Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
How Founders Build HR Tech That Recruiters Actually Use with Steven Lu
5 Plays13 days ago
In this episode of Behind the Build, Curtis Forbes speaks with Steven Lu, co-founder and CEO of Pin, about building recruiting technology by staying close to users. Steven traces his journey from writing code as a teenager to founding Interstellar (later acquired by Greenhouse) and launching his latest company.
They discuss why sourcing software emerged from sales tools, how top recruiters really operate, and why automation should enhance personalization — not replace it. Steven also reflects on failure as a prerequisite for success, staying hands-on as a founder, and why curiosity matters more than credentials.
This episode is perfect for product leaders, founders, and operators building people-focused technology.
About Steven:
Steve is the CEO and Co-Founder of Pin.com, a sourcing platform that's changing the way we search for candidates. Steve was also the CEO and co-founder of another sourcing platform called Interseller which later sold to Greenhouse in 2021. He entered the world of HR-tech in 2016 but still has his passion in writing code and building products.
Recommended
Transcript
Introduction to Steve Liu and Pin.com
00:00:07
Speaker
Welcome back, everyone. This is another installment of Mustard Hub Voices Behind the Build. In these episodes, I sit down with the people building, backing, and running better workplaces. I'm your host, Curtis Forbes, and my guest today is Stephen Liu. Steve is CEO and co-founder of Pin.com, a sourcing platform that's changing the way we search for candidates. Steve was also the CEO and co-founder of another sourcing platform called Interstellar, which later sold to Greenhouse in 2021. He entered the world of HR tech in 2016, but still has a passion in writing code and building products. Welcome to Behind the Build. Thanks so much for joining me, Steve.
00:00:44
Speaker
Yeah, thanks for having me. um So did you originally start off on the technical side, writing code, building products, and then building companies?
00:00:57
Speaker
Yeah, yeah. So I started writing code when I was 12 or 13. So i had a passion in it. i like My first app was actually on the Facebook platform. So I built one of the first apps using their API. Eventually, i kind of like that's when i was like,
00:01:15
Speaker
14, right? Then after college, I was like, okay, cool. I still have like that startup bug. I want to work with like really passionate people, like like build cool products, things like that. So I worked for a company called Compass back. That's the real estate company now. yeah I was like, they're...
00:01:33
Speaker
fifth engineer at the time. So I'm always kind of like passionate in just like building products and also just working with cool people. um Eventually, I was like, all right, I want to start my own startup journey. And actually took me like basically, i think about a year before I got into ah HR tech. And then from there, just started my journey ever since just doing like ah HR tech, sourcing software, that kind of stuff.
00:01:56
Speaker
Amazing. So early at Compass, early on the App Store, you're kind of always sort of right at the front of of technology and innovation. And I love that. So you have a pretty impressive career um you know in technology. And I think, like you were saying, most the experience hasn't been in HR tech space. I guess over the last nine years, it has about a decade or so. ah You
Transition to HR Tech and Founding Interstellar
00:02:20
Speaker
made that shift. um What brought you to the world of human resources?
00:02:25
Speaker
Yeah. so um So originally I was building sales software. So meaning I was building like outbound sales to email, that kind of stuff. ah When I was in college, I was helping Rutgers build like spam filters and email software. So I kind of was like, okay, I know this space decently well. I kind of know how to get around spam filters. I know like the the do's and don'ts. But back in the day, there used to be two companies, Sales Loft and Outreach, which I'm sure everyone knows. They had big war chests. So i was like, okay, well, we can't do this. But I had a buddy
00:03:03
Speaker
who went from Squarespace, who was like, hey, I need something similar for recruiting. Can you do this for me? was like, okay, let me build you something. And over a weekend, I built him an interface to work with the software. And from there, we just kind of like entered like recruitment. And then we just... I just saw saw like an opportunity that was untapped at the time.
00:03:22
Speaker
It was like, Okay, cool. Let me each keep building the software for recruiters. and it And it hit pretty well, and it expanded pretty, pretty largely. And then we started getting into like external recruitment, all that kind of stuff. But Yeah, Interstellar was originally built for sales, then shifted into recruitment. And ever since then, like that that that decision like ah made it very well because sales is obviously super saturated market. I know the the money is better there, but recruitment was more interesting to me because I'm always kind of like a people person. um
00:03:56
Speaker
My Chinese name actually means people theory, which is kind of hysterical. I didn't know that until like... a year and a half ago when my wife told me about it and I was like, okay, that's what my Chinese name means. um my My parents told me something else, but funny enough, that's actually what it means. but Yeah. They kind of mapped out your whole professional journey without you even realizing it.
00:04:17
Speaker
Yeah. Yeah. um Yeah. Just fate, I guess. So um I'm kind of curious when you say you saw the
Innovations in Tech Recruitment
00:04:24
Speaker
opportunity. So you were building you were building software um for recruiting, but on the email side, right, on the outreach side, to be able to sort of get around some of these filters and help reach people, what was the opportunity that you saw um that you, you know, that, I don't know, um maybe I'm not even sure, I guess how to exactly articulate that question. But you mentioned you saw the opportunity. I'm kind of curious, what was that opportunity that you saw?
