Don't think LLMs are going to replace recruiters. I think we'll always have recruiters because what makes someone a really good recruiter is that capability to build good human relationships. Is AI doing any interviewing upfront? Part of the screening process, yes. However, the further down the funnel you take it, the harder it becomes to envision AI actually doing those things. Like, there's not that many important decisions you make in your life. We want companies to really identify who's a good fit for the roles they're hiring for and then strategically reach out to those people and engage them. Okay. So twenty thirty, where do you see things being from a talent acquisition space? I think you'll look a lot more Welcome today, Ishan Gupta, cofounder and CTO at Juicebox. Welcome. Very nice to be here. Thank you so much for having me. I'm excited to talk about this because this one is kind of near and dear to me with recruiting and how AI can help in that space. Before we get into it, you guys raised eighty million recently at a, what, eight hundred and fifty million valuation. Yeah. Pretty incredible. I'm curious. And you guys what? Twenty twenty two is when you found it. Right? Yeah. You dropped out of Dartmouth at nineteen to do this? Yeah. Good decision? Good decision. Good decision. Can't complain. Awesome. Alright. So one thing I'm interested in is we know, like, AI is solving a lot of problems for candidates, obviously. Think I got a buddy who, his wife was, coming out of being a stay at home mom for a long time, she was looking for a job, wasn't really getting one. He wrote a little bot that went and looked at all the the found all the places for her to apply, rewrote her resume based on what it thought it they were looking for, and she got a job in, like, four weeks. So it was, you know, really and then she was having a really hard time. So there's a lot of great instances of it on the candidate side, but also that means that candidate volume is It's a lot more. Yeah. Is a lot more. So companies are deciding, am I gonna sort through all of that And you say I had to sort through, or am I gonna be more proactive and do outbound? Exactly. So a lot of customers that we have today, and we we have over four thousand customers, and there there's companies who are, like, startups, there's companies who are enterprises, there's companies who's, like, literally just the founder trying to find their, like, first founding engineer or something. All across the board, we keep hearing this exact same thing across industries, is that people are overwhelmed with inbound applications. Because any time you put a role out there, like anyone who's hiring knows this, that you instantly get thousands of applications. And the reason for that is, in some ways, LLMs have made it really easy for people to mass apply the roles. So inbound is not the signal that it used to be. Like traditionally, when people used to apply, they would be like, hey, I'm going to find ten companies that I really like, and I'm gonna send them my application, and I'll do a really good job of filling out that application. That's not the way people approach it now. It's like, hey, I'm I'm gonna get bored, and here's a thousand different companies that I'm gonna send my resume to, and I'm gonna see whatever sticks. Because I'm gonna do everything I can to be as personalized as possible, to trick them into hiring. It's way hard as a as a as a hiring manager or a recruiter to really know, like, who's actually interested, which is why a lot of companies we work with, they're not getting any signals out of inbound. More and more companies are trying to go outbound, and that's why we see the future of recruiting going. And that's what, like, JuiceBox enables you to do, is that instead of, like, sifting to, like, tens of thousands of applications, out of which most are not going to be relevant and people are not going to hear back from companies anyways, it's better for companies to be more strategic about who's a good fit. Like, there's not that many important decisions you make in your life. Maybe you you decide who your partner is gonna be, you decide where you're gonna live, and the third maybe the third most important decision is where you're gonna work. Like, what are going to spend your life on? So then that should not be based on happenstance. Right? So that's the thing we try to solve is we want companies to really identify who's a good fit for the roles they're hiring for and then strategically reach out to those people and engage them. And that's where LLMs come in. Like, that's what our AI agents do, they help companies identify the right people to be reaching out to. How does it do in the outbound? It uses email, so what really happens is you'll you'll type in a job or you put in like, you describe who you're looking for. Our agents will look for the profiles that match what you're looking for. So you and and you can have as much nuance as you want. Let's say you're looking for someone who has experience building large scale data pipelines for consumer applications. So why nuanced search? Someone might not exactly say that, but our agents are smart enough to sift through thousands of profiles, surface those profiles for you, and if you feel like they're a good fit, then you can just one click and it sends them emails and engages them and tries to get them into the process. Just out of curiosity, like to get into the weeds on this one because I'm curious about the email side. Would a client be setting up, connecting to their own email and it would come from their domain? Yep. Exactly. So they will be connecting their own mailbox, and then the email will be coming from the hiring manager or the recruiter who actually wants to get in touch with the candidate. So they have to be careful in terms of, like, getting their own domain blacklisted or anything since they go too heavy on this. We're saying candidates are spamming businesses. This is also gonna prevent businesses from spamming candidates Of course. Because they're gonna burn out of their domain. Yeah. Exactly. And you you what JuiceBox forces you to do is it forces you to be more strategic about things. Like, it's