Ep 104: Will AI Replace Coders? Terminal Co-Founder & CEO Dylan Serota

0:00How much is AI making software coders better? Is it more commoditizing low end? What's happening here? So one of the things I do every month, host a dinner in a different city around the US with some of the top CTOs or up and coming VPEs. And I asked that question to them. VPEs, VP of engineering. VP of engineering. And I asked that question to them. Do you think AI is going to replace your team or part of your team? And across hundreds of CTOs, the answer has been...

0:32Everybody who works with me knows how much I talk about the importance of talent in the innovation world. AI is changing how talent works. It's changing what's possible for recruiting talent. The frameworks of talent are very different than they were even a few years ago. Dylan Sirota is the founder of Terminal, which is helping a lot of top technology companies build their talent teams and manage their talent teams around the world. Excited for you to hear from him. I'm Joe Lonsdale. Welcome to American Optimist. Today we have our friend Dylan Sirota here with us from Terminal. Dylan, thanks for being here.

0:59Thank you, Joe. Pleasure to be here. You know, before we begin, tell us a little bit about yourself. What's your background? So my background prior to starting Terminal, I spent seven years at a company called Eventbrite. Kind of did a tour of duty across sales, product, strategy, last role, helped build teams globally, Latin America, Canada, Europe. You did that with Eventbrite, where the hearts is, right? Exactly. Good people. Kevin and Julia. Yeah. Had a great time, left right before the IPO, really with inspiration to build a company. and saw a big opportunity around global talent. And that's what led me to eventually start Terminal.

1:31We're going to dive into Terminal more later, but just at a high level, what does Terminal do? So Terminal is a global talent platform for hiring developers. So we're a managed marketplace of talent, the best companies, the best engineers coming together to solve great problems. And so we're elevating the talent pools from around the world and helping companies build global teams. So global talent, tell us a bit about state of engineering talent today and hiring today. Like what are the major trend lines? definitely becoming more global. There was a while where you couldn't hire people very easily at all and you had to go global.

2:01What's going on right now? Prior to COVID, there was definitely still the preconceived notion that you had to build a company and a team all around one kind of 30 mile radius office. I saw that breakdown when we were building teams at Eventbrite, just the strength of global talent pools really emerging. And that was really accelerated by COVID. But the challenge we still face in the U.S. is the deficit of technical talent relative to how much demand is growing. And I think the U.S. has been the innovation hub. We're increasing demand, increasing capital investment. We're seeing all different types of industry move into technology.

2:35And so you have a demand curve that is at an extreme slope. Meanwhile, we're not keeping pace with the amount of supply. So inevitably, we're going to see, and we have seen demand start to move to different global talent pools. So let's talk about the global talent first. And I want to talk about the supply question as well, because I think that's changing a little bit, at least right now. But on the global side, so I've seen a lot of talent. Originally, there was a lot of talent in China was number two. I don't really like to work with them as much right now, given the CCP issues. I've seen a lot of talent in India.

3:01I used to have kind of really back office, kind of simple things there. Now I have really advanced talent there. But I mean, I have companies hiring top talent in Scotland, in Israel, in Eastern Europe, in obviously certain parts of Latin America. Like where are you seeing the very top talent and where are you focusing? It depends on what type of talent you're looking for. But broadly, software talent is emerging in every part of the globe. Fact is, today, most engineers can do a lot of self-directed learning. Used to be that a lot of the technical skills you need was in the classroom. And today, with open source, with the ability to really learn independently, software engineers are popping up all around the world if people are hungry and if people are self-directed around their learning.

3:42So if you look at just major population centers, there's top talent in those markets. And it's compounded by the fact that a lot of the best U.S. companies have had to go to different markets. And so not only is the talent there, but then they're grooming them. So you've got Google offices all over, you've got Amazon offices all over. And so that is taking kind of the great potential of a talent pool, training them. And now they're really, there's a lot of different markets that are coming to the forefront with top talent. That's interesting. I actually noticed that New York became a lot stronger market to hire and after Google had been there for a while, 20 years ago.

