Artificial Intelligence In Recruiting: The Talent Engineering Playbook | Kyler Frisbee

Unlocked Professional: AI and Future of Work: Here for a treat today. This episode aligns so well to what we typically discuss on the show. You take your domain knowledge from the work that you do, sprinkle in some AI, and outcomes magic. My guest today, Kyle Frisbee, built his automated sourcing tool to reduce the mundane process of sourcing engineering talent.

And that's a huge part of what we dig in today, including his take on how broken resumes and traditional ATS platforms have become. He felt there was a huge opportunity to build a system like this. So he went heads down for months, spent way too much in tokens, but integrated his own domain knowledge in recruiting and learned a ton along the way. Building something that in the past would have been limited by needing engineers to build and iterate on it.

Since we recorded this conversation, Kyler's gone and launched a new venture called Seed Glass, alongside his co-founder Laszlo Bach, the former head of people at Google. Congratulations to both of them on that launch. Having spent 17 plus years in the industry myself, I can tell you this kind of innovation is a genuine game changer. It's exciting to see experience like his finally translated into something like this without the old limitations of needing an engineering team to bring it to life.

I'm really excited about where the future of this space is headed and where Kyler and Laszlo take it. I know you're gonna enjoy this episode. Don't forget to hit subscribe and tune in every Wednesday as we drop new episodes. There's a lot more of this kind of content coming.

This is a great space to be plugged into if you care about the future of hiring and talent acquisition. Enjoy. We did a head-to-head comparison with Frontier Lab. They had a new role that no one in their company had staffed for and didn't have time to ramp up on this new discipline.

They gave it to us and another agency at the exact same time as a competitor. And this is one of the advanced agencies in the industry. Within Three days we had received fifteen resumes from some of the top professionals in the field who had flagged that they'd be interested in interviewing and put themselves in for the role through us. It took three weeks for the competing agency to find one engineer who would eventually move on to that that field or or that process.

And and all four of the folks they picked from the first fifteen moved on to that final round. So it is light years ahead of what we would see as competitors in the market and light years ahead of what I could do because I As a person, as an individual, am absolutely in that same seat likely with other agency. Maybe I'm a little faster somehow, but not by much and not by this magnitude by any means. I don't necessarily know how we solve the ATS issue.

Applications are just literally falling into the void, and there's so many fake candidates that it is extremely hard to surface a relevant candidate from an internal ATS. Like You're looking at twelve fake profiles, every real person, let alone them being a real fit. It's extremely difficult to break through the noise. So I don't know how that's solved.

But what I would say is if you are even remotely open to new jobs, do not ignore the recruiters that are reaching out to you. They could be your most legitimate opportunity to getting a new role because ATSs just aren't getting you the calls that they used to even three years ago. And if a recruiter reaches out to you about a role, the chances of you getting a higher salary than when you apply directly are are real. Not only is that person bringing you a job that could be a fit because they thought you were great, but they're probably gonna be able to negotiate a higher salary for you than if you applied on your own to the exact same company.

Tyler Frisbee was a semi pro snowboarder before he ever built a recruiting career. And somehow that's the least surprising thing about his resume. He spent time inside Google and Unity and helped Unity's machine learning team grow from one person to more than 30, and then went on to build the hiring engine behind Rivian's run to the largest IPO since Facebook. In twenty fourteen, years before anyone called it AI recruiting, he was already running IBM Watson in production to find candidates.

Now he runs Optimal Tech Partners out of Whitefish, Montana, where he's built his own recruitment agents from scratch, taught himself to do it through vibe coding, and so far made a deliberate choice to keep the whole system proprietary instead of selling it as a software. He calls what he does talent engineering, and he'll tell you the sourcing side has gotten radically smarter, while the actual close, the part where you earn someone's trust, has stayed completely human. Kyler, I'm passionate about this emerging talent engineering space you're in, and I know this conversation will provide a ton of value to the audience, especially those in talent acquisition and recruiting.

Welcome to the show. Hey, thanks, Jeff, man. Excited to be here. So before Optimal Tech Partners, before helping building Rivian's engineering teams, before growing Google's AI and ARVR hiring orgs, where did you grow up and what was the early version of Kyler like?

I was somewhat of a little boardvum hoodlum. Grew up on the mountain in Whitefish and then like driving boats in the summer, skateboarding, but was just a massive skier snowboarder growing up and so followed that passion throughout like high school and into college. Picked my college because it was the closest to a mountain and really happy that I did. But there were definitely some some days where my family and my friends were a little worried whether I'm gonna make anything of myself, whether we were gonna end up in jail.

But I always loved like the business end. My dad was a stockbroker as I grew up and so always wanted a briefcase, always wanted a suit. Definitely got to do my banker stuff before I got into tech. But you know, it was it was a long journey, and I think some people were a little nervous for me, but definitely it worked out in the end.

I went to Montana once. We actually had to drive through it to go to Canada because I didn't get my daughter's passport in time. I'm curious, is it more of that remote? Part of Montana or is it a little more populated?

