Subscribe
Share
Share
Embed
Daniel Chait, CEO of Greenhouse (the hiring platform behind 22 million applications a month), coined the term "AI doom loop": applications up 239% since ChatGPT, but 75% fewer reach the hire stage. Inside: why software engineers are the worst auto-appliers and how Greenhouse is rebuilding hiring.
AI is reshaping infrastructure, strategy, and entire industries. Host Conor Bronsdon talks to the engineers, founders, and researchers building breakthrough AI systems about what it actually takes to ship AI in production, where the opportunities lie, and how leaders should think about the strategic bets ahead.
Chain of Thought translates technical depth into actionable insights for builders and decision-makers. New episodes weekly.
Conor Bronsdon is an angel investor in AI and dev tools, Technical Ecosystem Lead at Modular, and previously led growth at AI startups Galileo and LinearB.
Disclaimer: All views, opinions and statements expressed on this account are solely my own and are made in my personal capacity. They do not reflect, and should not be construed as reflecting, the views, positions, or policies of Modular. This account is not affiliated with, authorized by, or endorsed by Modular in any way.
Host: Welcome back to Chain of Thought, everyone. I am your host, Conor Bronson. Today, I am joined by Daniel Chait. He is the CEO and co-founder of Greenhouse, a hiring platform that many of you may have used. And it's, in fact, used by more than 7,500 companies, including HubSpot, Anthropic, Coinbase, and the NFL. Daniel's been building Greenhouse for 14 years at this point and has had a front row seat to an incredible transition in the way we hire, in the way we apply to jobs, and one of the strangest moments in hiring history. He coined a term for it, the AI doom loop. Applications are up 239% since CHAP GPT launched. Many of you have probably experienced this, but resume hacks like white fonting and prompt injections are up 500%. Recruiters are drowning in a 500 to 1 application ratio. Anyone who is hiring knows that it is extremely difficult to sort through the number of applications you're getting. And the people who are on the applying end are looking for their next role are struggling too. 75% fewer applications are making it to the hire stage. Both sides are losing. Nobody's happy. And this matters for engineers no matter what stage of their career they're in, whether they're doing the hiring, whether they're looking for their next role. It's simultaneously the worst job market for candidates and the hardest hiring market for recruiters. So what broke? What's Greenhouse doing about it? And can AI fix the problem it created? Daniel, welcome to Chain of Thought. It's great to see you. Where are you joining us from? Great to be here. I'm here in midtown Manhattan in New York City, and this is just a wild time in hiring, so a really interesting conversation. I'm looking forward to it. I'm very excited to chat with you about this, especially because I know Greenhouse just put out some new information based on your data around what job seekers are experiencing. And one of the headline moments of that was that 63% of job seekers have faced an AI interview. We're all seeing AI get embedded throughout Every process today is what a lot of the show is about. But before we talk about hiring with AI and what it means for job seekers, how should they be thinking about their approach? How should we be preparing for our next opportunity? How should we be hiring teams today? I do want to say a quick thank you to our presenting sponsors, Galileo. I want to say a massive thank you to them for sponsoring Season 3. As presenting sponsors, they have driven a ton of ability for us to have incredible guests like Daniel on the show. And they've also helped many of us, including myself, to evaluate our AI agents and our AI systems to ensure accuracy. And whether you're hiring, whether you're building agents for Salesforce, whether you're doing any other work with your AI systems, it is crucial that you understand what's happening with them and observe them, understand them, and evaluate them for the success. That's where Galileo comes in. You can check them out at Galileo.ai. And a huge thank you again to them for sponsoring Season 3 of the podcast. Daniel, I want to focus in here a bit because, you know, we've been talking about a couple stats already here as I've thrown out the open. But you have this term that you coined called the AI doom loop to describe what's happening in hiring right now. And that's pretty ominous. You know, we talk about P doom in this industry sometimes, but it sounds like there's already a sense of doom that's impending for anyone who's in the hiring market. For people who haven't heard that term before, walk me through this cycle and why it's different from any hiring downturn you've seen before. It is a really interesting dynamic. And you're right. It's bad news. I mean, you know, I think it came out of, you know, empathy for hearing what was happening to so many job seekers who were just struggling to find a job and more like not feeling like the system was working for them. It wasn't meant for them. And so they had to kind of fight it or figure out a hack around the system. And meanwhile, our customers, the thousands of companies that are out there trying to hire, are also experiencing these huge problems and they're not able to hire anyone either. And so, as we started to look into what's happening, here's what we found, is you kind of go back to 2020 when all of a sudden everybody moved from, you know, work was generally done in an office and hiring was generally done in person. And what that meant was, you know, mostly you applied to jobs that were near you. And when the pandemic happened and so much work went remote, all of a sudden people were eligible for millions of more jobs because they could apply to jobs kind of anywhere. And so application volumes started to come up. And then add on to that, changes in the economy, the end of the zero interest rate environment meant that jobs were harder to come by and people were finding, you know, a shortage in available postings. And so application volumes went up again. So you had these predecessors of just like economic and job market changes that were already making it less of a candidate market and more of a hiring market to start with. And then, of course, 2022, ChatGBT comes out and everything accelerates. Since then, basically, we're all using AI more and more in all aspects of our life. Just this week, I pointed ChatGBT at my fridge and asked it what I should make for dinner. And so it's impacting everyone at home, everyone at work, and it's impacting how people look for jobs. And so on the back of that, what's happened is job seekers feeling the softening job market and sensing that they need to up their game and apply to more jobs are using AI to do just that. And so they're really doing two things. One is they're automating the application process. It's going from a world where people used to think about which jobs to apply for and then put some effort to applying into them to having AI just kind of like find as many jobs as you can, apply me to all of them. And then secondly is AI can also automatically customize every single job application according to that post. So it'll read the job description, and it'll try to make sure that your resume and your cover