it's not like, hey. Let's look for all software engineers that work at Google and shoot them an email. That's not the way it works. Like, it tries to understand what kind of engineers are going to thrive at your company, identify the right fit, and then send good messages. So then you're not blasting email to, like, hundreds of people, you're only reaching out to the people who are actually off it, and you're getting better response rates out of it. If it's sending through email, then it's sending to their email. Is it sourcing their email from, like, Apollo or ZoomInfo or something? Yep. We work with a number of providers, like, such providers in these spaces, and we we tend to have this, like, waterfall of sorts, where any time you identify a candidate, we're going to look for their contact information across different providers and we'll keep looking until you find it. So you don't need to purchase credits to like ten thousand different softwares and then figure out where you're going to get their email. We'll do that lookup for you, and we'll find what email works and then reach out to them. What if you only have their work email? Are you We and their work usually don't reach out to work emails because, of of course, because of the sensitivity of offering someone a new role, we don't wanna be doing that. Well, they're not gonna respond. They can't Exactly. Weren't crazy for them to be like, that sounds great. Exactly. So we we don't reach out to work emails, but because of the amount of providers that we work with and because of the coverage we have across our datasets, they're usually able to find personal emails to reach out to people. What do you feel like the because obviously, we have, like, the the you're moving from, like, a Boolean search of of a specific keywords being then at risk of people just keywords stuffing their own, you know, LinkedIns and and whatever. Right? Because this is also this is scraping LinkedIn, would assume. Right? This is looking at people's LinkedIn profiles as well? We work with vendors to collect data from, like, different So whether it's on LinkedIn or mean, any data I have on LinkedIn would potentially be something that could be analyzed by this tool. We index data from pretty much a lot of, like, professional sources on the Internet. So whatever you have publicly available on the Internet, there's a good chance we have that on the platform. Okay. So if you're gonna be reaching out to people and you have a keyword that you wanna search, right, the old way was, I hope they have that keyword in publicly available profiles out Now you are looking more at skills. Like AI is able to say, okay, these are the types of skills you need of this role. They may not have the exact keywords, but this person has done projects or worked on these things that would mean they have these types of skills. Is that right? Yeah. Exactly. Exactly. So it that that's the problem we wanna solve. Right? Like, it it just makes no sense that if you don't have a keyword, you're not gonna match a search. That's exactly what JuiceBox solves. The way JuiceBox works is we're not gonna be looking for specific keywords or anything. We will read through your profile, we'll look through your experiences, and we'll try to build a semantic understanding of whether or not you match a skill set. And that's similar to what a hiring manager does. Right? So if if I'm an engineer and I'm looking for other engineers to join my team, I know exactly the kind of technologies I'm going to work with. Now someone might not work like, mention exactly those technologies. Most people won't, actually. But I know what adjacent skill sets are. I know what adjacent technologies are. I know the companies that are likely to have these technologies in their tech stack. So I'm gonna be targeting those companies. I'm gonna be reaching out to people who have adjacent skill sets and that really widens my talent pool instead of just reaching out to the five people who decide to mention the exact same keyword. That's the way Juicebox works. It's like if you're if you're asking for a specific capability, we're not just going to like find people who mention it, we're going to find anyone who could be adjacent and there's a good info, like there's a good reason for us to believe that they may have the skills that you're looking for. And that's how we're able to, like, identify a much wider pool set, but of much more, like, targeted scope, like, people who are truly a fit for what you're looking for. As we get deeper into all of this, and let's just say this this becomes widely adopted. Right? There's there's either you have recruiters using this and there's still human in human recruiters, or there are no human recruiters, whichever way that ends up going. Right? Are we gonna get to a point where the human humans are training, are basically giving as much data as possible to the LLMs so that they can be selected. Because I think right now, you're, again, scraping publicly available data, but, I mean, wouldn't, you know, something like a LinkedIn, you you purposefully go on there and give it information about yourself so that you can be found, so that people know what you're doing. I always think about, you know, DISC assessments if companies are into that or, like, communication styles or things that that if I knew more about that candidate, I could know whether they are a better fit or not to hire. And I just wonder if we're gonna get to a point where humans are giving as much data as possible to these LLMs so that they are more, you know, qual or thought of as more qualified for a specific role. To be honest with you, let let's start on the first point. We we don't think LLMs are going to replace recruiters. I think we'll always have recruiters because what makes someone a really good recruiter is not their ability to go on LinkedIn and find people who are a good fit, or not their ability to sift through a bunch of resumes. What makes a recruiter a really good fit, or like a really excellent recruiter, is their capability to build good human relationships. And that is the part we're not trying to replace. So once you have someone in your