4:12So you've seen that similarly around the world. I hadn't thought of that. Same thing in Canada. And then you have different homegrown companies, Roppy in Latin America. You've got Shopify in Canada, and they're also doing a good job of training up the talent in these markets. I want to push back on the demand side a little bit. It's definitely been the case that there's a lot more hiring. There's a lot more. It's a lot harder to find top engineers. I'm hearing right now, the last year or two, that a lot of new grads are really struggling to find jobs. I'm hearing that AI has made engineering in some cases, whether it's 20 or 30 or 40%, has made it more productive.

4:46So there's been a little bit of a pushback where companies hired a ton and then all of a sudden they're not hiring as much. Like, doesn't this go back and forth where some years there's like people are really struggling to hire and other years there's too many people on the market? How do you think about that? So historically, there's been about four open jobs for every one back-end engineer in the U.S. So there are more open jobs than there are available talent. The impact that we've seen in the startup recession over the last two years has had some impact, but it has definitely been outpaced by the fact that you have all of these traditional industries also moving into software development.

5:17So you talked about India used to be a place of kind of low-level outsourcing. That's exactly what a lot of the big financial institutions used to do. JP Morgan's a good example. He used to outsource a lot of their technology. Today, they're hiring thousands of engineers in the U.S. You have companies like Nike hiring a thousand engineers, CVS, thousands of engineers. So non-traditional tech has moved in to really drive up that demand and replace a lot of maybe what we've seen the challenge in the core tech market today. What is Nike doing with thousands of engineers? You tell me. Selling shoes, selling apparel, building websites, different consumer apps.

5:51Different data, I guess. That's interesting. I agree. A lot of people do need engineers they didn't need before. But I guess it's just shocking some of these companies that are hiring so many engineers. It's not what you'd expect. No, I mean, a great client of ours sells rugs, and they have one of the most advanced data analytics teams out there. Wow. When you build teams, I'm curious, because for me, like when I do a startup, I'm traditional, I'd like to be together and iterate, and then you often do have a second office. But usually for me, it's always been like a second office where you have a culture there, you have people come there.

6:20I haven't found that it works as well for things I'm doing, to have people just everywhere. Like, how do you think about this? Do you sometimes build hubs for people, like in one place? We do. Everything from completely distributed hiring, which some companies have a preference for, down to building a co-located team in a market. And I think the different discussions around remote, hybrid, distributed, to me, there's one answer, which is you're going to need global talent. The question is, how are you going to access that? And what is the tradeoff you're going to make? Do you want to be captive in certain markets and leverage your influence in those areas to build teams?

6:54or do you want to vastly expand your supply pool to reach a bunch of different talent? I think there's many different ways to build companies. When you guys are screening talent and you're looking for people, I'm also curious. So I remember back when I was in college, computer science had just become hot in the late 90s and then it kind of crashed in 2001 when I was there because the bubble went away. And I feel like something happened similarly recently where I think the number of computer science graduates rose 140 % the last few years versus a decade ago. So obviously it became really hot. one thing I've noticed is there's some people, I guess I hope to include myself amongst this originally, were just really interested in computer science.

7:29And you just like, I really like math and I really like figuring out how systems and algorithms work. And it's something that I thought was fascinating. And there's other people who are like, well, man, this stuff's not really what I want to be doing, but I want to get a job. And it feels like maybe that shows up when you're hiring people a little bit. Like how do you differentiate between people who are passionate about this versus people who went into a high area? Does that matter when you're selecting candidates? I think it always happens when you see kind of an expansion in a certain sector.

7:54You know, I graduated from Penn, Philadelphia. At that time, all my friends went to New York to do finance. I was kind of the one odd cat coming out to San Francisco to do technology. Then you fast forward a few years when things started really expanding and a lot of people flooded out here. So I think you're always going to see kind of that, that urge for people to kind of follow where the, where the puck is going and try to pursue those. What we look at is a lot of who are they, How have they actually, how have they developed their own skills? And oftentimes it's, what are they doing on the side?