letter are as similar as possible to what's in the job description in an attempt to kind of give you an edge as a job seeker. So those are some of the things that job seekers first started to do, and I don't blame them. I understand that when the job market is soft, and you just want a job, and you have these tools available to you, you're going to use them. But what it meant for companies that are hiring, is that all of a sudden, those application volumes went through the roof. So, you know, since 2022, and basically, Chatube 2 was launched, application volumes are up almost two and a half times per job, what they used to be. And recruiting teams have not grown. In fact, most recruiting teams have held steady or shrunken in size. And so they now have, if you look at the number of applications each recruiter is responsible for, it's even more. That individual recruiter may be looking at four or even five times as many applications as they were just a couple of years ago. And so companies are now using AI to say, okay, I can't look at a thousand applications when I open up a job. I got to get, I got to filter this back down. And so their first attempt is to basically take off the shelf AI tools and say, hey, I'm going to throw Claude at my inbox and say, hey, can you just tell me the 20 people or 30 people I should call and forget the other 970 applications? And so it shrinks it back down. And why we call it a doom loop is because, again, that's an understandable response on some level, but that also makes it harder to get seen and harder to get a job. And so the candidates respond to that difficulty by sending in more and more resumes, and the companies respond in turn by filtering more and more out. And that's why we've called it a doom loop, because as bad as things are, and I can't remember a time when both sides were as unhappy, AI is not making it better. AI is making it worse. I feel like you just opened a can of worms for us to talk about here, where there's a million different directions we could take this. But I want to address, and I'm mixing my metaphors here, but one of the elephants of the room, which is that there is a perception that software engineering jobs are in trouble. Um, and part of this is because we're seeing a change in how those roles are hired. We're seeing a change in the leveling of where those roles are hired. And yet, when you talk to the Bureau of Labor Statistics, they projected that software development employment is going to grow 15% by 2034. Um, when you look at job postings for engineers, I believe they're increasing. So, why does it feel so painful right now, and what's actually happening to technical talent that's on the market? I think there's a number of different things happening at the same time that make the picture really confusing. And so, here's the way I see it. On the one hand, you know, it's unavoidably true that the job of software engineering and software development is already radically different than it was you know, certainly two or three years ago, but even 60 days ago. I mean, like, the pace at which change is happening is astonishing. And, you know, keeping up with that is almost its own full-time job. And so I think there's a big shift in what it even means to do that work and how those are being seen. And some people are looking at that and saying, oh my gosh, a software developer can produce so much more code. can ship so much more functionality that we need less of them. And you've seen some big public pronouncements from companies saying they're having to make layoffs or staffing cuts because of AI. I think a lot of that's probably overblown. I think that's a lot of cuts that maybe were going to happen anyway. We overhired already. Let's have a good narrative for the stock Yeah, it's a much better story to say we're using AI to be so much more efficient than like, geez, I overhired or the business isn't going so well. So I don't necessarily think a lot of that represents real productivity gains yet by AI. But what I do think, to address your question of why there's still so much hiring is, what I do think is If software is cheaper and cheaper to build, which it is, I think there's more and more appetite for it. I think there's so many problems that software can still address that haven't been tackled because there's a lot of software to build and it's been expensive that as it gets cheaper, I think ultimately there's need for more and more of it. So certainly that's happening. But nonetheless, you'd say, OK, so then there's lots more job postings. Surely it should be easier and easier. But that's where you get back to the AI doom loop. And ultimately, what it becomes is kind of a signal problem. It's like a matching problem that if everybody's sending in hundreds or even thousands of job applications when they're looking for a role, which is happening. And by the way, software engineers are the worst, according to our data. They use automation the most. They send out the most applications. We've been taught to automate for years. I mean, hey, they're cutting edge. I do not think they're the bad guys in this story by any stretch. They're just trying to get jobs. I totally empathize. But nonetheless, the fact of the matter is, when there's that big of a pile of resumes and the pile keeps getting bigger, it doesn't matter how many job postings there are. It's like a needle in a haystack. And when you add to the fact that everybody's resume is starting to look more and more the same, because AI does that really well, right? Because all of a sudden, the signal is getting more and more attenuated, is that you're getting 1,000 resumes. They all kind of look the same. And so the experience of a job seeker is just getting worse. It's like you're sending in, you know, a needle to an ever-growing haystack and you're making your needle look more like hay. It's just going to mean, you know, difficulty in finding jobs and getting matched. And so something's got to give. The process isn't working. And that same pain is felt acutely on the side of companies who are saying basically the mirror image of that. I open up a job. And look, I have a software engineering background. I graduated as a programmer. I worked as a programmer. I spent 10 years hiring programmers before starting Greenhouse. And for decades, you know, getting an engineer to respond to an outbound recruiting was, like, the holy grail. Like, oh, my gosh, I got a candidate to talk to me. Like, this is the greatest thing that ever happened to me. Because engineers basically sit there at their job and get job offers, like, showered down on them. And so, you know, that's kind of where I think recruiters are used to is like, oh, my gosh, if I could only get some candidates, I'd be golden. And now, careful what you wish for, because there are a wash in candidates. There's thousands of them at every job. And it's still a bigger problem than ever. So that's where it's just like the system just isn't working for anyone. I want to talk about how to fix this system. But I feel like there are other problems that we haven't even addressed yet, which is, Beyond volume, we are also dealing with a trust crisis in hiring. Your survey last year found that 91% of recruiters have spotted candidate deception, 41% of candidates admit to using prompt injection to bypass AI filters, and 46% of US job seekers