process, Juicebox does not do anything from that point onwards. You're not doing your interviews through Juicebox. You're not making your offers through us. You're not actually trying to close a candidate through us. All of that is still the recruiter doing it. It's still the recruiter driving the process and actually closing candidates. And that's what they want to do. That's what excellent recruiters want to spend their time on. The extreme example of that are executive recruiters, like people who source executives, because they're rarely spending their time on sourcing and they're spending much of their time on actually selling to executives Right. And making a case for why they should be joining a company. And we believe that more and more recruiters are going to be spending their time on that, on building relationships and closing candidates, rather than just trying to find people all day. And right now, that's not true. Like, they're spending the bulk of their time just trying to find people, which is exactly what we wanna replace. You got a good point. I mean, there was a one of the companies I was at, we would use this one firm that was it was a hundred thousand dollar Yeah. Retained, not contingent, but retained search when we go to find like, I I was hiring at the time a a SVP of marketing. We paid a hundred grand for them to go find that person, and that's what I mean, if you think about the idea that relationships are what is is your your new currency. Right? You don't I think maybe you'd have recruiters that were able to get up, rise up a bit into that world. There's probably a lot of recruiters that go, oh my god. I'm on contingent right now, and and I'm only getting paid when I hire someone, and I've got a million searches, and all my time is spent on the outbound. Exactly. If they can have the outbound automated for them, they could potentially put more time into the relationship and charge more. Exactly. So then you're you're spending like furthermore, like, at least on the agency side, like you mentioned, it's even more overwhelming because you're not just doing the searching, you're not just closing candidates, you're also doing your BD work. Like every agency recruiter is also going out there and trying to find clients for themselves. So they're like triple overwhelmed. It's not just like one thing, it's actually three different parts of the job that they have to do. So if we are able to support one of those parts and we are able to take away the aspect of actually having to look through like millions of profiles and try to figure out skills you've never heard of, then you have a lot more time and we unlock a lot more productivity because you're spending more time finding clients and you're spending more time closing candidates and building relationships with them. Is there a way to allow AI to know who the people are that have worked out really well in the business? Yes. So that's also something we're planning to spend a lot of time on is is there's there's a whole treasure trove of data that every single company has. That's data around, like, who has been successful at the company. And there are a lot patterns in there that can inform your decisions about future hires. And right now, all of that knowledge lives amongst really experienced recruiters, lives amongst hiring managers who have experience hiring a bunch of people on their teams. And it's scattered across a number of different people, and there's no central unit that owns it. That's what we want to become. We want to feed back data from how people are performing in your interview process, how people are performing at your company, and take that data and help you identify better talent. So what that means is like looking looking at interview feedback, looking at different notes across the company, looking at people's performances, trying to figure out why the people who are being successful at your company are actually successful, tie that back to your sourcing, and make sure we're getting more of these people in your funnel. Every company also has their own way of defining what a successful person looks like at their company. Yeah. Right? Like you could be a great engineer or a great designer at one company, and you could be a bad designer at another company because they have different ways of working. Yeah. So what we wanna do over time is we wanna truly understand what is it that makes people successful at these companies. LLMs are able to parse that information and draw meaning out of it, save that as their memory, and then use it over time to get better. Yep. I worked at a company where you had to move incredibly fast when you first started in order to, like, be thought of as, okay. You've got this thing. Right? Anybody who came in who was, just a little bit slow out of the gate could have been a phenomenal x, whatever the role was, but weren't really given the opportunity because they had moved fast. So the only way that we really knew how to source for that was go, what are the other companies that have really strict DNA when it comes to makes know, so we'd end up just poaching. They get eight companies all just kind of going after each other because they know that they're the ones that if someone made it there more than six months Yeah. They're good. I know. That makes sense. It's very similar for us as well. We try to look for people who are doing fast moving internally for, like, our team as well. Yeah. And that's why, like, people who have been in fast environments, who have been in those high growth environments tend to thrive because they know how to operate with that mindset of clearly doing things quickly and making faster decisions. But there's gotta be more data points available. If you had access to more data in terms of whether it was the Firefly recordings of their, you know, interviews or even even transcripts from their internal meetings to see how they communicate and how they work, there might be, if you again, if you had access to all those data points, you might have a much bigger opportunity in term than just going, well, these eight companies are move fast, so let's only pick from them. Yeah. Right? So it's it's all about