, whether they're local or remote, bring your own key, all of that. But what we're really trying to do is, you know, originally we thought that this was going to be a consumer application and we have some popularity, like people who have the acumen to, you know, clone a repo, build it locally, use it. But what we're finding is this like modular interoperability that was the part of the original thesis, is very attractive to enterprise customers. Reason being they have, you know, their own agents that they've built and they want to be able to connect them using ACP or they built their own MCP servers. They want to just drop them in, drop them in and treat Thunderbolt as nothing more than an interface. I m think that there's still some ambiguity. People will disagree with me on this. This might be a hot take. There's a little bit of ambiguity right now between the term harness and like what is otherwise a front end. And there's, there's some blur there. I would call Thunderbolt more of a harness just because it's, it's interoperable, but specifically for agent, like agentic work. Not. Yeah, not just a pretty front end with some like opaque back end. So that's, that's the core of the product. We're also partnered with a company called Tinfoil, which provides confidential compute, which they're also a partner of Red Hat, which is cool.

Brian Gracely: Okay.

Jordan Cab: So in a future version we're going to be able to offer confidential compute. So we have no idea right now what you're sending if you use the Thunderbolt backend, and soon we'll be cryptographically attestable that we truly have no. No clue.

Brian Gracely: Yeah. Now when you talk about things being local and again you sort of said like it can, it can run on anything. Is it, you know, is it assuming that like you're, you're running something like Ollama and you've grabbed some model, you know, you could have grabbed Quinn or something, you know, something that's going to fit on, on any given machine or you know, does local mean something different? Because you're, you're going after like sort of team based use cases and collaborative stuff. Does that make sense? Like, you know, does it mean, is it, is it something that like I'm going to run on my laptop or is it something that like the IT group's going to set up? Because, you know, this, this group is going to run 30 agents that are going to do a bunch of autonomous tasks or is it a combination of both?

Jordan Cab: Combination of both. So we have people, internal Mozilla who have, you know, they're definitely like Tinkerers and have home labs and run Channel 4 or CLEM, you know, at home and use Thunderbolt and just connect to a local endpoint. And then with some customers, some partners we're exploring like they're going to stand up a very, very serious like hardware infrastructure and something scalable and run a very serious open source model or even a private model that they've trained. So both ends of the spectrum. And then obviously if you want to drop in the, you know, your Anthropic API key and connect to Opus 4.8, the Frontier model at, at Anthropic's API endpoint and go for it.

Brian Gracely: So really anything nice. So as, as you're, you know, as you're engaged with, with, with people, you're doing sort of FTE tasks. What is, you know, why are they, why are, you know, what, what are the problems they're sort of solving where, you know, this implementation of Harness, you know, makes sense. Like what, what are they, what are they trying to, to guard where you're all around. What are they trying to, to build personalities around? Like what are, what are some of the most common types of use cases you're seeing sort of in the early implementation days.

Jordan Cab: Yeah, so it is early. But I would say that there's this transition right now. There's almost a gradient where, okay, people, you know, your median information worker. If I'll say that at like an organization, large organization, they're now comfortable with the idea of interacting with a, an LLM. They try chat, GPT or Claude. It's cool. It can learn a lot, it can do a lot. They're comfortable with that. And now they're starting to transition into. Yeah, but I want it to do something right. I don't want it to like, tell me how to, you know, create this Excel formula. I wanted to open Excel and I wanted to make these changes and email me the, you know, finished PowerPoint or this finished Excel sheet. That's where we're going. So I would say as an enterprise, if you are, or an organization, you know, a large private organization, if you're looking to transition away from like informational Q and A and off to like execution, like agentic execution, having agents work for you and do things for you, I think it's a great time to start considering a harness and how you can bring a hardest in, you know, into the fold, into the stack.

Brian Gracely: Yeah. And how do you, you know, if you're putting your, you know, maybe not your sales hat on, but you're sort of explaining hat on, like, how are you explaining it to people? Like, what's the, what are the conversations tend to be where people are like, oh, okay, cool, that's, that's where I should start. You know, like you said, like, they're starting to do execution tasks, but, you know, like, what are the questions they're asking sort of before that and then once they get going, like, what are they, you know, what's the learning curve tend to look like? You know, as far as like, okay, cool, how do I take my day job and like put it into this thing? Or how do I start thinking about, you know, processes that are like, identic first, where I'm like, oh, I don't have to think about four people being in the loop. I can just kind of move through stuff like, what is, what do those conversations tend to look like and, and where do people get hung up and where, you know, what are the light bulb moments?