Host: say they lost trust in hiring this past year. This problem seems to be continuing to worsen, particularly as we're now seeing AI interviewing candidates and 38% of candidates walking away from hiring processes because it included an AI interview, which I would love to talk to you about. But how bad is this fraud problem? Is this a true underlying concern that everyone needs to worry about as well? You know, I think it's definitely a big problem and it's growing in scope. There's a bunch of different angles to it. Let me pull it, because you talked about a bunch of different things. Let me pull some of those apart. They're quite different. So one thing that's happening absolutely, as I mentioned a minute ago, job seekers are using AI to sort of magnify the amount of jobs that they're applying to and to make their resume kind of look more and more alike. In the middle of all that, There are people who are basically making fake resumes or fake job applications that aren't really at all true about what they've done. And that deception just ends up as noise in the system that has to get worked out. if you move a little bit further down into the interview funnel, you have job seekers, legitimate job seekers that actually want the job using AI to help them pass interviews, right? We've seen interview coder and other kind of interview co-pilot tools get better and smarter and harder to detect. So that as I'm being interviewed typically over Zoom and they're asking me questions or watching me code, that really I'm using AI as a job seeker to like, a lot of people would call it cheating, to basically give me the answers to that I'm being asked by the interviewer. And that's a real problem, and that starts to erode trust as well. And I sort of hesitate, or I put scare quotes around the word cheating, because, you know, it raises these really interesting questions. After all, aren't we hiring people to use AI at work? Isn't the company using AI to run the job process, to write the job description, to help interview candidates? And so, why am I at a disadvantage? I can't use AI. And companies are saying, well, look, you know, I don't want to know what Chachi Vicky thinks. I want to know what you think. And so I think the real answer is probably somewhere in between. And society kind of has yet to figure out the rules of the road as these technologies have advanced so fast about what is and isn't fair AI to use in a job interview setting. And then you have the real bad guys. So those are, we just talked about sort of legitimate job seekers doing stuff to try to give them an edge with varying degrees of okayness, then you've got the real bad guys. Because in the midst of all this noise and chaos and AI slop, it creates opportunities for bad guys to sneak in, right? Because there's like, it's hard to mind all the doors at the same time. And so there's a growing industrial-scale North Korean effort to infiltrate remote IT jobs using techniques like deep fakes. So they'll have typically a call center full of people all pretending to be the same job seeker trying to get in that one job, and they'll take turns on interviews. They'll deceive their way into your organization. They'll set up laptop farms in the U.S. And basically they're trying to infiltrate supply chains, perform espionage. And so there's some real scary stuff out there as well. And so I don't want to, you know, those are all symptoms of what's going on right now. They're obviously very different in nature and you need different tools to try to weed them out as well. And it's definitely part of why we're seeing a return to flying candidates in for in-person interviews, even if it's a remote role or onboarding in person to try to help solve this. But it seems like the problem of fake engineers applying is only increasing. What's the scope of this challenge for companies that are trying to hire remote engineers? Yeah, I mean, I think as a percentage of overall, you know, resumes that you may collect when you open up a job, it's not a huge percentage. So in our data, we see about one and a half to two percent of job seekers fail an ID verification, meaning, you know, the technology that we use in greenhouse to assure that this person is who they say they are, about a percent and a half or so are just failing those. So they're definitely not who they say they are. So it's not that many. But if you think about opening up a job and getting 200 or 300 applicants, and you realize that, like, maybe a half a dozen of those or so are simply not who they say they are, like, you better hope you don't hire one of them. And so it's less about you know, the how many of them are there and more about are you sure you're not hiring one of them? Because if you do, you know, obviously the ramifications, I don't need to tell you or your audience of hiring a North Korean spy are pretty bad. Yeah, maybe not an ideal day at work to make that hire. And listen, I mean, it sounds kind of fantastical. I can certainly tell you when I started this job and launched Greenhouse in 2012, I didn't think I'd be fighting North Korea. That wasn't part of my game plan here. But here we are. And I've seen the evidence. I have watched videos of people purporting to apply for jobs that have been discovered to be North Korean spies. I've seen the forensic evidence when they do the detailed logs of looking at IP addresses and device fingerprints and, you know, the sort of digital exhaust of when you apply for a job. And so this stuff is quite scary. But again, I think the more common thing that people come across is simply, you know, this is misrepresentation. This person either is sending a friend to interview for the job, you know, for them. And so the person I'm interviewing isn't actually the person who applied because they're trying to help them, you know, get the job or they're using some, you know, side tool to help pass an interview when it wasn't really their thinking that went into it. Those things are almost universal. Those things are extremely common. we're just seeing both sides of this process, both the interviewees and viewers lean into AI to help sort through or, you know, spread themselves farther. But it's not really solving the problem yet. How should we be trying to solve this doom loop that's occurring? Yeah, so, I mean, that's what I spend my days thinking about. And as much as we've talked about the problem and given it this catchy name of the doom loop, you know, the real job is how do we fix it? And it starts by understanding the problem deeply. I want to be clear, AI is not the problem itself. I think AI, I'm an AI optimist. I think there's a ton of possibility. There's a ton of potential. AI has already had tremendously positive impacts in almost every area of work, including in hiring. But it's the way that people, it's the tools that people have had available to them and the way that they've used those tools that's created this problem. And so that points to a way out, which is to say, what if we created better tools, better ways to use AI to help people get jobs, to help people make hires in ways that didn't make the process worse, but rather made it better? And so that's the thing I think about all the time at Greenhouse and what we're trying to do. is how can we take advantage of the power and the potential of AI to make things better and to help bring humanity, bring trust back into the process, and to help accelerate the process, make it easier, make it more efficient to hire