00:04:51
Speaker
Yeah, so so back then, it was basically a strategy that a few kind of like tech recruiters kind of created. The best ones, I would say like the top 5% of those like tech recruiters, utilized a strategy that was ultimately like, let me go find the candidates and reach out to them directly cold.
00:05:09
Speaker
um And the idea was that like people don't typically apply to your role at like the tech level. back then, um especially the high end talent. So you have to reach out to them like kind of like a sports agent and basically poach them from where they are. So I saw that opportunity there. And what I did was I built software to make that more efficient.
00:05:30
Speaker
Typically, what I saw was like, ah They would send an email, put it on a calendar to like then send a follow-up reminder, and then send that email and put another calendar email or calendar event to send another email. And basically what they were doing is just kind of like scheduling this out. And what we found was like... you know Typically, most candidates reply in the second or to the third email that you personalize. So I thought, why don't I just make that fully automated utilizing software? So you create a sequence, you send the first, second, and third email, fully customized, replacing the name, do it in batches. And basically, where do we get the data? you know Pull in some LinkedIn profiles and then start sending those emails in bulk. So the idea was that like I could do that
00:06:19
Speaker
portion of their role about 100 times faster than doing all of those parts manually one by one. So I saw an opportunity to build software like that. And back then, it was actually just us and Gem at the time. They were called Zen Sorcerer. And they like we were basically competing against each other as we kind of grew up because we both started at the same exact time. So Steve Bartel and I were like kind of like the incumbents of building software around this new methodology that the tech recruiters were using.
00:06:49
Speaker
I love that. So it was Interstellar your first
Interstellar's Success and Acquisition
00:06:53
Speaker
company? Was that your first, the first one that you founded? Yeah, so that was the first one. i would say the first successful one. i think the ones beforehand, there was a lot of unsuccessful ones, things like I was writing jewelry, or sorry, like ah process code for jewelry. So we were like putting, making like wax molds, and then like putting that into kelm and yeah, producing jewelry. um Yeah, I had a lot of failed ones before that. But yes, Interstellar was the first business that was successful in my career, at least that I founded.
00:07:27
Speaker
i um I mean, I think that's pretty typical, right? ah you know, at least when it comes to entrepreneurs, you know, there very few are or batting a thousand, if if any at all. But i am always curious about, ah you know, what inspired other founders or the co-founders to kind of start their own, um you know they're you know, their own businesses.
00:07:45
Speaker
Yeah. um What, you know, after Interstellar was acquired by Greenhouse, what made you jump back into entrepreneurship? And did you stay with it even after post-exit?
00:07:58
Speaker
Yeah. so So post-exit, I was like, you know, part of the team just to integrate the product straight into Greenhouse. So that took a little bit. And then I had a kid. So funny enough, I actually started my new company also while having a newborn child. ah But yeah, the the thing that got me back into the HR tech space was actually
Creation of Pin.com and AI in Recruitment
00:08:20
Speaker
AI. So I i was like, kind of like, there's there's a new frontier of computing that i was like, okay, cool. This is kind of interesting.
00:08:27
Speaker
There's like basically layers of ideas that I had prior to the acquisition where I was like, okay, what if I eliminated the operator of these tools? What if I just like fully automated this? I had this idea ahead of time, but it required a lot of engineering manpower to do. But LLMs and AI is basically bring that fast forward, that frontier for me. Right. So what typically would take like three or four months to build one model, I can do in a matter of three hours by utilizing AI and LLMs that are off the shelf today.
00:09:00
Speaker
So I found kind of that opportunity. I was like, AI is really interesting. I want to figure out how to change the space rather than just make the space more efficient. So i was like, okay, cool. Like, I really want to figure out how to build like a really good search engine. Like today there's Google and today there's like LinkedIn for for recruitment, but I wanted to figure out how do we do sourcing better than what we're provided today? And so that kind of got me back into the HR tech space. into building the best search possible ah for candidates.
00:09:32
Speaker
which Which brings us to PIN, right? So tell me about pin.com. What is it? What does it do? Yeah, so we are an AI sourcing platform that started from the ground up. So what we do today is we've basically obtained like 200 million plus resumes, maybe more than that now. And what we've done is we basically like let AI loose on every single person's resume. So typically, like on a public resume, maybe only 20% of them actually have like valuable information and data in it. filled by the candidate and the rest of the 80% is really empty or sparse. So what we've done is we've trained an AI to look at the entire talent market, then rewrite every person's resume by scratch, which then gives us the ability to look at even the people in the darkest corners that have described themselves well, but could be basically the top 3%, top talent that you would hire. And so ultimately, we're trying to deliver a new search experience, deliver a new like AI portion for for finding those candidates. And so I like to say this, which is like the difference is on LinkedIn, you were typically taught for the last 10 years to take a search that's in your mind of like, this is my perfect candidate, and then translate it into computer code, which is Boolean search, right? And I'm taking keywords. I'm trying to find the specific things similar to like Google. And I'm trying to find specific keywords. And what we've done with AI is you can just literally naturally tell us who you're hi like looking for. And AI will just evaluate every single resume in person for you. So that way we're delivering like hits every single time. So typically at PIN, our like acceptance rate on PIN,
00:11:16
Speaker
delivered candidates is like 70 to 80%. Whereas like most other products that depend on that Boolean aspect is lower than 30%. So we are trying to deliver just a way better search experience that you can for recruiters to find that like perfect candidate that you want, rather than you needing to scan like 70 bad resumes before you get to one good one.