the data you feed in. And I think it's interesting with me, I I I have a marketing background, and I hate giving any kind of data to, you know, cookies or anything like that because I'm like, I don't wanna be served that. But now with how helpful LLMs are, I'm like, ****, take whatever you want, you know, it's helpful. I guess the the the trick for you guys becomes how do you get companies to want to feed that data in, right, to give you more of that data so you can do that kind of analysis. That makes sense. Yeah. Of course. We we deeply integrate with a number of systems, but as you mentioned, there are, like, good amount of, like, nuances around that because a lot of companies use legacy systems, a lot of companies use systems that haven't been touched in years and don't have a good API. So there are a number of challenges that we navigate as a team around being able to integrate with all of these different kinds of systems. Recruiting is also sort of a space where it's a it's very fragmented in some ways. The the ATS market is huge. There are like so many different companies that are in that market, so there's a number of players. Whereas in CRM sort of, in sales CRMs, for example, it's it's all concentrated towards, like, Salesforce and HubSpot, but that's not really true for the ATS market, which is why that's actually a challenge we deal with, is being able to integrate with more and more softwares across different teams. Because you guys will act as a bolt on, do you just own? You're not trying to own the whole process. Yeah, not really. We we deeply integrate with your existing systems. We wanna meet where you are, like, we wanna be embedded in your workflows, We don't want to add more friction. Like, the way Juicebox becomes successful is not by you doing like a six month migration to move all of your data, it's plug and play. You set it up this week, you set up your systems, you integrate everything, next week it should feel so organic that it's like a part of your everyday work. Okay. So how big are you guys now? We are forty five people now. Forty five. Okay. So forty five people, eight hundred and fifty million valuation, you got about eight eighty million from Yep. The last round. What are gonna do with it? Like, where's that going? Lot of things. One is we wanna invest in product and r and d. We wanna go deep. Our our goal is not to go broad. Our goal is not like, hey, we wanna build a massive platform that does everything in the world. That's not what we are trying to be. We wanna really go deep on this problem of search and identifying the right talent. And it's it's a pretty wide problem. Like, it's it's way different to hire an an engineer in the Bay Area versus a designer in New York versus a an HVAC maintenance worker in the Midwest. Like, completely different kind of searches, completely different kind of nuances. So how can we build a software that understands your nuances across all of these different industries and help you find the best talent? A lot of our effort is going to be to go towards building our agents, building a good memory layer, and investing in product and R and D. A lot of it is also going to go towards aggressively scaling sales. We've had an amazing year where we're growing pretty consistently and AEs at JuiceBox are very successful and it won't make sense for us if we weren't like actively scaling our sales team. We recently hired a VP of sales as well, who's been building out the function, hiring more and more AEs, hiring more SDRs. We're also planning to open an office in London and go international and work with some more European clients as well. That's awesome. Okay. So company's growing, you're going deep instead of wide, that makes sense. Where when you say deep, are you like, right now, by the way, you guys are doing the outbound. Is AI doing any interviewing upfront? No. Okay. Do you feel like that should be where AI is involved in this process? Part of the screening process, yes. Like, if if you want, like if there are small little things that you wanna check off your list to understand, like, what someone is looking for, and, like, more transactional questions in that sense, those I think AI interviews can actually do. However, the the further down the funnel you take it, the harder it becomes to envision AI actually doing those things. Because a lot of recruiting is human relationships. Like at the end of the day, you work with someone because you trust them. Even if it's not trust, you at least build some degree of conviction to build more trust with them. Right? So you have to spend time with humans. You have to spend time with people you're going to be working with to really get a good understanding, which is why the AI interview space is one where we, to be honest, in some ways, we are we are watching closely, and we're trying to figure out to what extent can we actually use AI interviews as screens to, like, drive someone through the process. But I I don't foresee that happening anytime soon. I don't foresee interviews getting replaced by AIs. One of the things that I we find interesting, because we, you know, at at Cadre, we've got we do a lot of of implementation and consulting for businesses. One of the big populars is voice agents. Yep. Interesting when we do let's say we have a a voice agent that goes and does, like, an interview with a a sales team or something to figure out, like, where some of the pain points I find that the answers that someone gives to an AI agent are often less nuanced and less verbose. They're less verbose than they would be with a human. And I I I think there's part of it is just this this natural thing of, ugh. Talking to a robot, I just shut up and, like, let me get through this. Right? And I wonder if people really would be able to come off as well as they they would with a human if they were getting interviewed by AI. Yeah. I I agree with that. Like, interviews are not like, you know, here's a person reading out a question. It's it's a relationship that you're building with someone over a conversation. You're trying to, like, get to know them well. And there's a strong human element to it. And I feel like if you take that out, it's