and to get a job. And so that's, I think, where the way forward lies. One of the ways you can obviously leverage AI to help solve this is to have AI screener interviews. We're starting to see this be a common first line of defense. How is that going so far? Yeah, I think it's going to become increasingly common. You know, it got off to a bit of a rocky start. We've just published a survey, you know, highlighting some of the concerns that job seekers have had with sort of the first generation of, you know, AI-led interviewing tools. And you know, not surprisingly, that first wave, like, the voice quality was a little spotty, and there's a lot of things where it's like an uncanny valley. You feel like you might be talking to an image of a robot instead of an actual person. It interrupts you when it shouldn't or, you know, vice versa. And so I think it hasn't been always a smooth or a great experience. And I think more to the point is, additionally, companies haven't always been transparent, surprisingly, about whether you're going to speak to an interview with a person or whether you're going to speak to an AI. And so, if you're expecting an interview and you show up and there's an AI bot, it's really jarring. And so, in our most recent survey, almost two-thirds of people have now come across an AI interview somewhere in the wild, which is a big growth from where it was a year ago. But over a third, almost 40 percent of people have walked away from a process because the process has been so janky. And so I do think that there's a ways to go. I think the new generation of AI interviewing tools is much better. I think the quality of voice AI has gone way up. I think it's much more naturalistic. And I think we're being a lot more transparent now about, hey, here's how this works. And here's what you're going to expect. And I think when you explain it to people, what they start to see is like, hang on a second. The fact that the company's using AI to do, say, a first interview, it means that you can remove the gatekeeping. It means that that whole problem of getting lost in the pile, you know, sending in, you know, a needle in a haystack to thousands of other resumes, hoping that you're one of the few people granted an expensive, time-consuming, hard-to-schedule interview, Instead, you could really open up the funnel and say, pretty much anyone can get a first interview and give yourself a fair chance to show what you're capable of. And so I think as long as the experience is good, and I think as long as there's transparency about what is and isn't going to happen, I think people are going to start to really warm up to the idea that, again, this is a way where AI can actually open up opportunity, where AI can be less biased, more available. And there's innumerable ways. I mean, if anything, I've built my career here at Greenhouse on the notion that most job interviews are bad and need to be helped. And so I think the bar for the current, you know, human-led job interview is pretty low. And when you think about — I mean, I can give you so many examples. Who hasn't had the example where, for example, you know, you take a job interview, the person spends 57 minutes of the hour, you know, hammering you with questions, and then they go, all right, we've got three minutes left. What questions did you have for me? It's like, ah! you know, you don't really get a fair shot at asking what you want to know. There's all this pressure to not ask about things like benefits or vacation. But if you're interviewed by an AI, of course, you don't have those problems. It can be infinitely patient. It can answer any question, you know, that you want. It's not going to judge you. It's not going to care if you have an accent. It's going to be available to work around your schedule. So if you're working full time or if you're a caregiver, So there's a lot of ways in which the current system just doesn't work well for people taking face-to-face interviews that an AI-led interview can actually make things much better. So I do think it's a very promising technology for bringing trust and making process better and more efficient as well. I like this perspective of AI interviews, when communicated effectively and set up effectively, actually opening up opportunities for applicants. Because one of the natural things we've seen from companies as a response to this mass amount of applications has been to lean into referrals. And while referrals can be great to find qualified candidates, and those of us who are farther along in our careers often benefit from them, especially if we've been intentional about building networks. They can also be challenging for early career job seekers looking for opportunities, looking to get involved. And so having other tools in the process that can help people get an opportunity and stand out does seem like the right approach to avoid us continuing to push, you know, newer and job market entrants from being kept out of some of these roles. Yeah, I think that's right. You know, historically, just to put some context, you know, Greenhouse, across our entire customer base, collect about 22 million job applications a month. So we see lots of hiring data. And forever, employer referrals has been the most effective source of hire. In other words, they have the highest chance of getting a hire when a referral is made versus any other way of getting a candidate. The problem is there's just not enough of them and there's no real way to make more. And so for a company, that means, you know, there's only so much you can do to get referrals. And to your point for, you know, if you're a job seeker, like, You know, there's not that many referrals available. They're not available to everyone. And so it's just not a systematic answer to the problems of the job market as a whole. So my advice to a job seeker is absolutely, if you can get a referral, get one. But most people can't. It's not really going to scale. But what we can do is think about why does that work and what can we do to help it scale. And so that's why we created this feature of DreamJobs. Last year at Greenhouse, we launched this tool for job seekers called My Greenhouse and gave them access to this dream job feature functionality, which basically is simply lets them apply to all the jobs that they want across all the Greenhouse customers. But one time a month, they can indicate one job application as their dream job. And simply doing that is a really powerful signal of intent, because after all, that's a limited resource. I only get one a month as a job seeker. And so the job seeker's going to think really carefully about, where am I going to use that one? It's got to be a job I really want. It's got to be a job I think I'm really good at and have a chance to get. And so it's a little bit like kind of an early declaration of college admittance, where it's a signal to that college that you really want to go there because of the scarcity. It's the same on the job side. And so Dream Job, on the company side, when they open up their role, and if you think about that image that we've talked about where they have thousands of applications, they'll have a prioritized inbox. that will show them internal applications from other employees for that role, employer referrals that they've collected of people who work there who say, you know, I've got a candidate to refer to you, and dream job applicants. Why? Because those are