Pretty populated. We're still pretty small. We're in the Flathead Valley, just south of the Canadian border. We share the the Canadian Rockies and the the US Rockies with Canada.

But yeah, we've got a lot of celebrities that live here. Like Emilio Estevez has a house. Johnny Depp used to drive his yellow Lamborghini around town. Drew Bledsoe has a place out here.

John Lithgow. Who's gonna be Vol or he's gonna be Dumbledore and the new Harry Potter remake. A lot of really compelling folks live out here. Who is the ⁓ coach Phil Jackson of the Chicago Bulls?

He's got a place out here and used to bring all the players out every summer. So we're among some really amazing people. And the valley has grown a tremendous amount since COVID, but still has that Montana feel to it. I think everybody wants a little slice of Montana after after the tombstone exper or sorry, the Yellowstone.

Yellowstone, yeah, Yellowstone experience. Yeah, exactly. And it has ⁓ it's been really fun to see how much of ⁓ popularity it it Montana's grown into. So you mentioned on our call that you're a former semi pro snowboarder with skateboard art all over your walls, which is what we have in common.

Are you still riding these days? ⁓ yeah, so I actually just tore my ACL at the end of this last season. I'm about nine weeks out of surgery. and still ride as much as I can.

The winter wasn't the best out here in Montana and just had a couple a new new babies, so we've got two little daughters that are under two. And so we're pretty busy. But yeah, I I started a C B D company with like this legendary skier Tanner Hall. ⁓ it was called Lion's Gold and that was just before I started at Rivian.

And so I got to do some semi pro skiing again to promote the brand. So I'm still pretty active on the skis. I love it. More of a social aspect for me, but I can get a little gnarly every once in while.

So, Kyler, you built a recruiting system and you said that your system finds candidates that tools like Clay and Apollo miss since everyone's using the same standard stack and surfacing the same names in the sourcing world. For an engineer on the top side of that, what is that actually getting noticed by a system like yours look like? And what is it picking up on that a keyword search wouldn't? So a long story short, how does it work at a high level?

Yeah. We browse the entire web, very similar to how we used to do it manually, but we take in data from everywhere, including published papers, private and public repositories and databases, GitHub, social media from Twitter, Stack Overflow and LinkedIn obviously. And I feel like if if an engineer's trying to get noticed by our system, they've probably already done the work that they needed to do to get there. But if somebody's hoping to be noticed in the future, those publications And building in public is probably the biggest thing.

Having something that is clean, legible, understandable, and works in your GitHub makes a big difference. And then yeah, just being involved in anything you can throughout the community as far as like the discipline you're in, but all of that leans to a signal that we pull from. So essentially having some type of footprint digitally out there, promoting your work, pushing it out via social platforms is a good idea. I think Not only just in tech, but in general, to have some kind of social presence is key these days in terms of job search.

You ran the first Watson-based recruiting platform back in twenty fourteen, more than a decade before AI recruiting had a name. What did that early attempt get wrong before agents were in play? Yeah, so IBM Watson was phenomenal technology at the time. I believe it was like winning chess matches and games of go.

But It was used in natural language processing to match up resumes and job descriptions to really determine culture match and synchronicities between the words and synonyms and so forth, but it wasn't trained extremely deep on technical abilities, aptitudes, or context. And so what you would end up finding is the signal didn't always work. If you were generating a difference between a back end engineer and a front end engineer, a lot of the times it worked and could maybe filter out those those folks.

But if you were looking for somebody who had specific expertise or a specific experience in building early stage or enterprise, it had a lot harder of a challenge doing that. So the new tools are just substantially different and better. Did it take it to the next level past what a typical keyword search might be in the past? Yeah, no, it definitely took it past that.

It would analyze the style of writing, it would pull real context into what somebody was doing. But I think the training models for IBM Watson were probably minuscule compared to the data sets that are training the models today. So the ability for it to really understand what somebody was doing or what they might be able to do was not very deep and so you'd more or less get a match of like culture fit. And that's based on writing style.

So it's like, okay, this person's resume is written in a similar fashion with similar wording and ⁓ stylings and content as the job description. So you're really like matching up the person who wrote the job description with the person who wrote the resume. And often we we know that those aren't the people who are hiring and a lot of professionals don't actually write their resumes. So it's it just fundamentally flawed in a lot of different ways, but it was an advanced solution.

But essentially I let that go and and ended up sourcing manually for 10 years after that. Sourcing tools aside, we'll talk about recruiting as it is today and how you have resumes being created by AI, job descriptions being created by AI. So there's there's a gap there in between what's actually human out there and what's just matching up these separate keywords. What does a system like yours do that sets itself apart from just making that match between what could be two AI agents?

So the newest systems, especially the one that we built, is extremely sophisticated and understanding the full depth of context. Not only can it understand what your company is doing, what this role would be doing, where the person would most likely come from, whether it's target companies or position that they've held, as well as like how well they might fit in the new role. and in the new trajectory, how do they have experience as a founding engineer? Have they increased velocity as they ramped up to an IPO or through later stages of funding?