Jordan Cab: So I think, as is classic with anything in tech, we are, we're famously and traditionally bad at naming things. Um, it's like, I think the second hard problem in computer science, naming things. We have a bit of A branding crisis, I think, across AI agents, harnesses, models. Like, it's. Unless you were here originally, it gets confusing. Okay, so that, that is something that I like to take a step back and help people understand. Like, shelve those terms. Shelve those, like, those concepts for a second, and I can help you figure out if you need to get there or not. A good question. A good place to start is like, what do you wish these traditional tools did, right? Like, do you just use chat GPT or something like that? Or, or, or whatever model or whatever tool just to, like, ask questions, get an answer, and go about your day? Like your professional day? Then maybe like, agentic, like an agentic harness isn't the way to go. Maybe you. You might, Your needs might be met by just trading LLMs as like, a Q and A system. Or are you. Is there a task that you do every day that you could explain to somebody like me who has no idea how to do it? Like, I'm not in the insurance industry, but you could tell me how to, you know, fill out a vendor on onboarding, you know, worksheet or form? Oh, yeah, Yeah. I spent 12 hours of my week doing that. Okay, well, then maybe it's time we start experimenting with, like, building out a simple agent on your behalf. And that's where deepset has been tremendously valuable because they build the agents and then thunderbolts the front end. Okay, but those are how those conversations go. It's like, if you have a piece of work or a part of your day that is monotonous, repeatable, you know, all the. Of the inputs, then it's time to say, okay, let's try building an agent. And then, um, we need a harness to connect it to, so why not thunderbolt?

Brian Gracely: Yeah. Okay. Yeah. And that's kind of really what I'm, what I'm, What I was curious about, because again, you know, I think you're. You're very right in that there's a lot of things that, you know. I mean. Well, in general, technologists are bad at naming things, but, you know, there's a lot of things that get lumped into stuff. So, you know, you hear the term agent, you hear the term harness, and, you know, in my mind, I'm always like, okay, what does that, what does that mean to you? What does your thing do? Because, you know, no two may not necessarily be the same. They might be doing different stuff. And so, yeah, I, that's. I was, I was very curious as to, you know, were you coming around and just Getting the point of saying, like, you know, what is your role? What does your task do? How would you map that out? Does it make sense to now start to use this second technology as opposed to just what you could do in a chatbot? Okay. It becomes like, use a chatbot. Then maybe there's another tool that you use and you're like, oh, that tool connects to these four things that right now you're just copying, pasting stuff to. Or every time you get to this stage, you send an email off. That could just happen automatically, but, you know, you may not, you know, for things like that, you might be like, well, you know, I don't need a full blown harness or I don't need all. But yeah, so it was, I like the way you explained that. It was, you know, let's, let's forget about the name so much as, let's get into what you're trying to do, how do you do it today? And then let's start thinking about, okay, where could technology start to augment what you're trying to do?

Jordan Cab: And yeah, and we're still so obviously early. Like it's, it's indicative by the naming of things. We're so early because, you know, like Sarah at Progressive, who's the senior claims manager, is not gonna go click on Add MCP server, type in a websocket endpoint or an endpoint M. Right. And like, she's just not, not because she couldn't figure out how to do it, but it's that like, come on, you know, like, we all know that that's not how it's supposed to be. Nobody's cracked the code yet of how to like obfuscate and abstract truly, all of this away. I know, like, the Frontier Labs are getting close. Like they have better branding, like Claude Cowork Claude Design, where under the hood, it's taking advantage of these things. And that's one day where I'd like to see Thunderbolt go. Yeah. But yeah, the names are a symptom of. We haven't figured all of this out yet.

Brian Gracely: Well, I think it keeps coming back to this thing that I keep highlighting whenever I hear that some company is like, oh, we decided to lay off people because of AI and I, I keep coming back to being like, I. I'm betting in most cases, most of the people in your company still have no idea really how to use it because to a certain extent, to be a pretty good, powerful user, you needed a little bit of programing skills, a little bit of, you know, some techie you needed some sort of techie stuff like, you know, what an IP address looks like, or, you know, what, you know, an FQDN looks like, or, you know, what an API looks like. And I'm like, you could, you could pair together your, some of your, your techies that you might be like, well, they're underpowered as a software programmer, but if I could, you know, if I could pair them up with, you know, Sarah, who works in claim, the claims department, like you might create a superhero out of that. Right? Like we view, if you take the person who knows the business domain, you marry them with somebody who has got enough technical stuff to be like, okay, cool, I analyze what you did. Let me set that up for you, or build you something. And maybe that is know, maybe that is exactly what, what FDE is doing. But it does feel like this marriage of, you know, two domains that, you know, maybe, maybe in the past thought of themselves as different, but now, you know, the, the tools and the technology are sort of making them interchange, you know, intertwined in, in, in pretty powerful ways.