00:11:40
Speaker
That's incredible. um you know that I mean, that though those metrics alone right are are sort of mind-blowing. I'm just out of curiosity. The fact that you're you're taking all of these resumes and rewriting them, um I'm curious, is there a risk that you you know essentially become – putting words right in in the mouth of those um those candidates, right? Does that risk overstating their abilities or even understating them to some degree?
00:12:13
Speaker
Yeah, so it never replaces the interview. I think it just is to help basically surface people more than anything. um And there's like a point, like you're probably like describing hallucinations that are typically like valid in AI in the LLM space.
00:12:27
Speaker
But we're what we're doing is more kind of like bringing factual evidence from data in places that are surrounded that person. So thinking of like the company that you've worked for, the colleagues that were beside you, or like things that we know about your technographics, or your funding rounds, or any of that kind of information, we basically kind of feed to the LLM to write that resume after. And we build kind of like like this like pseudo resume that's specific to the facts or that like of the data that we know around you to be like
00:12:58
Speaker
factually evident, right? Like this is really good in the sense of like, there are people who can like describe themselves really well, and like can write every single keyword in their resume. And what we do is actually we rebalance it. So ultimately, these people who like are perfect resume writers basically can either lie or falsify information right on a resume. What we do is we rewrite that also. So we bring that down and bring like the ones that don't have any information back up. So we're just ultimately leveling the playing field for everyone and surfacing candidates that you would typically never find in a search, and then providing that for you. So that way you can get in touch with them, do a proper interview, conduct basically kind of like whatever skills based interview interviews that you have. um We're not replacing the like the interview stage, where we're we're really just trying to like, again, build a better search. So that way you're talking to the right candidates.
00:13:51
Speaker
I love that. So tell me about some of the ah the organizations that use PIN or the types of of organizations. We don't have to use any names here, but I'm kind of kind of curious, who who do you primarily serve?
00:14:02
Speaker
Yeah, so we do mostly, i would say external agencies. So I think about 65 to 70% of our business is external agencies, 30% like corporate recruiters, internal recruitment teams. I would say ranging from all types of industries, whether it's healthcare, finance, accounting, tech, designers, whoever basically like you're trying to find. We're even trying to find like construction project managers all the way to like architects. yeah I would say anyone in that category of, I would say, knowledge-based work is kind of kind of like our target target customer.
00:14:40
Speaker
Is it primarily US-based, Western hemisphere, global?
Global Expansion and Market Challenges
00:14:46
Speaker
Yeah, so we we are North America, South America, um and Europe. So we are mainly those three territories. um We are looking to expand into APAC, mostly like Australia, New Zealand, in those areas soon. It is a challenge because every every new market brings and introduces a new challenge in terms of the talent space. So like the North American talent isn't the same as European talent. So we have to build and construct different models, different AI LLM prompts to ultimately kind of digest those different areas differently.
00:15:25
Speaker
And the good news is like, also the nice part is AI also can read many languages. So I would say the coolest part about AI that when we embarked it like a year ago was like, was like oh man, this thing can read Spanish. This thing can read Chinese. And also, like all of our users can also now search using English or whatever language that they want to then surface anyone that they want rather than the specific people that are like structured to their language. So that was the kind of cool unlock that we got for free, which was kind of nice. But yeah, each and every talent market has very different needs. And so we have we have to adjust differently when you try to search for different markets. So APAC, we're learning about. I'm not very familiar with the APAC region, so it is like a team effort to like learn what are the needs there from our customers or potential customers.
00:16:18
Speaker
So why, why do these folks, these organizations, why are they choosing pin over other solutions? I mean, what pain points specifically, if you put me in the mind of the customer specific, um, does pins like specifically solve for these companies?
00:16:32
Speaker
Yeah, I think it's time and effort, right? Like you waste a lot of time talking to the wrong candidates and you also waste what I call is a lot of social capital or like basically like outbound capital when trying to to reach out to the wrong people. So a lot of times, you know, the process starts with like finding the candidate, then reaching out to them, then basically like seeing if they're interested, then interviewing them. There's this kind of like funnel that forms. And everyone's super focused on this middle part of the funnel. And so the idea is that people think like, I need to put in 5,000 candidates into this funnel to get like 10 people to respond back to me. Our idea is, let me get you 10 people that you need
00:17:14
Speaker
then everything else downstream doesn't matter. Like basically your copy can be exact and precise rather than generic. You can customize like the email specifically towards that candidate. Everything you do is basically kind of like based on that, like that funnel and the time saved and the effort like required. And so like search,
00:17:34
Speaker
If I build it perfectly towards that funnel, everything else, yeah, again, that funnel just doesn't matter. The percentages, the the open rates, the interested rates are significantly higher. And what we've seen from our customers is that like, typically if they're receiving like, let's say a 15% response rate, typically utilizing their old search, like a LinkedIn search or any other platform,
00:17:55
Speaker
They're getting relatively around now 50% response rates with close to one-to-one interested percentages.