But beyond that, being able to really understand context of tools, how those have been used, where there might be gaps in seniority or in in application, but the kind of understanding that it can pull from context alone, from somebody's five jobs, what they've done, their career arc is pretty remarkable. And then when you mix that in with any types of additional content, whether it's from those proprietary databases or from GitHub or from whatever else it might be, the signal becomes pr ⁓ absolutely incredible.

And then the systems are not only able to ingest data around the company and the role as well as the candidate, but being able to match that up and surface those candidates and then To generate the outreach in a way that is compelling. The the number of compliments we get from candidates themselves, just acknowledging the amount of research that we've put in to to really knowing what they're doing and where they're coming from is pretty remarkable. We're talking like senior staff engineers at LinkedIn itself who have given us compliments saying, like, you've hit a niche or a spot that almost no recruiter has been able to hit, and reference how Usually it's AI jargon based off LinkedIn profiles.

And this particular person, we found their GitHub and really understood their personal projects and how it applied to the industry and and ended up finding somebody who was one of the most advanced in their field. So the last example I'd probably give on that is we did a head-to-head comparison with Frontier Lab. They had a new role that no one in their company had staffed for and didn't have time to ramp up on this new discipline. They gave it to us and another agency at the exact same time as a competitor.

And this is one of the advanced agencies in the industry. And within f three days we had received fifteen resumes from some of the top professionals in the field who had flagged that they'd be interested in interviewing and put themselves in for the role through us. And it took three weeks for the competing agency to find one engineer ⁓ who would eventually move on to that that field or or that process. And and all four of the folks they picked from the first fifteen moved on to that final round.

So it is light years ahead of what we would see as competitors in the market and light years ahead of what I could do. Because I as a person, as an individual, am absolutely in that same seat likely with that other other agency. Maybe I'm a little faster somehow, but not by much and not by this magnitude by any means. So for some context here, most of the positions that your agency recruits on and the tools or the and the tool supports are more tech roles.

Correct? Absolutely. Okay. And for the audience, it's a an important distinction that the types of roles require different types of searches, different types of industry knowledge, different types of domain knowledge by the respective recruiter and or recruiting tools.

And so I'm finding this this trend based upon conversations that I'm having tech Or even outside of tech, that traditionally the way things are set up is that you're doing these keyword matches through the ATS and the resumes. And I'm curious from your perspective, let's say specifically in tech, do you think resumes are ultimately unnecessary to some degree in many of these positions now? Like your systems and the tools that are evolving to do the sourcing of talent, they don't even need a resume.

They're actually finding again that social footprint and other signals that that align better than that resume would. I think at the highest level for executives, there's that idiom that a resume is outdated by the time you get it. So for some of these most high-level folks, whether they're chief officers or board members or even maybe some of the senior vice presidents, that's probably true because their role is so dynamic and it changes what they have to do. I think in the tech industry the resume As much as it is hard to differentiate yourself in an ATS at the moment, is still just probably one of the most important artifacts for an engineer.

We work with recruiters and product managers, marketing managers, but our bread and butter is that software engineer and the reason software engineering sourcing has always been such a different beast is because you could get an idea of what somebody does based on the technologies that they list or some of the projects that they've put in. whether it's on their LinkedIn or in their resume. So I think twofold on the ATS side, when you're applying for jobs, your resume has to be not just polished, not just keyworded, not just context relevant.

It has to really explain what you built, what were results that you attained, and actually key I guess benchmark those those key metrics. for a recruiter to see or a hiring manager to really understand what you've delivered. And especially in this AI age, it's even harder. So if you can say that I used a genetic tooling to build this thing for my discipline or for my job, your resume speaks volumes higher than the traditional or fraud fake resume that's keyworded perfectly and from Apple and Netflix and so forth.

And just I've worked with clients and they've given me access to their ATSs and the inflow of fraudulent Credentials is at an all time high. On the headhunting side, which is really where we focus, that resume is more of a validation for us. So all of our key marks come from everything beyond the resume. So you I think you're right in the headhunting sense, we aren't necessarily using the resume to find people to identify who could be a good fit or why they'd be a good fit.

But as soon as we do and we open up that conversation, the resume is still that validation point that we use to sell the candidate to clients, as well as maybe internally benchmark against potential future roles and leveling and so forth. So I do think that it has a tremendous amount of value. And I think even beyond getting the initial job, when you go up for a promotion, the first thing that those hiring teams or that HR team is going to pull is your resume. They're gonna look at what has this person done, what's their track record, are they qualified for this next step in their career here?

And so I I still think that the resume is extremely valuable, even if not the most valuable. Yeah, absolutely. And I think tech differs from a wide range of other skill sets that are out there, as we discussed. I agree with you with what you're saying.

The resume, it's more of yes, it shows your track rec record and history, but I think traditionally what it's been used for in the past is more of your marketing. So it shows outwardly when you send that resume in to the recruiters or to the hiring managers, this is exactly who you are. But I think the important distinction here is again having that social footprint and having some type of presence outside that is really again, if you're especially if you're in like tech and you're looking for some of these premium, high level roles that are hard to get into and you're looking for somebody such as Kyler to find you, then you're gonna have to have some type of social optimizer or social presence out there.