Jordan Cab: Yeah, no, I think, I think you're. I think you're exactly right. And I think, you know, right now the commercial value of an FDE makes sense because there's like, you know, there's a lot of value. There's a lot of, there's a lot of sales to be made to, you know, to be candid. And you want to put your engineers in front of the customer with the customer teaching them how to use it. But I do hope that there becomes an equal pattern, like, uh, you know, the, the internal version, like what you were talking about, the internal fde. I don't, I don't know a good name. Internal deployed engineer or. Yeah, I mean, Mozilla is a tech. Tech company, but I can see it, you know, again, progressive, to use the example. Have some of your engineers become internally deployed. Go work with the claims department, go sit with them, go learn what their job is. Because there's probably something there. There's probably something to teach them, and I'm sure you'll learn something from them as well. Right?

Brian Gracely: Right. Okay, well, cool. I'm gonna wrap up. Cause I know you're busy and you've got a lot going on. If you were to give kind of folks guidance whether they're going like, oh, FDE sounds like kind of cool. I've been doing some full stack stuff, or I've been doing some platform engineering stuff for a while, but I want to get more, you know, you've now, been through this for a little while. You've gone through some career changes. Like, you know, what would you, what would you suggest to those folks as far as being like, uh, hey, this is, you know, this is kind of a great way to extend your skills. Or this is some stuff you should be looking at to be, you know, be really effective at this type of role.

Jordan Cab: Just general advice, don't start a company. That's a good place to begin. You'll learn a lot. But don't do it unless you, uh, Unless you really hate yourself. No. So there's this balancing act to do, I think right now in the industry. If you're in any technical role and you want to get in front of the customer, do not get married. So there's two pieces of it. Do not get m. No, no, I. That was a bad. I paused at the wrong point. Do not get married to any technology, is what I was going to say. Or any standard or any protocol. I made that mistake when I was building the first version of our platform Limelight. I was like, oh, yeah, Rag. This is it. There's no possible way we can improve on Rag. Like, this is. It'll be the special sauce on top, but, like, rag is it. There's so many better ways to do things like that. I mean, now we have knowledge bases. Memory is becoming the hot new thing. So don't get stuck on one thing because it works kind of well. Constantly be learning about what else is out there. But again, don't spend too much time, like, spinning your wheels on trying to learn every new thing. So to sum it up in one, I would say, and my boss, Chris Emozzo, is, has an uncanny ability to do this. The better you get at being able to discern something that's going to stick versus something that's, like, kind of cool and interesting and is going to be, like, front page for a couple of days. The more valuable you're going to be. Like, Chris picked out agent client protocol from, like, early on and was like, this is where we need to go. And it's paid off in spades for implementing that in Thunderbolt. So if you can sharpen that sword, you're going to. You're going to do very, very well.

Brian Gracely: Yeah, no, I, I would, I would agree. We've always said, like, you know, to, to succeed in this industry, you've got to have a natural, uh, level of curiosity. You know, you're just kind of like, you're constantly sort of looking around the corner like, what's, what's Potential, what's next? And then. And then. Yeah, I think you're exactly right. It's you, you have to have the discipline to say, you know, when new things come m along, I'm going to go evaluate them. But, you know, unless, like you said, you, you just absolutely know for a fact that it's going to be the thing, you have to be able to sort of come off that belief or that like, oh, I love this thing, because so many new things have come along and you have to be flexible enough to be like, okay, cool, I'm fine leaving that thing behind and taking on that new thing. So I think that's great advice. And especially in a super fast moving pace like we are right now. I mean, I think back to stuff we were talking about a year ago and two years ago, and we're like, oh, yeah, ah, no one does that anymore. That's, you know, new stuff has come along so well. Very cool stuff. Well, uh, Jordan, you know, thank you for coming on. I think, you know, it's, it's really good for us to, to just get a chance to talk to people who are, you know, living this stuff that are, you know, kind of in the trenches. Exciting stuff, especially for people that are, you know, starting to look at sovereignty. Uh, if sovereignty is either something that you're interested in or, you know, is sort of being pushed on you because of laws and regulations, it's cool that, you know, the things that you're building with Thunderbolt are emerging and so forth. So we'll definitely have to have you back at some point to, you know, if willing, you know, talk about some of the, uh, challenges of building a company. Or maybe we'll just do that over a local beer or something like that. But, well, listen, thanks for being on the show. Uh, we really appreciate it, folks. We appreciate Jordan's time today and giving us some very, very valuable insights. Uh, hope you enjoyed it. Thank you all for listening. Thanks for telling a friend about the show, helping us grow the community. And with that, we'll wrap it up and we'll talk to you next week.

Jordan Cab: Thanks for listening. Check us out@theenterpriseaishow.com for past shows, newsletters, and all things enterprise AI.