AI's Impact on Recruitment Efficiency
00:18:03
Speaker
right So like the better your search is, the the better you're going to get in front of those like candidates without like basically landing in spam or any of that kind of stuff. right So like less is more.
00:18:14
Speaker
That's kind of like typically the model that we're trying to follow. follow That's amazing. Um, you know, I'm kind of curious, obviously, you know, we've talked about AI a little bit. It's everywhere.
00:18:25
Speaker
Are the companies that you work with, um, or you feel like maybe organizations in general, are they pretty eager to to embrace AI? Do you see any reluctance with some of the folks that kind of come to you?
00:18:37
Speaker
Yeah, I think it's more like less about reluctance and more about like, I'm really curious, but I don't know how this operates. So there's actually, I call it more like a learning curve than anything, because we're changing kind of behavior that we've taught you about computers in the last 20 years, right? In that like realm of let me Google that for you. It used to be like, and let me type in the few keywords that could surface up that information for you. rather than let me just describe to you exactly the like candidate that I want. So typically when we like onboard a new customer, they're trying to do what they did typically back in the day rather than trying to utilize it in the form that they wanted to a long time ago. So I would say...
00:19:21
Speaker
there's no reluctance. There's no, I would say people saying like, Oh, AI is taking my, like taking over my job, but more, i think there's more just overall general curiosity more than anything. Like we book, I think like, like on our go to market team, we talk to about 30 people a day at this point, just to tell them a little bit about like what AI is doing to our industry. Uh, And it's not replacing them. It's rather more just like changing the way you work and making that process more efficient and less like mundane of just trying to like spam 1,000, 2,000, 5,000 people and then maybe talk to like two people out of 10 in that day that could be good fits. right So um our objective is to help them just multiply their time.
00:20:09
Speaker
Is there an ATS component to this or is it purely sourcing at this point? Yeah, for us, it's just sourcing. I think like for me, I over index on it. This is kind of like search is the most important piece. I think there's like roughly, if I could count ah probably 100 different ATSs that I know. So we integrate with those ATSs by like basically sending that information over the communication and all that kind of stuff to the ATS. But we basically, we don't really touch the ATS space other than like, once you hit the first interview, we're like hands off from there.
00:20:45
Speaker
um But like, like what we're typically trying to do now is do like kind of like information, call recording, that kind of stuff. Because that information can tell us a little bit more on how to improve our search.
00:20:58
Speaker
So the more we can like obtain during that and interview process, the better we can find better candidates that may have not been a good fit down funnel and make that updates, make those updates upstream.
00:21:13
Speaker
I don't want to talk a little bit more about the AI, um, aspect here, cause it does kind of fascinate me a little bit, especially to get the perspectives from a lot of different founders who are, you know, knee deep in this space anyways, uh, at least, um, you know, I think a lot of, first of all, we, we know a lot of people are not embracing it as much. I mean, some having anxiety about how it's going to impact them. Some worry that it's going to take their jobs, you know, some, uh,
00:21:39
Speaker
Entry level, less skilled workers, you know, they're worried they're going to become unemployable.
AI's Role in Future Employment
00:21:45
Speaker
um Should folks be worried? You know, how how can business leaders help make people more comfortable with the ah proliferation of AI?
00:21:55
Speaker
Yeah, so there's definitely ah I definitely see kind of like a risk factor towards junior and entry level roles for sure. Absolutely. um I think whenever I get kind of like asked this question by like basically the younger generation, I ultimately tell them to say that AI is basically a tool, and right, that you can utilize And what you need to do and become is kind of like a consultant or a jack of all trades. And so you need to be able to be that person who can understand the problem really quickly, be very quick on your feet on how to like attack a problem and utilize the tools to solve the problem. Right. So there's no more, i would say, like in the junior and entry level roles, like you need to be less um specialized and more generic at basically solving problems across like the organization. And that's how you become like extremely valuable short term to an organization when you're in a junior level role. like The other side of this problem, though, is this kind of like other side, which is more specialized people. So it's never, ever going to replace a 10-plus year veteran in engineering or a 10-plus year veteran in architecture. They just have so much like minute details that isn't written down anywhere in the experience of like making a building a writing software that they've done for the last 10 years. It will basically help them kind of like, I would say like for me as an example, I can write code faster, right? I look at it more like words per minute. So typically I would type at 120 words per minute.
00:23:34
Speaker
before AI, and now I can typically write somewhere between 400 to 500 words per minute. It's just accelerating what I typically do rather than me trying to write out every single little tiny detail on the yeah on the keyboard. So for junior entry-level roles, I highly recommend being kind of like the consultant jack of all trades. I think like for the most part,
00:23:56
Speaker
I hope bachelor degrees get cheaper, significantly cheaper and school gets cheaper ah because of the world of AI, because like learning so a specialized trade is more accessible now. um And like, kind of like you need to become more of like, you need to be trained more on how to be like a very good problem solver. And that's more kind of like maybe a one or two year program or maybe an apprenticeship as like an example. Yeah. Rather than you needing to go to school for four-year engineering degree to learn how to code.