And not only just the presence, but I think it's important that like an another important component that you mentioned is like the amount of fakery that that's going on and fraud. And especially in tech and have people from overseas that are trying to take jobs remotely and present themselves as somebody who they aren't. And so that as that issue like pro proliferates, I think if you ha if you can present yourself with a strong social presence, you can present the work that you've done, you can validate that.

Your resume is still important, but it's not the only thing that matters. And I kind of see the world as it's evolving that maybe we get into a position where the resume becomes less and less important. I don't know exactly what that world looks like just yet because that's such a tr traditional process that we've followed. But I've I brought on some guests recently and talked to them that are more on like the on the side of the house, HR tech side of the house where they're b they have systems that are doing skill based assessments.

Differs again from like the sourcing component that you're talking about, but it but I still think it matters. It's maybe more important again to show what you can do versus just put a resume together. And I think that's the the the most important message that I'm trying to get across is that it's evolving so quickly. And I don't know.

Any thoughts on just where we're at in terms of the evolution of resumes? I think building in public is far more valuable than a resume. Right? What you've done in the past is great, but what you're doing now is way more relevant.

And for me, building this system over the last f six, five, six months has led to more opportunities than me applying to any job. So For instance, l the last company I was building was really successful for a short term. And I vibe coded a lot of it at the end of the the the the time we were building it and we revenued almost two million dollars over four years, so it wasn't a crazy amount of money. But anyways, regardless, we closed that down and so as we were closing it down I was trying to figure out what is the next kind of opportunity.

I threw a resume in to a couple of these Frontier Labs to just kind of see what would happen. Didn't think anything of it, got a couple of rejection letters, and really got excited when OpenClaw came out. And so I started building every single day, 16-hour days minimum, sometimes 18-hour days, wake up with a laptop on my lap, go to bed with it on my lap, and just consistently building the entire day. And Some days we were spending I was spending five hundred dollars in Claude API tokens.

Anyways, fast forward through a number of like these successes, especially that 15 to 1 kind of result, word kind of spread through the grapevine through one of my personal advisors and two of the largest Frontier labs have floated aqua hire opportunities across the table for me. And one of them is the one that I applied for last year. And the crazy thing is that the opportunity is almost four times the salary, equity, and so forth than what I applied for. So it's and it's levels above what I would have accepted at that point.

So I would say like building something for your discipline, getting it proven out, showing folks that you want to work for, how you've increased productivity, how you've added value to your workflow, how have you differentiated yourself from every other person. In this day and age, that's how you show that you are the one that they should hire because AI is changing everything. Every single clip on YouTube, on social media, it says it and almost getting numb to it. But the truth is if you don't know AI, you are so far behind.

And the folks that are using AI every single day, beyond just like chatting with Claude or OpenAI, you're you are pioneering a part of your industry that Very few people are. I would just say if you're building in public and you can show some type of results, that's a million times more valuable than your resume. Yeah, and I think that's a awesome story. And I think that's definitely it's a key distinction that people should consider these days in terms of both building out their career as well as just trying to align to the future because as mentioned before, things are evolving very quickly and understanding that, you know, not knowing or AI is just not gonna be an excuse anymore.

You've deliberately kept the system proprietary for now instead of selling it as a product. And it's scaled your output by a lot. Without giving away the actual build, what's the one piece of the system that every recruiting team should be building for themselves right now? That's a really good question.

I think it's unique to each team and each individual. And that's really the special part about it is like building your software for you, for your specific workflow is far more valuable than trying to chase what works for us. For us, for me, sourcing has been the biggest unlock. Having a team of fifty sourcers at Google that were passing me leads every single day was amazing.

We'd have five to six closes weekly because we'd have twenty five candidates passed over the table minimum every week that were interested and so with this system we're able to do just that, even more. Every day I wake up and I've got Anywhere between three and ten new contacts candidates who've submitted their resume to us and asked to be submitted to the role we reached out to them about. The reason I think that we keep I keep it proprietary is because I believe that this kind of sourcing engine will be table stakes over the next couple years.

And one of my good friends told me that sourcing software has really two ways to go. You either become an agency or or you become an ATS. And I think there's like very true because when you look at some of these systems that are AI sourcing, they're trying to figure out what's my next thing, especially if they're getting data from the companies you mentioned, Clay and Apollo. Clay and Apollo are phenomenal for business development, but they only have like twelve percent of personal emails for their contacts.

So it's extremely hard to use that as a benchmark to build out your recruiting agency. And that means that anybody who's using those tools to reach out to people, probably reaching out to the exact same people every single other person is using for that in that tool. So really building a sourcing agent or engine that can go out and really dig into the web the way that you would or finding the things that are important to you is important. And then we believe, I believe, that the brand is going to be the biggest differentiation.

So getting to lead the way early keeping that brand name showing that we were ahead of the curve a little bit and continuing to provide world class service on the recruiting side along with that speed and precision on the sourcing side for our clients is the differentiation that our brand is gonna stand on. And and so five, ten years from now, my goal is to have hands dozens and dozens of clients and hopefully a lot of them were with us along this run. But even when it's table stakes and everybody has the fastest sourcing engine on earth, that Quality, that care, that real understanding of what people are looking to do with an agency is critical.