8:24Are they contributing to open source projects? Do they have side projects? Have they built companies or have they built products before? So a lot of the way that we're looking at screening them is being able to actually dive into their history of what they've done and then do some personality assessment on what are their ambitions? You know, is it just about money or are they actually someone who wants to be mission driven? And that's a big aspect of our screening. Tell me, how does the AI affect this? Like, did you, first of all, do you use AI in your screening at all? Like what's the, what's AI doing for you to help you look at candidates?

8:53A lot of the way we're using it is I would call it marketplace operations. You know, we're, we're, we are building a global talent marketplace. And one of the hard things is getting completely differentiated supply of talent, organized and merchandise on the marketplace. Unlike, you know, a car marketplace where you have make model years, people are all different. And so we leverage our ability to screen and interview someone, capture a lot of those unique insights, organize them, tag them, and then merchandise them. So we're using it on the marketplace operation side, first and foremost. We also can do different levels of scoring because we can predict how likely is someone, based on their experience and background, going to be to pass one of our customers' interviews.

9:34And that helps us prioritize thousands of engineers up to really the cream of the crop. So you can't do make, model, and you can kind of classify it into certain types of makes and models, huh? We try our best. It's a very difficult supply pool of humans, unlike other inventory of marketplaces. And when you look at what AI is doing on the other side in terms of their jobs, how is it changing your business? Because first of all, how much is AI making software coders better? Is it more commoditizing low end? What's happening here? It's interesting. I'm definitely seeing huge productivity and efficiency gains, but I don't think the role of software engineers going anywhere anytime soon.

10:12So one of the things I do every month, host a dinner in a different city around the U.S. with some of the top CTOs or up and coming VPEs. And I ask that question to them. VPEs, VP of engineering. VP of engineering. And I ask that question to them. Do you think AI is going to replace your team or part of your team? And across hundreds of CTOs, the answer has been zero. No one thinks it's actually going to replace. People are seeing productivity gains and different shifts, but calling it more akin to having spellcheck. It's going to make you more efficient in getting the work done. It's going to allow you to abstract some of the work to actually core engineering versus writing the code itself.

10:49But it's not going to take away the need and desire, at least at the enterprise level where you're building products, serving millions of users. It might change a little bit how you're hiring. I'll push back and see what you were saying. Because for me, what I've heard, at least, and I'm not running these teams myself these days. I obviously invest in companies and I help build companies. But for a lot of our companies, it feels like the value of a really senior engineer has gone up relative to the value of a junior engineer, thanks to some of the things they could do. Is that what people are seeing or no, not in your case?

11:18I've seen that the value is coming a little bit more from the junior engineer who is having to do a lot of the task-related work. The senior engineers that are oftentimes thinking about there's 10 different ways you could build a bridge, what is the right way for us to build it? When they're thinking about those things, that is not being aided today by AI. But the junior and mid-level engineers are able to be a lot more efficient, feel like they're able to actually learn by having the AI suggest code. So I'm seeing it a little bit more of the opposite, that it's actually up-leveling a lot of the mid-level engineers.

11:51The senior engineers, and I think this is the trend, keep getting abstracted to more architectural engineering decisions versus spending so much of their time writing the actual code itself. I guess that's similar to what we're seeing with some sales guys where Michael Dell was telling me earlier that you actually can take general salespeople and make them specialized salespeople using some of these AI technology. And I guess you're seeing this as making that more efficient. So that's the analogy that you're giving here for the engineers. Exactly. Which is different than some other possibilities.

12:22Like chess in the old days, like AI, as it got better, would commoditize the tactics, commoditize certain parts. But then like the judgment parts of position play, you still needed like the very, very best players. And so it used to be that if you want to be the best at chess, you can have a person plus a computer. And you have a really good chess player who understood position well enough. And then the computer would do all the simple things. So one argument of this kind of like hybrid human computer is that you want a really wise person for some really hard decisions. Then you want the computer.