the high-signal jobs, really a strong signal of intent and fit, and you're going to be much more likely to find a hire in that group. And it's really working. Two thousand people have already gotten their jobs through Greenhouse Dream Job. And they converted about five times higher of a rate than every other job seeker who applies to those jobs. And so we have shown that when you bring some signal, you know, to the table, you can actually cut through a lot of this noise. You can sort of break it or interrupt the doom loop and make the process actually work better for people. This is a lot of great data, and I think it's a very interesting innovation here, too, to provide candidates an opportunity to more strongly signal, hey, this job is important to me. I think I'm deeply aligned with it. What would be your advice broadly to people listening who are job hunting or are thinking about finding their next role? You mentioned finding a referral if you can, indicating your top jobs if you're applying through Greenhouse Dream Jobs is a great option. What's the broader strategy that candidates, particularly technical candidates, should be using in today's job market? Yeah, look, I think there's no silver bullet. It is a difficult job market. And unfortunately, the best advice boils down to a bunch of hard work and elbow grease. What I would say is, generally speaking, those auto applies don't really work. It's sort of a false progress. You feel like, I've applied to so many jobs, you know. But you really haven't. You've just pushed a button and spun the wheel. I
Host: talked to someone earlier today. He's graduating from undergraduate. He's got a mathematics degree. And he's like, look, I went into Claude. You know, he's the son of a friend of mine. He called me for job advice. He said, I went into Claude, and in about 30 minutes, I created a bot that's automatically applying to Job Stream. I've applied to so many, it's not working. I'm like, yeah, I know. I see the data. Everyone's doing that. It doesn't work. And so, you know, it can feel good to sort of take that step and see the numbers go up, but it's not really getting you anywhere. And unfortunately, the real advice is, you know, as you kind of said, like, if you can find referrals, definitely do. By the way, you don't always have to ask people, hey, can you nominate me to be hired for a specific job? But asking for advice is a great way to get a job. There's an old adage in startup circles, if you ask for money, you get advice, and if you ask for advice, you get money. Uh, same applies here, like if you ask people for, you know, help or thoughts in your job, you know, they might, oh, gee, I actually have a place that you might want to apply to, or I can call a friend and see if they've got an opening. So, you know, you can, you can do those kind of things. I think being involved in online and offline communities is a great way because recruiters know that too and they're looking in those communities. So this has been happening forever. I started, for example, the first .NET programmers community back in the early 2000s really as a way to find talent. And so we'd buy pizza and we'd host 30 or 40 people a month to talk about programming topics. And when you found the smart ones, you know, when you had a job opening, you knew who to call. So going to those things and meeting people can really, you know, be there. And then it really is just about patience and hard work. And, you know, count yourself as applying to a real job where you've done the research, where you've sent a personalized message saying something real about the company, saying something real about the job, ideally even getting someone that you may know there. Like those, if you notch those repeatedly for a period of time, you'll get much more progress on a job than just like spray and pray. big plus one for me to a couple of things in particular. One, the idea of putting the legwork in to meet people at the companies you want to work in. This has been something that's been really successful for me. When I was a junior, senior in college, I did a whole like informational interview series with like 20 plus people across various companies to A, understand where I wanted to apply and B, meet a network of folks and had an incredible impact. The and I'll note on that front, by the way, anyone who is listening, who is an early career professional who wants to talk to me, please reach out. You can reach me on LinkedIn, on Twitter, through the Substack app. But if you're a listener and you would like to chat, I would love to talk to you and see if I can help push you to your next role. Super thankful for all of you listening. And I think it's a really important thing to do to give back. And anyone who's listening, who is an executive or is a leader here or is farther in their career. I think almost all of us have had this experience where someone we knew, whether it's someone we worked with or someone we'd met at a conference or whatever else, has been impactful for us around finding a next role, finding a next promotion. And so it's so important to give back when you have the opportunity. And this obviously creates reciprocal positive network effects. The other thing that you mentioned that I love is this idea of joining communities and building in public. Open source software is having an absolute renaissance. You can code more than ever. You can buy a cloud code subscription or codex or whatever your favorite is. You can use open code. And there is an opportunity for you to build in public and gain recognition for it. And even something as simple as like a cloud code skill I developed that now has like 1200 GitHub stars is out there. And like, you know, that doesn't matter to me today. I'm not looking for a new job. But it's a great signal of, oh, someone is someone who's building. There's a reason I did this podcast. You know, part of it's that I enjoy it, but do I also think about the long-term benefits of knowing people in the industry, being public about my work? Absolutely. And so I highly encourage anyone who is thinking long-term about their search to, or even short-term as well, to say, maybe you should write a blog post about something you're building. Maybe you should, you know, post to LinkedIn like this, I built this thing on GitHub. You should go chat with this person. I know it's higher effort, but these things pay off and they compound long-term, too. Yeah, and I want to highlight something in particular that you said around, you know, the opportunity to build something and put it out there. You know, I think one underappreciated dynamic of the fact that these technologies are evolving so quickly that can be really powerful if you're a job seeker is the fact that when there's a new technology, it's new to everyone. And so if you're on the job market today, you can kind of have as much experience as almost anyone in the world at building cloud code skills and whatever they're going to ship next week. And so, yeah, it's hard to be the world's most experienced, you know, at some technology that's been out for 10 or 15 years. And so I think, you know, people worry, oh, geez, what am I going to do as an entry-level person? I would argue that you have a real opportunity. particularly in today's market where the technology is changing literally every month, to be a first mover and to develop real expertise at something that very few people