So that's why I've decided to like utilize this kind of early traction to build out a brand. Do you feel like we were talking about of the frontier tech companies out there? Do you feel like that this is a system or recruiting in general is something that's on their radar? It's so big.

One of the biggest AI companies in the world right now that recently came out of stealth. has a multi multi billion dollar, we're talking half a hundred billion, it's like forty to fifty billion dollar valuation. And this is one of their biggest desires is to have a sourcing engine before they start to scale all the hundreds and thousands of roles that they need to staff up to accomplish their goal. And then you look at the Frontier Labs and they're trying to figure out the exact same thing.

They have a number of folks in there, but figuring out and finding somebody who has that exact sourcing expertise and understands how to use the proprietary technologies, databases, and so forth, ways to search the web that are developed over a career, and then codifying that into an automated system is not an easy thing to do over the course of a short period of time. And so we're seeing a lot of folks in the recruitment field and different companies dabble with it, building smaller tools and then maybe the HR team will scale out one of the tools if it's a great idea.

But a whole ground up sourcing engine is kind of the holy grail at this point because we are as a company able to do volume that we would never be able to do as a four person team. It's absolutely remarkable what's possible. We'll come we compete with the biggest in the world right now on on number of candidates per role and we're beating them at speed to market. speed to to interviews.

Yeah. I couldn't be more excited about where we are and I think those Frontier Labs, they feel it. They want to have it before everybody else because that's their competitive advantage. If their competitors have it, they're gonna hire faster.

And as soon as those folks get into one of the Frontier labs, it's not as advantageous to leave because the equity's already started vesting. What are the top three recruiting or sourcing tools? I think juicebox, noon, and then I don't even know what a third one is. I just don't use them.

But there's definitely a bunch that people like. People like Gem. I had a conversation recently with someone and we were talking about a tool such as Juice Box, which is a sourcing tool. And it has some similar characteristics of the tool that you built, which it goes out and identifies candidates and brings those candidates to the the recruiter or talent acquisition team for them to consider as a candidate for the position.

It does that talent match. And What's funny about it is that he actually runs an agency and they're doing the recruiting on behalf of the company for them. So ultimately, even though they're still a sourcing company, they're still leveraging recruitment companies to help them find their own internal talent. And so what that what that says to me is that nothing against the tool.

I'm sure the tool itself is strong and has a lot of value, but ultimately the part that you still can't replace. Is that human part? So I'm curious, like from your perspective, where do you think the gaps still really exist and where do you think the challenges will be in the future as these tools evolve? Like what parts w will still have to remain human?

Absolutely. So first off, Juice Box is starts off as a phenomenal experience. You use it for the first time and it's amazing that you can just have a conversation, load up some raw data, and you start seeing qualified candidates. What you don't realize until you're a couple months in is that A very different search will yield very similar candidates.

And so you'll see very similar candidates or the same candidates for kind of different disciplines on the back end, regardless of like what that discipline specifically is. And then you start to realize that those same folks are getting emailed by every other person that uses JuiceBox. So I think it's a phenomenal tool, but The more people that use it, the less effective it is, and it is echoed by the fact that they are using an agency to fill their roles. But that exactly hits on one of the biggest disciplines in recruiting.

It's why we our motto is the stack and the driver. The stack is the phenomenal world class sourcing engine, but the driver and the drivers are still the human side that you can't replace. Once a candidate says, Yes, I'm interested in that job. It's the recruiter's responsibility to jump on the phone and sell that person.

And that's been my favorite part of the whole job for my entire career is always imagine myself in Shark Tank and I'm pitching an investor on a new business and it's the startup realm. So I absolutely love it. Why is this interesting? Why is the problem incredible?

Why is the role incredible? And how is the equity valued and you know looking over the next two to five years? And so that initial sales call is absolutely critical. And then facilitating that conversation along the way, making somebody know that they have somebody who they can talk to, even when the HR teams in that company aren't necessarily responding that week 'cause they're moving all the pieces around to get you through to the next step.

You know, you have somebody who you can talk with, alleviate some of those concerns. And then when you get to the negotiation side, being able to facilitate the logistics of an offer, whether it's benefits, moving, higher salary, more equity, different title, whatever those caveats are that are important to the candidate. Having an advocate on your side who can go to the company, ask for those things, negotiate with the company themselves, and not ruin that relationship for your first day.

You walk in still as you know fresh as can be to that hiring manager and to that engineering manager or your boss who just negotiated the crap out of this contract with the recruiter. And so it's a tremendous amount of cover for you to alleviate that awkwardness that you might feel walking in having really negotiated hard. And I think those fundamentals will never be replaced, regardless of how fast we can source. Yeah, agreed.

It's important that the human part still remains human. And even just as AI comes and starts getting more involved in both work and just on our general workflows. There are those human components that will always need to take place. I know we just talked about it before, is like sales, having good relationships with people, understanding the personal side of where they're coming from.