00:24:30
Speaker
Now, basically, AI can do it for you. You just need to learn how to utilize these tools and also be a really good problem solver at the same time. You know, i see in a lot of lot of HR rooms, right, there' there's a big concern over not necessarily replacing jobs, although that, you know,
00:24:50
Speaker
It's a real concern for some folks, you know, regardless, right? You talk about the specialization aspect of it, but I think that there's real concern of that talent pipeline drying up, right? Because if you're able to replace so much of the minutia of the, you know, or entry level, um,
00:25:08
Speaker
work with AI, that's generally where you used to develop your talent pipeline, right? your You have your entry-level workers, they build skills, they build experience, they become mid-level workers and so forth. And if you're replacing that whole bottom level, right, with AI, all of the sudden you're drying up your talent pipeline. And where does that come from?
00:25:29
Speaker
Yeah. So I think like... the main kind of aspect here is there's going to be, I don't know what it's going to be, but there's going to be a new realm of jobs. There's just like, ah like it's things that we practically can't fathom today. Like think of it as like when we started to have computers, who knew data entry was going to be a role, right? Like we really only had that once that became a thing. And so I think like there's going to be just an unlock of new talent where like we typically hire junior engineers to help senior engineers write code faster and their ideas faster because I can hand that off to them. Today I can hand that off to an AI and then review that code, but there's going to be something else that next unlock of efficiency for like the GDP and the economy.
00:26:20
Speaker
I just don't know what that is today. yeah and I think we'll probably see that in 10 years time. But I do think there's going to be a new wave of jobs that's actually going to be like, probably, I'm very bullish on AI, as you can tell. yeah But I do think like that newer generation is going to be able to find that next level of entry level job. Whether it's like a consultant, whether it's like, you know, helping that, like helping a senior person with AI, still like a senior engineer with AI, all that kind of stuff, right? Maybe it's like, maybe it's like, I can read the AI machine code more than anything. Yeah. or i'm everyone starts training their own AI model, not just three companies that exist today, right? There's a lot of different things that could occur. I don't know what it is, but I'm very bullish that like AI is actually going to help us rather than eliminate jobs.
00:27:13
Speaker
You know, I... um and So I'm 100% behind you on that. I actually fully agree. I'm very bullish on all that stuff, too. um I do think it is going to create um a lot of new jobs that we don't even know what they look like or what they exist. You know, I have the image. I've long had the image in my mind that it looks a lot like like an AI dog walker.
00:27:36
Speaker
Right. It's, there are so many agents, you know, and every company is going to start using these agents to perform these menial tasks or this entry level, this, or this entry level, that, and there's going to need to be an individual that can sort of walk the dog, right. And kind of keep them in line and where they're doing and where they're going. And it's going to require a lot of generalized ah experts. Well, you know, we talk generalized expertise, right. Which might be sort of a, um,
00:28:04
Speaker
What's the word? um Oxymoron. Contradiction. Yeah, yeah but but it's the truth, right? It's going to require some some guidance, right? in In some capacity. And just kind of like that data entry, you know, I could see that certainly being a thing.
New Job Categories and Skills in AI Era
00:28:20
Speaker
Going back, like, there's like AI engineers today, right? Like that job didn't exist back in the day. Yeah, exactly. Because like wrangling all of these different AI models and prompts and everything is actually a challenge, right? And so now we have AI engineers to help us kind of manage that.
00:28:36
Speaker
Yeah, and prompt engineers. And ironically, you know a lot of these jobs don't require the college degree. And those who are excelling in as those prompt engineers happen to be, you ready for it? Because we've been saying it for the last 15 years. Those are the people who excel at storytelling.
00:28:52
Speaker
The people who actually know and can understand how to connect in a straight line. you know All of these really important things. you know, details and articulate them in a way that you can start to talk to an LLM so it has all the inputs it needs to to actually give you what you need.
00:29:08
Speaker
Those are people who are really excelling at those roles. The storytelling, I mean, you know, has been such an important part of sort of our our business lexicon in the last 15 years. Yeah, yeah, communication, right? It's like a good manager communicates well.
00:29:24
Speaker
And so similar to AI is like, if you can communicate and story tell very, very well to an AI, you're gonna get great outputs, right? yeah But if you tell if you tell me like five words and expect me to like guess what you're doing or like what you want. You can like this problem exists today with humans, right? So like, like if you, if you can't communicate communicate well to an LLM, you're not going to get what you want out of it, right? It takes a lot more effort than you think it takes originally to talk to and like an LLM. But that's because basically today's technologies has gotten us to a be lazier, scroll on our phones, type five words into Google. Today's technology like basically built bad habits. But AI, I do think, is getting us out of those bad habits to make us communicate better. Maybe maybe this also solves like some other like ah you know other problems in life, which is... I'm sure it will. And I mean that right there, right? what What we've covered even the last five minutes is a perfect commercial for pin.com because the reality is, is better inputs, better outputs, right? in In LinkedIn or in Google or whatever, when you're talking about those Boolean searches, right? Where where it's either on or you're off, you're searching for keywords, you know, you'll type in five things and that's the best you're going to get, right? yeah
00:30:43
Speaker
the The outputs that you get from that isn't going to get any better. But now that you have an opportunity to work with a ah solution that you can actually talk to, you know, in a native voice about exactly what it is that you want and be descriptive, right, and articulate those things in ah in a, you know, real meaningful way, you're probably going to get much better results.