can and showcase it. I think that is by far the best opportunity to stay current, to develop skills, and then ultimately, Conrad, for your point, to develop a bit of a public profile and some reputation that can help you in innumerable ways, including in getting a job. I love this mindset reframe because I know it can feel really scary right now with how fast change is happening. And look, we've all heard a million executives talk about the importance for growth mindset to long-term and, you know, it can feel overblown. But in this moment, I cannot recommend enough leaning in and building and trying things because as Daniel's saying here, there is an uncalculable level of opportunity and there is an extremely level playing field. If you get a 5x Codex subscription or a 5x Cloud Code subscription and you run a couple parallel agent sessions, you can get a lot done. Even if you're not using a virtual machine with a million sessions that are running on your repos in parallel without you thinking about it. There is so much you can do with these tools. There's so much you can learn. There's a million different learning opportunities and events happening here. And if you lean in now, it is going to pay off massively in dividends in your career. And I would actually argue, if you fail to lean in, and you abdicate the responsibility for growing a career, that you are at major risk the next couple of years as well. I think that's right. Look, I think, you know, being old doesn't have a lot of benefits, but one of them is it has a benefit as a perspective. And so when I graduated college in the mid-90s, software development was expensive. I mean, to buy a compiler or an IDE, which was kind of new at the time, could cost you $1,000 or $900 in 1995 dollars, you know, just to buy the tools to write code, to develop and ship it, to get it to customers. You literally had to burn it onto physical media and deliver it to them. There was no way to showcase what you had done to anyone, so no one would ever find out about it. And when you think about the analog today that you can start building free software anytime that you can immediately deploy it worldwide for free. And then you can create your own YouTube channel or go on TikTok and showcase what you've done, write a story about it on LinkedIn. Like the possibilities to create value and to communicate broadly, like it's hard to realize how new and how powerful that is, but I can tell you how different it is than it has been. And I think still not as widely appreciated as it could be for what it means to getting a job and to having impact as a young person. And this narrative that we're talking about here, this area of opportunity, this era of opportunity, I think it pushes back at some data that I saw Greenhouse put out, which is that 62% of Gen Z entry-level workers have lost trust in the hiring process entirely. And meanwhile, you're seeing on the opposite end, your 2025 survey found that 70% of hiring managers do trust AI to make faster, better hiring decisions. This can feel scary, but as you put it there, there is this opportunity to lean in. What do you make of all this back and forth and how candidates should be approaching it? Yeah, again, I think it's wise to be aware of the downsides and the risks of AI. So we started this conversation talking about the AI doom loop and sort of the systematic problems that can happen when spam and AI slop get out of control. But I think people also should, you know, and are very cognizant of the fact that AI can represent and magnify existing biases. that AI can make unexplainable decisions and invent, you know, hallucinations. Like, those things are real to be aware of. But I think if the thinking stops there, it's a huge missed opportunity. And I think instead, the mindset of, given that those are possible risks, how can we design systems to overcome them? And actually then thinking from scratch, what are the possibilities to make things entirely better? and to not be beholden to the way things used to be. And so, you know, I'll give you an example. I think, you know, hiring kind of just works how it always has, where people started the process with a resume and, you know, it used to be printed out on paper. And now you've taken that piece of paper and you've put it up on the computer screen, the same rectangle, your name at the top, a bunch of job listings. And job ads used to be in the newspaper. you know, help wanted. And now they're, you know, up on Indeed. But it's kind of the same thing, job title, a couple words, click here to apply. It's kind of the same thing it always has been. And I think the opportunity in AI today is not, how do I make a so-called faster horse and buggy? But can we stop and rethink hiring itself? And can we reimagine what hiring could become in the AI era? Why do we have job applications? Why do we have resumes at all? And I think if you're willing to think about those questions, from my perspective, I think we see huge opportunities to not just guard against biases or protect ourselves against the risks, but actually sometimes completely transcend them altogether. Like, there's a well-known bias in job application response rates. There was a study out of University of Chicago decades ago that showed a randomized sample of sending out resumes with different names at the top and who got a callback. a tremendous amount of bias if the name at the top was stereotypically black or white, or stereotypically male or female. And so we know that that bias exists in people. And if we're using AI systems to replicate that behavior, it'll probably exist in people as well, in AI as well. But what if we don't have job applications at all? What if that's not a thing? What if instead, if you're interested in the job, talk to the AI, have a half hour conversation, answer some questions, show that you've got the skills, and then you can almost define that whole category of problem completely away. So I do think that if you think about the possibilities of AI and you have the underpinnings underlying it of a solid structured hiring process with good science and good thinking behind fairness and appropriateness, I think there's a ton of possibility to make the process work better for companies, feel more human, restore trust. And so I'm really hopeful that in the next era, when AI is really dominant in hiring, I think we're going to see a huge improvement. I love this reframe, Daniel. And I think your perspective here about here's a problem, here's how this becomes an opportunity is so fantastic. And it brings to mind to me something else we've talked about in the show quite a bit, which is the idea that, you know, we have unleashed this massive throughput example or throughput improvement for AI coding. And yet many folks see that and go, oh, God, but, you know, code review is now this huge blocker. You know, we're having these huge systemic problems. We need to rethink these processes much more thoroughly than simply, okay, we have now improved code generation or resume generation to this rapid rate. You know, we can apply more, we can put more code out there. We have to more radically rethink the systems that we built. There is a reason for that. It's because the tools we have available to us are a step change different. They are, they are digital employees that we can put to work for us. So, you know, I wrote an essay about this back in 2024, and it's only