Those are things that like tools such as what you built don't exactly do. And I'm confident that the future still looks like a place where people can still be people. I agree a hundred percent. I've always thought the the results of AI is going to be more human connections, freeing up humans to do more human things.

So let me throw you a curveball, Kyler, on this. And this is maybe a little bit outside of your scope, but we're talking about tech roles. But let's use an example, something like a contractor, let's say. Now, if you're a recruiter and you're going out and you're trying to identify talent for you're looking for an electrician, then you're not going to find that person's resume a lot of times on LinkedIn.

Maybe they don't have a presence there at all. They're not really on GitHub. They have no presence there. They're not on social media.

So if you had to build a sourcing tool to try to identify that type of talent. Where's the first place that you would start? I feel like this is an interview. I would actually start with the permits themselves.

That would probably be the first place to look. What companies are issuing new permits or pulling new permits in that specific field. And then I would work my way down to see who are the top companies in that space. And From there we would just be looking for people on Facebook, which is the traditional place where people put their title and their company, but maybe nothing else.

We'd also just be looking for reviews where names pop up and trying to be like cross referencing those. And now in the age of AI, all of these things can be written into an engine that would do all of this in mere moments. So the ability for us to just run out and find people and even to just map out every painting or every electrical business inside of a specific region is more than easy to do. And then that ability to source from different databases that we already use to be able to find those folks.

I don't think it would be too much of a challenge. What would be harder is quantifying their expertise or their experience prior to a chat. And a lot of these non tech roles do take that. They take a tremendous amount of legwork for a recruiter to be able to do upfront before even identifying qualified talent.

So yeah, you'd have to do a lot more interviews for sure. I like it. Traditional market mapping. A lot of times I think AI gets overblown and it's literally like what you just mentioned there, Kyler, is you're taking your expertise of a good recruiter.

You're trying to bake in the tools and put a system together that leverages those tools to think in the way that you would. And there's parts obviously like we we just mentioned that you can't augment But there's parts that maybe you could and and those tools hopefully will develop over time. But I think a lot of times people get overwhelmed when they think about building with AI and leveraging these tools. But I oftentimes it's much more simpler than that.

If you understand the domain, you understand how to do the work that you do already. You again sprinkle in a little bit of AI into it, spend some time beating your head against the wall, and hopefully you get some strong output. You know, that's exactly it. And it's so easy to over engineer in this age of AI saying I'm gonna create agents to do everything.

When I first started building with OpenClaw, it's a multi agent harness and the ability to just be like, I have a biz dev agent, I have a sourcing agent, I have a recruiting agent, I have an outreach agent, I have a writer agent, all of these things. But then you start to realize like the memory is actually the issue. How do you deliver context between those agents? ⁓ it's still database transmission for the most part.

Otherwise you're just like stuffing context windows like to the gills and it's extremely costly. So figuring out how to do this in a way that is not extremely expensive is probably the other balance, right? Because it's easy to be like, Okay, this agent is gonna have access to the internet. Now go out and find every candidate that you could.

That's probably going to be a multi thousand dollar, if not more, endeavor to get as many candidates as you need and thus it's probably not even less expensive than ⁓ manual sorcery. It might be a little faster, but is it worth it? You're still getting paid the same amount to fill the job. And if you're a recruiter internal, you're spending a lot more money than they're hoping you would spend.

Your it's on top of your salary. So yeah, I I think that was one of the first lessons I learned is how easy it is to overbuild. Now we talk about just recruiting where we're at right now. What do you foresee popping out your crystal ball?

What do you think the recruiting life cycle looks like? three to five years from now compared to how it is now and how it's been traditionally. Obviously, like if you look at the you right now we're in the introduction of a agentic age. There's been a lot of interest and time spent on the sourcing part of it.

I'm just curious, what do you think that process looks like the recruiting life cycle three to five years from now? It could go two ways. Either either every company is going to have their own internal sourcing agency or internal sourcing engine. And their recruiting team is going to be fed with that consistently and they'll have more than enough conversations and they'll move extremely quickly.

But then you're spending a lot of money to up upkeep that and it's a lot less expensive than a team of sourcers. But the other opportunity or the other thing that I see is where maybe the biggest companies or startups they outsource all of the sourcing. Where maybe they have a couple of recruiters who are extremely good at selling. Maybe even some of the biz dev guys or some founders or whatever.

I don't know exactly what the title or the role will be, but essentially they'll be sales recruiters who will handle the inflow of candidates interested in your job. Because the difference with an agency is that I have 320 individuals at this moment from the biggest companies in the world who are interested in looking at startups. Trying to see what the next step is gonna be. And these are folks from like Google DeepMind, NVIDIA, Palantir, Andural, right?

The biggest labs, the biggest enterprises, and the most exciting startups. And the ability for some of our clients to be able to tap into that, we call it the paddock, but essentially it's a talent bench. The ability for them to tap in and immediately have candidates that they can chat with day one who are in that discipline, in that field, and if they're interested in that specific area or that venture. then they're having conversations day one.