00:31:07
Speaker
Right, context, right? Context is so important. It's like one of the key foundations that we have here. And and and and then we have like ah personalizations that are like our second level in in that search space. So yeah, exactly right. It's like Boolean can only get you so far. And what you really need to do, at least for us, was we rewrote the entire engine three or four different times now at this point to learn from all the bad things that occurred. in order to build something that is really good and reliable. So yeah, we're on our search 3.2 and we're building and releasing 3.3, I think in a month, but 3.3, I think is probably should be a new major version, but my engineers think otherwise, but yeah, anyways, but sometimes you got to listen to your team.
00:31:53
Speaker
Yeah. um Steve, um I want to go back to, you know, Some concerns that AI could could introduce bias into the hiring process. It sounds like you guys do a great job, like you had talked about, right, of sort of leveling that playing field. But I've heard a lot of situations where AI has been perceived to discriminate, right, based on things like race or age or gender.
00:32:18
Speaker
Can it be avoided? Are tech companies taking steps to prevent their AI from becoming biased? Can you? How? Yeah. Yeah, so I think the major thing is like you can feed it everything or you can feed it just the facts, right? And so like in order for AI to not be biased, what we can do is remove the name, remove the photo, remove the gender, remove... basically any identifying piece of information or like any information that can introduce bias, right? And so if the AI doesn't know, then basically the AI can't tell and make decisions based on that bias. There is a certain kind of aspect where I'd say like, same thing, inputs and outputs, right? If I'm starting to train the AI on inputs that are like,
00:33:05
Speaker
terrible as an example, like my biases, it's you're just going to get like a biased output from the end. That problem has existed since 10 years ago. It has like just resurfaced more because now like the implementation of AI is much easier, right? So now governments and people are concerned about this more because companies can introduce this into their hiring prep process faster. And so I do fear that like there are some companies out there that don't know how to utilize it correctly. Just because they're so junior or new to the space, I think they will evolve over time, which is which is great. But for us, I think the main thing for us is like, how can we surface, like for us, it's more like how can we surface candidates that have never ever gotten like an outbound email from from a recruiter in the past, right? I want to basically make the talent space, again, more accessible more than anything, ah whether or not you tried really hard to like write your resume. Typically, it's like, typically if like the the hiring process all about like how well can I advertise myself?
00:34:12
Speaker
Whereas like what if I just eliminated all of the games that you played back in the day and just made it so that like I have this very specific skill that i I ideally would work for these 30 companies, right? Because I have a very specialized skill. and Now all those companies can find you because like basically like you didn't have to describe yourself. We did it for you.
00:34:34
Speaker
Hmm. Everyone's going to do it anyways. that's yeah That's my theory. Everyone's going to utilize AI to write your resume. I think that's just my theory long term in five years. It's like, I'm too lazy. I don't think I've written a resume in so long. If you asked me to write a resume, it's going straight to the AI. I definitely don't think you're wrong. It actually makes me kind of curious, like,
00:35:00
Speaker
You know, we've talked about jobs changing. We've talked about maybe some going away or, in you know, ai introducing new types of of roles.
Evolving HR Roles with AI
00:35:11
Speaker
What about HR specifically? Like, how do you say HR related jobs changing as as more HR tech kind of rolls out, the more that AI is being used? um And it kind of like,
00:35:24
Speaker
would be curious, you know what do you think, aside from that, what do you think the HR leaders I think need to learn or should be prepared for just to be able to remain competitive in their roles, right? ah Being a generalist obviously is is one, but as we talk about HR specifically,
00:35:43
Speaker
you know you're you're changing a lot of what it even looks like to source candidates, right? So talk to me about how you see the HR jobs changing. Um, uh, it's going to be weird to say, but I think we're going to go back in time.
00:36:01
Speaker
And what I mean by that is we used to like ah HR used to be kind of like more like talent management. Your, your, your role ultimately is to identify talent and foster talent.
00:36:13
Speaker
But we got to this world where software made basically this process so efficient that we forgot the main job, like the main role. part of our job, which is to like identify talent, manage that talent.
00:36:27
Speaker
Right. And I would say like talent management, h r is supposed to be really good at that. But now they typically fell back into the role of like, let me just like manage payroll. Let me just manage like, you know, like 360 reviews, let me manage a few other things. I think the most important piece is that when talent management got introduced, I think it was like the 1980s or something ah in in that manner, where those people knew how to like foster the top 5%. CEOs would hire like basically the top chief people officers because they knew they could attract the top talent. like that's That's the key thing is that like you need to be really good at that talent management piece now. which is that human driver, that human aspect of basically the HR space versus being very generalized into a role where you're like either managing benefits, processing payroll, ah like basically managing 360 reviews and just calling it a day, right? Yeah. The objective is now that like those types of people can, can really foster talent and attract that talent. That's what I think that like basically like in, in, at least the sourcing in the talent management space in my aspect needs to become better at. I think we lost that because of structured hiring as an example. yeah I think we lost that over the years. I mean, we just need to relearn that straight now since,
00:37:57
Speaker
the space changing so significantly. You, ah I like that. You're also probably the only person who has said, we're going to be going back in time. Yeah. That's why I think it's like an odd thing to say, but but, but, but in adding the context, I think you make a really good case for it. um it It makes me wonder um because I do see some of these ah HR roles changing and changing.