12:51That'd be the different model, but you're not seeing that one as much for coding. Not yet. And I think at least the companies that we're working with, there are so many interdependencies when it comes to the systems that they're building. So in a very small subset of code, it can help and aid in how to finish a complete and endpoint. But when you think about security, user data, different transactions, we're just not seeing the ability to go kind of towards that state of fully autonomous for the task related work yet, but it is making people just more efficient at getting the work done. So it seems that AI can make this a little more complicated.

13:28Everyone's going to run their resume through chat GPT. It's going to all look the same and really impressive and optimized for what you want. And like, how do you dig through this? Like how are you using technology or your business to help people figure it out? It's a really hard challenge. And we're seeing that today with many of our companies complaining meaning that they're getting 5 ,000 applicants. But then when they start sifting through, they realize that 95 % of them are fake. And so they're getting fake resumes, fake applications. And it puts a lot of onus on teams now to go back to more of a traditional in-person or face-to-face type of interview.

13:59During the kind of post-COVID boom of hiring, people didn't have time. And so they're sending people tech assessments and just looking at kind of online resumes. A lot of that is now breaking down and failing. And so a lot of companies are going back to actually coding with people live to make sure that they're good. There's different ways that I think we can help to solve for it with different tools. But it's kind of this cat and mouse game between what are candidates going to do to try to get in versus what can we do to really ensure that someone is who they say they are and that the quality is actually stands by in front of a resume that might be fake.

14:35I mean, that's a crazy problem that if, I guess if you wanted to screw with your competitor, you could just send them like thousands of fake resumes credibly with AI. This is like, I hadn't considered this, but it was kind of evil in a way. And then it's also really interesting. There's all these rumors of like, there's maybe a really good engineer, but he might have four jobs in Eastern Europe. And, and like, like, how do you figure out he's not doing that? Cause that's like, that's your job now is to be on the ground. Like, what do you do? That is a huge problem. And we, we see it. There's articles coming out around even different gangs or different countries that are trying to infiltrate companies by posing as engineers, getting jobs, stealing code.

15:10So we have built out a whole identity verification process and location verification. So we want to know that when someone is applying for a job and we do a background check, that the person who is interviewing and is going to do the job is who they say they are. And they're in the location that they say they are. We've seen many companies hire folks, and then all of a sudden, that engineer is logging in from China or from other markets. So it is something that I think is becoming much more prominent and is definitely something that companies have to be very wary about, but an opportunity, I think, to solve that problem with some interesting tech.

15:44It's fascinating. There's a security aspect to your business then as well. Exactly. So let's talk about Terminal. I was a little bit involved at the beginning with Jack Abraham and stuff when you were building this and obviously saw a huge need for talent. So I'm tied to it as well in some ways. Why are you passionate about building a talent platform? What are you trying to solve for? Really two things to me drive my passion for Terminal. The first is I'm a tech optimist at the end of the day. I believe that the more innovation we can push, the more of society's problems we can solve. And if we can unlock more innovation, we're going to solve a lot more problems and make everyone happier and better and live more fulfilled lives.

16:20The one gating factor to that that I see is that deficit of technical talent. Again, our demand for innovation, our demand to build is outpacing our ability to reach that supply locally. And the solve for that is raising and providing access to that global pool of talent to unleash a lot more innovation. And then secondly, firmly believe that people's lives can change with professional opportunity. And some people aren't as lucky to win the geographic lottery to be born in the U.S., But the way in which we can actually provide some of the best values of the U.S., meritocracy, innovation, capitalism, is allowing those opportunities to come to people around the world.

17:00And so being able to unleash American innovation, be able to bridge that towards providing professional, global, world-changing opportunities for people really drives me. I guess in some ways it's positive on the U.S. If you believe that we don't have enough engineers who are graduating here, we're still going to be able to build things here despite that lack of time. Absolutely. It's a bottleneck that we can unlock that's going to enable us to build better companies, faster companies. It's going to enable us to hire more salespeople, marketers. It's going to enable the U.S. to still be the innovation hub, even if we're leveraging different pools.