Whereas as a sourcer, even if it takes three to five days to spin up a new search and to start getting candidates in, that's still extremely fast for an internal sa sourcing agent. But the ability for you not to have to carry that weight is I think advantageous. And so we'll either see every company have an internal sourcing engine or we will see agencies become far more useful and far more utilized by a every company. So whether you're a construction company or a tech startup, the future looks like you're gonna find the right talent much faster.

Yeah, absolutely. I think so. And hopefully that reduces the friction within that process that's that exists right now with candidates trying to match and find the right roles that's for in on the customer side, trying to find the right people. There's just a lot of people out there that are frustrated with the process of job search in general, both on the customer side as well as on the the talent side.

So if you can fix that gap and more accurately match the right people to the right roles. compensate them correctly, find the right packages, find them the right home, then it equally benefits both parties immensely. Applications are just literally falling into the void and there's so many fake candidates that it is extremely hard to surface a relevant candidate from an internal ATS. Like you're looking at twelve fake profiles for every one person that could be a fit, every real person, let alone them being a real fit.

It's extremely difficult to break through the noise. So I don't know how that's solved. But what I would say is if you are even remotely open to new jobs, do not ignore the recruiters that are reaching out to you. They could be your most legitimate opportunity to getting a new role because ATS is just aren't getting you the calls that they used to even three years ago.

So I think we will see folks really appreciate the an early outreach. And if a recruiter reaches out to you about a role, the chances of you getting a higher salary than when you apply directly are real. And not only is that person bringing you a job that could be a fit because they thought you were great, but they're probably going to be able to negotiate a higher salary for you than if you applied on your own to the exact same company. So how accurately does your tool weed out fraudulent candidates?

We don't I don't think I've had a fraudulent candidate yet because we are reaching out to people who have verified track records and we're pulling people who are actual practitioners of their role. We don't open up applications ever. And so it'd be super hard for us to even find a fake candidate, I think. Yeah, so it's like that's that has never been an issue in the last five months, this whole time.

What if that's the evolution of this? What if candidates start creating fake Social presences. It's definitely possible, but it's extremely rare or it would probably be extremely rare because y we're looking for a multitude of signals and for for somebody to fake that, it would take almost a whole career to be able to fake the signals that we're pulling from. Okay, Kyler, so building these tools, can you tell us a little bit more about your experience with coding agents and what the future of those tools look like?

Yeah. I think coding agents are Probably the most exciting piece of everything that we're talking about is the ability for anyone to become a product manager and then if they are an industry expert in their field, guide those coding agents to create exactly what you want. There's a number of really amazing people out there who are software engineers by trade that are using these coding agents and will teach you the things that you need to be careful of, the pitfalls of security and how to make sure that your code base is hardened and not able to be penetrated easily.

But For me, the education portion of these tools has been more valuable than almost anything else. Over the last eighteen months, I've learned more than I did in my fifteen years of tech recruiting from how code is put together, how it works with itself, version control and all these pieces that for a software engineer are probably common knowledge, but for me I only learned like specific disciplines and very high level, like what's What are the unique components of cloud computing?

What are the unique components of machine learning and AI? And how is computer vision used? And all of those pieces I was fundamentally solid with. But how could you build a stack, an engine, or a code base to approach those problems?

Those weren't necessarily like mm on my wheelhouse or in my wheelhouse. So learning while I'm working with these tools has taught me so much. Like w I went from the very first website I posted. was on my local machine and I sent the local URL to a friend was like, hey, check this out.

And just like all the memes and the funny like software engineers make fun of, like your first thing that you made and you're trying to get somebody to look at this thing you built on your local machine. It's not possible, if you're wondering, but that was where I started. And then getting to the point where like I understand version control, rebase, the order of commits, how to merge, like I was already getting good at that by managing a software team in my last role, but the ability to now understand like the order and how to use these tools to build is something that you can't replace.

So as much as these things are doing the work for you, they're also bringing you up to speed or bringing me up to speed at a rate that I definitely could not have gotten even at a university. So are there specific people that you follow or tools that you use in order to get that information or are you just grinding it out with Claud code yourself. I'm grinding out with Claude Code and I'll use Claude to find any real specific things that I can't. But where the benefits come from I think are on Twitter.

There's like AI and CodeGen Twitter where I'm able to like just scroll for a day or fifteen minutes and even within the first like five minutes I will find something that could be applied to my stack that I didn't know about, or a new evolution, or a new release of a repo. or something that will make my stack far more advanced and that happens every day. The more I spend on Twitter, the more advanced my system gets. And that's so much different than like my c earlier role over the last four years in like digital assets and so forth where it was just a toxic environment.

Digital assets are a very like toxic environment. It's all about number go up and and aping in to like the hottest new coin for the first thirty minutes and then pulling out and dumping as fast as you can. Anyways, but in the AI side, people are really like posting things that are extremely helpful. And yeah, so like you you just go on Twitter for a minute and eventually your feed curated.

But I've been lucky that I've been doing this for like eighteen months and now my feed is extremely curated to where like I can just scroll for a couple minutes and I see a lot of stuff. But I really like this guy named Alex Finn, I think is his name. I found him on I met him on X and then he became this YouTuber and he guides through a lot of like code builds and this Russian guy I see I don't know names, but YouTube has been massive. I just go in and I'm like Claude Code or Replit or Codex and I can get lost in in in in in depth.