00:38:24
Speaker
I think you do a good job of painting that picture of what that might look like. But,
00:38:31
Speaker
and I feel like this might be a very real scenario, how do organizations plan for roles that aren't even fully defined yet or they don't even exist yet? You're hiring people because you know that You're staring into a void or an abyss and you're expecting this person to handle that. you know The role can't be fully defined.
00:38:56
Speaker
how are they going to plan for things like that? Yeah. So so undefined roles. I kind of treat hiring in two buckets. It's like...
00:39:11
Speaker
I hate to say this, it's it's like butts in seats in like like solving a need, right? I think for us, as at least part of but like my organization, I'm a really small startup, so I can't really say for like large corporations really more more more than often than not. But I would say for me, it's like, I've got a problem.
00:39:30
Speaker
I don't know how to solve this. I need to go find someone that can solve this for me. Or the other side is like, I need someone either to take this over because I don't have the time to manage this and it needs to be managed, right? And so when I am doing and conducting hiring and interviewing, that's generally where I start before I open a role versus like, I typically like,
00:39:52
Speaker
In my last company, i would say I hired like, okay, I wouldn't say the best, but I always thought for me, I was like, I need to hire another software engineer. I didn't exactly know why I needed to hire software engineer. I always thought of it more as like, okay, cool. I need ah i need to do this because it's going to help me solve time. And what I actually got was actually like, I'd spent more time. Like hiring like people then trying to train them and everything of that nature. Whereas like what I should have identified was more like what are some problems that exist in my organization today, at least on the start, like at least I would say startup side 200 or less that I need a person to help me do this. And then what I'm doing is I'm interviewing people based on that need and that skill. Have you done this before? Have you interviewed, like, have you've implemented this? And the stories that those candidates tell me give me confidence that, yes, if I hand you this problem, I know you're going to solve it for my yeah organization. um
00:40:50
Speaker
I think that that's, yeah yeah that that sounds really like, that sounds like a lot of really good advice. And honestly, it kind of, for me, it paints a little bit of a picture of,
00:41:01
Speaker
you know, as we move into this sort of next stage where jobs change, roles change, things are undefined, right? We talk about living in this sort of state of uncertainty, right? Which in a sense is sort of a very entrepreneurial thing, right? As by definition, right? Entrepreneurs are generally like creating value, right? um Oftentimes trying to create business value, right? In ah in a state of extreme uncertainty, um, which, you know, in a sense is kind of what you're hiring for in those particular roles. Right. Right. Um, so I want to keep talking about the future predictions for the future of work. What do you think is going
Future of Work and Employment Models
00:41:45
Speaker
to change? Will there be more shifts in how we work and where we work, changing worker classifications?
00:41:54
Speaker
What's your big, what are your big predictions for 2026 beyond? Yeah, I think the starting role will be a manager. That's kind of my prediction. interesting Interesting. Like managing AI agents or other people?
00:42:09
Speaker
People and agents, right? I think the key thing is that skill special, like like so to specialization will become less and less unless you need a very, like very specific, like problem solve. I think the people who are already in like the top 1% of like, let's say their space are going to be their own.
00:42:29
Speaker
like basically companies, because ultimately they have the skills that nobody else has, AI doesn't have it, and they will work for many companies at a time. I think that, let's say like like third world or third world countries will actually like evolve even faster than what we typically have seen the United States grow from the industrial revolution to yeah computer age, right? For us, at least in the United States talent market, I do think that we are going to become more managers than anything.
00:43:02
Speaker
um And eventually, as you learn more in the company that you are with, you're going to become specialized and run your own company. I fully believe in the world of small business. I love smaller businesses rather rather than bigger conglomerates for multitudes of reasons. But I think the most important piece is that I can unlock more talent if basically like, you know, there's this like fractional kind of world where Ultimately, now you've got really specialized people that I can consult with and engage with now because they're not tied up with another company. And like one company isn't eating up the entire talent market space in that like all the junior level roles and mid market, mid level roles, sorry, are all, all, all just managers at this point. They're like, cool, I need to manage five prompts to do these five things now.
00:43:52
Speaker
I need to manage 10 different AI agents to solve all these different problems for me. I'm creating process, I'm managing process, i'm I'm doing the work of a manager, typically that like is unlocked at five years experience or eight years of experience. And now like a junior level role can be a really good manager on day one.
00:44:15
Speaker
Those are a lot of bold predictions. the We've covered like modernization of third world countries you know at ah at ah at a faster, more more rapid velocity than ah than we've seen over the last 150 years.