increased. You can look at, you know, we talked a bit earlier about some of these AI layoff decisions. And I think one of the most prominent ones was Block's decision to cut, what, 40%, 50% of their staff. And this, you know, I wrote a very long form piece around this, and it's in the Chain of Thought sub stack at newsletter.chainofthought.show for anyone who wants to check it out. But there is a tension here between overhiring by some companies, by, you know, stuck hiring processes, by the desire to accelerate every role with AI. And then some companies, and this is what Block says they're doing, and it'll be an interesting test case to see how true it is over time, are reimagining how their org structures work, are trying to be much more horizontal. You know, you can look at, I believe it's Spotify, are saying, hey, we're actually not hiring new roles unless you first prove you can't automate that role. And I might be mixing it up with Shopify because I do it all the time, but it's one of the two. Um, and we're just, we're seeing some companies and. Begin to reimagine what their whole structure looks like, what the organization looks like. Uh, I was just at Salesforce TDX and they are now open sourcing a ton of your platform and this is Salesforce. They never open source stuff. This is like an enterprise company that, you know, they're very closed moat and they're, they're realizing, no, we have to drive developer adoption to, to have a longterm future here and to, to win the longterm. That's why they're pushing so much of agent force open source. I think we all need to be confronting this reality that our systems may not work in an AI future. We need to reevaluate them, not just today, but in six months and another year. And so what I want to ask you, Daniel, is how have you reimagined the hiring process at Greenhouse itself? And how are you thinking you're going to have to reimagine it again in two, three, five years? Yeah, it's a great question. And, you know, I totally agree with the premise that, you know, these are a step change technology means we have the opportunity and also the challenge to really reimagine so much of what work even means, what a job even involves. And can this work be done by a human or a bot? all these kinds of things and how an organization works and creates value are challenges that everyone, including me, are grappling with every day. For us at Greenhouse, how it showed up in how we hire, I would say the first thing we did was we realized over a year ago that there was this real misalignment between what the expectations and what the possibilities were of using AI in a job interview. Candidates, simply put, didn't know if it was okay or not to use AI. And so one of the very first things we did that got noticed in terms of changing our interview and how we hire was simply publishing, here are our expectations and rules of the road for what we expect of you as a job seeker to using AI when it's okay, when it's not okay. And, you know, I think that was appreciated by job seekers who, again, you know, they're using AI in their day jobs. We're expecting them to use AI if we hire them. And so the message of don't use AI in the job interview process is confusing. But at the same time, I need to assess you. And so putting down on paper, like, what is and isn't okay, I think was a really needed step. Now, you asked also, how is it going to change in the future? The societal rules of what's okay and the technology rules of what's possible are changing rapidly. So I am sure that advice will not survive a five-year time horizon. We're going to have to continually revisit what's expected, what's okay, and what's not. Beyond that, I think what we've started to realize is some of the modes of assessment and interviewing that we've used historically don't survive contact with an AI world. So classically, you know, Greenhouse followed sort of like proven science-backed best practices in hiring, which includes a lot of show your work. And so we would give lots of case studies and take-home work for you to spend time and thoughtfulness on to show us what you're capable of. And I think what we've come to see is nowadays, increasingly, people, and again, not blaming anyone, but people are using AI to create those. And so I give you a take-home assignment, I say, here's a scenario, you know, create a case study. You're not really understanding what that person's capable of. And so we've, in a lot of cases, upgraded that assessment to be a lot more of a live session. And in some cases, those live sessions can very much involve using AI. Like, I want to see how you use, you know, the AI tools that you use, you know, that you're going to use on the job when I'm interviewing. It's an important part of assessing your fluency. But shifting from more of a take-home to more of a live interaction has been a better way to assess in an AI world. And then lastly, as I'll say, just the use of automated kind of AI-led interviewing I think is on the rise. We're going to be using more and more as our customers. And as I said, I think it's a great way to open up the top of the funnel. I think it's a great way to give more people a shot. And at the same time, I think it's a much more efficient way for the company to put its time in spending our humans' time being more human. I think an important point, though, that you are making here is that transparency and communication about these processes matter. If you simply throw someone into an AI interview without awareness or you throw them into a coding interview without telling them what tools to use, you are not setting the candidate up for success and you're not setting your process up for success. That's right. And you know, I've always thought that one of the hardest things about hiring is separating out who's good at interviewing from who's good at the job. 100%. And like, it sounds simple, but really, you know, that's the core of what interviewing really is there to do. And it works both ways. And so it also means you got to separate out who's bad at interviewing, or who's bad at the job, because after all, I don't care if you're bad at interviewing, if you're great at the job, I want to find those people as well. And so if you do your best to set those job seekers up for success, I want to see your best work, I want to find out what you're capable of. And job seekers are going to be nervous, they're going to be self-conscious, they're not doing this as often as you may be, you know, interviewing people. Like, it is an imbalanced dynamic that is really smart to be aware of because, you know, you're going to miss out on people. And again, like, I'm an engineer, you know, the old saying about how can you tell an outgoing engineer they look at your feet when they're talking? Like, It's true. We're not always known for our people skills. And so breaking through that and getting at who is this person I'm meeting with and are they capable of doing this job is difficult. But I think setting them up, giving them the rules of the road, being transparent, it's not just about being friendly or being touchy-feely, but it's really scientifically the best way to find talent. I want to ask about engineering hiring specifically here, because I think many engineers who have been applying at top companies have been churning through leak code problems for ages, and that's changing. We all used to use Stack Overflow in that. Sorry, Stack