So there's just so much information out there, not only to like how you use this, but just using it itself will give you an education that's worth hundreds of thousands if not more. So if you're somebody that's new to this space, your recommendation, if I'm hearing it right, is get a Twitter account, sign up for Cloud Code, and start building. 100%. And it's even easier than that.

Like replit.com is probably where I started first. And even like love lovable.ai, those are like code agent platforms where you can just have a conversation the same way you do with cloud code, but then you hit one button and it publishes it to the web and you can build anything you want.

You don't have as good of an understanding into the back end of the systems, but you can connect APIs and you can get it activated. And so I just launched a YouTube tutorial for talent engineering. And in the first episode, the last five minutes, we go through building your first website and launching it onto the web. And it's literally one click.

It's editable like by text and like through natural language chatting in English. And that was how I got my first like experience in. and built websites and built systems that actually generated substantial amounts of money and then learning how to use cursor and actually using cloud code on a server, paying API costs was insane. I did that for months and then realizing like finally getting afraid or like getting over my fear of using like cloud code on the desktop app and that has changed everything.

And I've saved so much money. I was spending some days three five hundred dollars In in one month I think I spent like five grand on coding tokens and now I don't I think I'm spending like sixty dollars extra on top of the two hundred dollar subscription. So it's it's pretty unreal what's possible and available out there right now. What were you saying you were doing and s you were like how were you spending that much?

Because you were using So was using the API. Yeah. So if you use your API and you're using the models like Opus. To generate your code on like an external server.

Like I spun up a digital ocean droplet. We'll get into that in one of my like tutorials. But it's basically like a virtual server. And you load it with all the tools it needs, libraries and things to function, and then you load Claude code onto it, and you can talk to it the same way you do with Claude, and it starts building the code base natively inside of the server.

But it doesn't have all of the context, it loses track very quickly of what it's built. And then like reloading context and and like even updates, even just like a days of run cost me sometimes three hundred to five hundred dollars. So it was it was like insane. And so now what I do is I build inside of Cloud code, I push all of those updates to GitHub, and then I just pull those updates from GitHub onto the server and I don't spend any of that API money that I used to.

Yeah, that makes sense. I used Claude today and I was like, Yeah, like You know, flood over Open AI. Let's go. I'm definitely on that bandwagon.

But like I've got friends who are I need a website. I've got this company that I'm building and I just don't have time to build a website. I'm like, dude, you need to just use an agenic code system. And that's why I'm such a big fan of Replit and Lovable, because right on the web platform you log in and you type in an idea and you get to start seeing it built in real time reflected back at you.

You don't have to load it into some local port on your computer just to see it on you know screen. There's not like a y confusion of like how do I get this from this state to the web. Literally you're already seeing it there for you. And I think that's where people starts to unlock and it did for me.

One thing that I just want to outline, which I think is important going back to the fact that we've both been in that talent acquisition or recruiting space, is that you've gotta be great at asking The question why multiple times until you you get an answer. And I think to the audience, that's something that I want to promote. And a lot of what you've built and explained here aligns to that message as well. Is if you hop in and you ask plenty of whys with the tools, then you'll get the right answers.

It may take you some time. It's may that it may take a period of evolution, but it's concentrating on continuously asking why. And when you don't understand something, asking for a new update or a new explanation. And seems like you you've gridded that out, right?

You vibe coded this tool in that way of figuring out the why. A lot of the guests that I'm talking to that are in this space that are innovating have gone through that experience. And maybe that's kudos to recruiters out there and people from that talent acquisition world having that experience of talking to so many candidates and asking them why. Why do you want a new job?

Why are you looking to make a move? Why are you asking for $200,000 more than what the market bears? I'm coming to some kind of like epiphany here on this one, Kyler, which is funny, man. Maybe recruiters are will dominate the world in the future.

I always think recruiters are the best CEOs. I always believed that. I know you mentioned you just recently launched this YouTube channel. Where can people go to follow your work and stay connected with what you're building?

Yeah, you can see the agency at optimal.tech. If you want to follow along with what we're building and the lessons that we're putting out there, you can go to talentengineer.dev.

And the YouTube series is now live on Talent Engineer on Tal Talent Engineering. But yeah, my my handle on YouTube is just Kyler dot frisbe. You can find me out there and then Twitter and LinkedIn is probably the best way to chat is on LinkedIn. But ⁓ yeah, I'm available.

A lot of people ask how do we build what you build? And I can't necessarily give them the exact secrets. And that's why we've decided to build this this little tutorial series of how and what and just to get people that first little step into the space. It all starts with sharing some knowledge and then giving it a shot.

And that's it's great of you to share that. Everything that we talked about today from building his own recruiting agents through vibe coding, sourcing, getting smarter and Everything that needs to remain in order to stay human. It's gonna be in the show notes. Thanks for helping the audience stay unlocked, Kyler.

Don't be a stranger.