00:44:28
Speaker
We've talked about entry-level manager roles. We've talked about a move towards you know probably more independent, fractional 1099-type worker classifications to work for many companies, a move to small businesses.
00:44:43
Speaker
I like it. I like these bold ah i like these bold predictions, Steve. Look, like i love I love seeing the at least the way we are today, I think, feels old and stale. i don't know. That's that's just me. no I have a feeling this is going to end. These predictions will you know we'll age well. You know you heard it you've heard it here first. In 10 years, I want everybody to point back to this episode right here. It'll either age like fine wine or spoiled milk. So we'll see. Even Lou, the soothsayer, said it here on Mustard Hub Voices behind the build. So ah a couple of other other questions before you know we wrap up
00:45:22
Speaker
yeah I think that um I want to talk about what what business leaders need to be preparing for. Are there certain skills or competencies that they should focus on if they're already in leadership?
00:45:33
Speaker
um And knowledge gaps that need to be addressed so that they can at least lead right effectively. We talked about the workers, right? Tell me about some of the leaders, right?
00:45:45
Speaker
Yeah, so I think I can talk about startup leadership true more than anything. i think the two things that I probably find that make like really extremely good startup leaders are people who can identify people.
00:46:02
Speaker
who can, are, that are different from one another. And I mean, more kind of like the, you know, the skills and experiences that they have and being comfortable managing and helping basically craft this like crazy team where like, like, for example, my team, no one has the same experience overlap, which is kind of hysterical and crazy and being very comfortable with that and kind of managing that. And number two, I would say,
00:46:30
Speaker
Also, a very weird thing to say is most really good products come out of basically leaders who are extremely opinionated about what they're building and like are driving ultimately could be the wrong opinion for sure, but are very strongly opinionated on how things, their products should work. And so I would say opposite of learn and listen to the customer and be more opinionated more heavily around what you want to build. Because I think currently today, the space is really like everything. Every space that involves technology is super saturated now. So you've got to be really opinionated on what you want to build rather than being a generalist application to to everything, at least like the product that you're building. So I think the best leaders kind of like know how to lead and command people towards the right mission, towards the right goal, towards the like 10 year outcome. ah
00:47:29
Speaker
Bad leaders, I would say are more kind of like, I would say confused on what they're trying to like do and build and like, ah and lead. You can kind of clearly see that when you like, they talk in an all hands meeting, one-on-ones, et cetera. Whereas like really good leaders are like,
00:47:46
Speaker
this is what that finish line looks like. i need i need these three things. And so they have strong conviction on like basically what the outcomes are going to be in 10 years time. I like that.
00:47:57
Speaker
I like that. Last last question. if If you had a single piece of advice for business leader when when it comes to adopting ai in their ah HR processes and people ops, right?
AI Integration and Business Efficiency
00:48:09
Speaker
You're on an elevator top floor right before the doors open and they walk right out the door, right? You just got ah one one piece of important advice that you can share with them. What is it that you leave them with?
00:48:20
Speaker
ah Probably an unpopular opinion, which is the amount of money ARR per headcount ah in in your organization or your department. So the way I like to evaluate it is, let's say I'm making $5 million in annual revenue. I have 10 people. right That means each and every employee is generating half a million dollars ah per headcount.
00:48:44
Speaker
per headcount, right? So, so I think the key is to think a little bit like that, which then helps you figure out if this task, job, role process, etc, can they either be automated, or needs a human in the like in the loop.
00:49:01
Speaker
And so the way I like to think about this is like just trying to find good efficiency gains without losing quality. um so That's how I think right now. It's just literally like, what percentage, like, is this role needed? Or can this be replaced with AI? Right. And that's what I'm trying to do in my organization, because we're an AI first company. And a lot of our processes should be led by AI.
00:49:26
Speaker
ah Versus like, I need a human to, let's say do so like, CS, typical CS work, or customer service. um Just to give you context, we have zero people in customer service. We have we have basically like two CSMs, which like kind of help out.
00:49:46
Speaker
But CS is like, ah for us, it's like a entire company job because the way we we operate is if we can train the AI to answer all complicated questions that you may have. on all levels of our organization, we're building and instructing an agent to help future people going forward. So I will say this, like, maybe one thing, maybe not leadership, but just everyone in general,
00:50:11
Speaker
If you can you know tell me your problem in the most descriptive way possible, I can help you even further rather than if you just gave me five words. I can't really figure out how to help you from that. i think i think I think AI has made people some people a little bit lazier because they just give like me three words. I'm like, okay, I'm going to try to help you, but I need more details than that. but Yeah. um Yeah, I would say ARR per headcount it should be a metric that you should be thinking about. The higher that number is, the better you are performing in the AI, like basically bringing your company to the AI space. Yeah, I love that. That's great advice. I really appreciate you taking the time to join me, Steve. This has been great.
00:50:58
Speaker
Yeah, of course. Thanks for having me And ah thanks to all of you for joining us. This is Mustard Hub Voices behind the build. Be sure to subscribe so you don't miss the next episode. Visit mustardhub.com and learn more about Mustard Hub.
00:51:11
Speaker
Get started for free and discover how we help companies become destinations for workplace happiness and turn culture into a competitive edge. Until next time.