Overflow, if you're listening. That's changed. You can listen to my recent episode with Anoush Elengovan at AMD, CVP of AI there. And former software engineer, didn't code a lot for years. is now coding more than ever has 10 plus agents and running in parallel for him across different repos. And yet he's not touching a code editor. I'm not really touching a co-editor. I'm doing like little changes every once in a while, but like mostly I'm telling Cloud Code what to do. I'm either verbalizing to it a longer prompt or I'm just typing in a quick, hey, go do this. I'm throwing some notes in there. I'm feeding it my note system. I'm saying, hey, what are the tasks you need to go accomplish? How does a coding interview look in a world where I'm not coding by hand anymore? It's different. I mean, and again, these are evolutions that have been happening for many decades. Again, a lot of those like lead code interviews came out of an earlier mentality, which was about, you know, I need to understand your sort of fluency with core algorithms and data structures, like real comp sci stuff, because the generation earlier that I was familiar with that was writing software, that's what we had to do. Like, we had to write data structures to make software. We had to write link lists. We had to traverse trees. We had to do those things. And so we thought, oh, the way to find a good software engineer is, like, ask them all these puzzling questions about how to, you know, reverse the letters in a word or whatever that was. And, you know, already by then, even in the early 2000s, like, that wasn't really the job anymore. But, you know, people thought of it that way. And I think now, the entire, as you pointed out, the entire job is different. You're not sitting down and, you know, opening up a new thing and saying main parentheses, semicolon. Like, it's not how code is written anymore. And so, I think the job of a hiring team is to break apart, like, what we would say in my world, competencies. Like, what are the things that the person needs to be able to do well in order to succeed at this job? It probably ain't typing, and it definitely ain't linked lists. And then how are you going to know? Like, what's the test you would give someone to know if they're good at those competencies, if they could do those things? And it's just way different than it was certainly a year ago, but even six months ago. And so I think we're seeing a big change in the job. And so you're seeing a big change in what's needed to evaluate it. And, you know, how you evaluate those things just has to evolve with what's the job. We've talked a lot about the changes happening. And you've, I think, very kindly shared much of your playbook and thought process around it. What are your predictions for what's coming? Because, you know, Greenhouse has a ton of data. You're seeing all these insights. You have this great survey that just came out about AI hiring. What's next though? What are you predicting is actually going to happen? Because as much as we want to believe in this future where we all figured this out, it's going to be great. And you know, everyone's going to get hired for the right job for them. And we're going to find the right candidates on the hiring side. It's probably going to be a little messier than that. Yeah, what's the old saying? Predictions are hard, especially about the future. I hesitate to make predictions about AI because it's just evolving so rapidly. I don't know that anyone predicted that AI would invent its own religion, but it happened. So, you know, I can't say for sure by the time this podcast goes live, let alone, you know, six months from now when someone's listening to it, you know, the underlying landscape is just moving so fast. But here's what I can say is, you know, I think a lot of people are coming to the realization that the hiring process is not working and that the way AI has been deployed in a lot of ways is making it worse, not better. That has to change. Like, it can't continue. And from my standpoint, I can tell you within our customer base, And within the jobs that they're posting for, it absolutely will change. It will become both more AI, more AI permeated, let's say, more AI powered in the process of applying for a job, in the process of qualifying and being interviewed for jobs, even in the process of getting and doing a job. And job seekers are going to bring their own agentic tools to the process and help them in innumerable ways in getting jobs. And I think the opportunity that we have is doing that in ways that make it more human and more trusted. To make that more concrete, I think you should expect that you're going to have more conversations with AI, you know, as you go into the job market than you have in the past. I think you should expect the people that you come in contact with to have a lot of information, a lot of data that they can bring because every conversation is now recorded and analyzed. And so there'll be much more data-informed conversations. And I think you should expect that, like, we're going to see better kinds of proof on both sides. People are getting scammed by fake job postings. You should expect to be able to see some proof that this job posting is a real company that's not going to scam me. And you should expect to show proof that you're a real person and that you are who you say you are. And so I think if we can bring those pieces to the table, ultimately we can break apart this dune loop and make the process more efficient and take advantage of the power and potential of AI while removing or mitigating a lot of the risks and problems. Daniel, thank you for a fascinating conversation. It's been a delight having you on the show, and I think there's a lot of great insights in here that will be valuable both for job seekers and for those looking to hire. Let's close on that note. For people listening who want to get better at hiring or who are looking for a job right now and want to get better at finding one, where should they start? We've got tons of information on our website, so going to greenhouse.com is a great place to start. If you're a job seeker, you can sign up for a profile at mygreenhouse.com, create a profile and start searching for and tracking jobs today and using our tools to help you do that better. We've also published tons of information on our YouTube channel and on our blog that give great advice and insights about what's happening in the job market, so definitely check that out as well. amazing. And listeners, while you're at it, maybe just make sure you're subscribed to our newsletter at newsletter.ginathought.show to get this episode and all the insights in it in your inbox when it comes out and future episodes as well, because hopefully we're gonna be sharing a lot more great data from Greenhouse and from other companies. Daniel, thank you for taking us through this journey of how engineers, leaders, and everybody else should be understanding today's job market and approaching it. I hope that the advice you have given here will be valuable and the data is going to guide us in the right direction. And again, for anyone who was listening and is looking for a new role, if I can help in any way, please reach out. Would love to hear from you. Would love to be helpful if I can. And Daniel, again, thank you so much for coming on the show. Thanks so much, Connor. It's been a pleasure. It's a great topic.