really hard to incentivize people to really put in the same amount of effort you would expect in a one hour interview. In a one hour interview, feel okay, this person is listening to me, right? Like they are reactive, they're intelligent, they understand what I'm saying. And you're also trying to assess them. Like interviews are not at all about like, hey, let's assess this person. Especially for early stage companies, interviews are also an opportunity to sell the company, tell the candidate about the company, and you you lose most of those elements if you if you entirely rely on voice AI. Especially where when AI comes in right at the beginning Yeah. It could just be an off putting Yep. Experience right off the bat. So that makes sense. Okay. So twenty thirty, not that far away. Where do you see things being from a, you know, talent acquisition space at that time? I think it'll look a lot more outbound. I I I don't see inbound being the future. I don't think most people are gonna be applying for jobs. Think most people will get sourced for the right opportunities. That's the first piece. The second piece is I I really believe, and this is what we're trying towards, is more people will be working in roles that are fulfilling to them. And I want to see that happen. I want to see more people be in opportunities that they are meant for, not because they had some job title or they had some company or some school in their background. It's because they are truly a fit for their area of expertise, and more people should be thriving in the roles that they're working in. I think the way recruiting would look is recruiters and amazing talent people will be able to spend more time on strategy, one, which is deciding how to think about the overall strategy for talent, like where to open offices, what are places we want to tap, what are talent pools we want to tap. I think they will spend more time with candidates, they will build better relationships, and as a result, candidates will have a better experience. Because they're not sending their application into the void, never hear back again, they're sending their applications to companies who truly have the capacity to handle those applications or talk to those candidates or actually drive them through the interview process. So I think more and more recruiters will be spending their time on building relationships, closing candidates rather than spending their time on just sourcing all day. So I got a question for you. Do you believe that there will be some some kind of platform? Maybe you guys are gonna build this. Maybe someone else can build this if it even is a thing where people, again, give as much information as possible when they're on the job market. Because, like, if you're saying people are not applying, that makes sense. But when you need a job Yep. It's weird to just sit back and be like, well, I'm just gonna hope somebody recruits me. So do you think there will be these platforms where you just you give it as much information as as you want and can? And then that business, just like you said, that business can go, You know, instead of just going being opportunistic with who we we go after, let's just first focus on this city. Because, man, it'd be great to hire five people there and have a little mini office. I think something that like that will happen, but I don't think it'll look like a platform. I don't think it'll look like another job board where you just go and put in a bunch of information. I think it'll be a lot more personalized, and I think it'll be a lot more intentional in the sense that, like, any time a company finds that, hey, this is a candidate and like I'm missing some information to figure out whether or not they're a fit, there maybe can be an LLM or some sort of an interface that like reaches out to the person, has a more more one on one conversation and tries to figure out the missing information. So it'll look a lot more like like one on one interactions, learning about candidates too, and a lot less like what a job board does today. Well, I'll tell you, as someone who spends a lot of time interviewing, wonderful. It would be really nice to get to a point where we have we have what we call a bench. Right? We we have we we're lucky. We're in the AI space. A lot of people want to be in, you know, an AI consulting company where they could do all know, be able to work in AI across all these different businesses. So we have a lot of people that are actively saying, hey. We wanna wanna commit on board. The thing that is really valuable for a company growing like ours is having that bench of people where it's like, we've already looked at them. They're great. We're in a spot right now, but as soon as we do, we get three or four to pick from. It would be wonderful to get to a point where we are where all businesses are not as reactive. Right? And there's not this huge lag between realizing you need because right now, I don't know what the number is, but, like, it's gotta be four to eight weeks in be be between deciding you're gonna open a role to even finding, you know, a good candidate for it. Of course. Yeah. It depends on the role where, like, it can sometimes be even more than eight weeks. Sometimes you're spending so much time, like, just sourcing people, talking to people, getting them through all the stages. So it's a it's a very intensive process and companies are spending a good amount of time and resources on those right now. And I agree with the bench point as well. We that's actually one of the values of Juicebox is we got this module called an internal talent tree discovery where we tap into your existing network. We try to figure out who you've already had contact with, who are people your company, like your employees, have worked with, and try to really tap into that network and identify who's already passive talent that you've had have interactions with and you could reengage. So that's that's another way I think a lot of talent is going to shift. Will build these passive, like, talent pools that they will have continuous access to. Love it. Alright. Sean, thanks for coming by. Thanks for having me. Take your time. This was really fun.