17:32So I think we're going to see this trend of much more of a globalized workforce. You know, I think we saw kind of globalization happen at the manufacturing low skill level. And I think we're starting to see that in a pretty quick manner at high skill level. I think engineering is one of the most effective ways of doing that globally. So there's already obviously a big outsourcing market. What's wrong with the incentives in outsourcing today and how's terminal different? Outsourcing has very perverse incentive model. And so their core value from the beginning when outsourcing companies were being developed, it was that it was nearly impossible to go into a new market, find talent, open an entity, employ people.

18:10And so what they did was they scooped up local talent. They kind of held them captive and then they would lease them out to companies. The problem is their incentive is to charge a company as most as possible and pay the people as least as possible. And that's their black box margin. And so they literally sell engineers on a rate card. So there's no agency of the engineer to decide what they work on. There's no agency for them to demand a higher value based on what they're going to contribute to the project. And so this intermediation is, I think, putting companies at a point where they're beholden to these outsourcers where they can't ever break away because it's not their team.

18:46And the engineers really don't have agency around their career because they're kind of held captive. And meanwhile, there's this black box incentive structure that I think drives the wrong incentives. So how's Terminal different? Why does it work better? We're transparent and we try to provide agency around the team. So when companies are building teams on terminal, they're building on the terminal platform, leveraging our talent, but we're not getting in the way of them controlling the team and having the agency around them. So they're negotiating salaries directly. The engineers are able to ask for what they want to make.

19:17They're providing equity and upside. They're really part of the teams. It's not that they're a terminal engineer. It's that they're part of any one of our client teams. It just so happens that we're enabling those transactions and that economic interaction to be possible. So other than finding talent itself, what do you do to manage things for them? So at the top, we're a managed marketplace. And so we are sourcing talent, vetting talent, and making sure that the best engineers from all of our markets are available to be matched with our clients' opportunities. The hard part is when you think about kind of global commerce around talent is you got to do it compliantly.

19:52You got to think about IP assignment, payments, compliance, legal benefits. And so we have different transaction types that we can enable for customers if they want to have full-time equivalents or if they want to build teams of contractors, they want to have a contract to hire model, or if they want to build their own team and their own entity. So we have different kind of transaction enablers that allow that talent that's on the marketplace to be able to cohesively join and be part of our client teams, which is something that hasn't been built before. And so people could actually try contracting maybe a few of these engineers in Columbia or whatnot that you have on your platform.

20:30And if it's going really well, they could then hire them with equity or something like that. That's the big trend right now is basically this try before you buy for talent. Go into a market, prove that it's good, very low risk when you think about contracting. Our customers tend to see that they never want these engineers to leave them. They'll convert them to being full-time, and they'll scale a large team around them. Who are some of your customers that you could say, and how do you win them over? We've been lucky to work with some of the fastest-growing tech companies. Hims and Hers, fastest-growing health tech company to an IPO.

20:57We look at companies like Chime, like Grindr, like Nextdoor, and a lot of companies I can't probably talk about. But many companies that most of your listeners, I think, would be very familiar with. behind the scenes, we're playing a key part. Anywhere from 10 to 20 % of our customers' teams are powered through terminal. What are some of the big challenges you guys are facing as you're scaling? Challenges around supply. Like I asked, like I mentioned before, you know, building a talent marketplace is unlike other types of marketplaces. It's, it's double opt-in and you've got this unique challenge around supply.

21:30There's a lot of people who are available might not be the best engineers, right? The people get laid off. You don't want to hire always those guys. You got to vet the supply and it's a human on the other side. So you've got to also make sure you're kind of fighting through a lot of what people are selling you at. And then it's also perishable. People are on the market for a certain amount of time. So there's a lot of complexity around how do you aggregate the supply, balance the liquidity for the demand that we have coming in, give both sides a really great experience and try to optimize for speed and quality.