#33 - Using AI to fix job destruction, skills, and hiring - Tigran Sloyan

In a labor market being rewired by AI, CodeSignal is betting that skills, not resumes, will decide who thrives.

For this episode of The Deep View: Conversations, I talked with Tigran Sloyan, CEO and co-founder of CodeSignal, the company building a new standard for hiring and career mobility in the age of AI.

CodeSignal’s mission starts with a simple but painful truth: resumes and interviews are a flawed way to hire talent. Countless candidates have the skills to thrive in high-paying tech roles but never get a fair shot, while others with polished credentials sometimes land jobs they’re not prepared to do. 

CodeSignal is flipping that equation with skills-based assessments that help employers discover candidates with real ability, and a free learning platform that helps candidates level up for the next opportunity.

In my conversation with Tigran, we talked about:

+ Why resumes haven’t meaningfully changed in 100 years, and why it's breaking hiring

+ How CodeSignal measures skills, and why simulation beats multiple-choice

+ What AI unlocks for assessing non-technical roles such as sales and support

+ The dark side of AI: what CodeSignal’s research shows about cheating attempts

+ Why entry-level jobs are turning into tasks, and what that means for training

+ How CodeSignal makes free learning content work economically

+ The future of re-skilling at scale, and why AI tutoring changes everything

We also dig into what’s changing fast right now: the rise of AI-assisted work, the surge in fraud in hiring assessments, and why foundational skills still matter even when AI can do the task.

Tigran shares his background from Armenia to MIT to Google, his most contrarian leadership advice, and the AI tool he'd recommend you start using every day. If you want to understand how AI is being used to fix the problems that AI is causing in the job market, this is the podcast for you.

Subscribe to The Deep View: Conversations podcast in your favorite podcast player for more unique conversations with the brightest minds solving the biggest challenges in AI. You can also subscribe on YouTube.

And don't forget to sign up for The Deep View daily newsletter. We don’t just cover AI, we decode it. In a world flooded with hype, we deliver sharp, no-nonsense insights that keep our audience ahead of the curve and help them put AI to work every day: subscribe.thedeepview.com

Creators and Guests

Host

Jason Hiner

Editor-in-Chief of The Deep View

What is The Deep View: Conversations?

From frontier labs and enterprise platforms to emerging startups reshaping entire industries, The Deep View: Conversations podcast interviews the brightest minds and the most influential leaders in AI.

Jason Hiner (00:01.922)
In this episode, I talked to Tigran Sloyan, the CEO of CodeSignal. This company has a unique mission, built around the idea that many people have the skills to excel in high-paying tech jobs, but they often don't have the resume to land those jobs. And the reverse is also true. Some candidates have resumes that land them jobs where they don't have the needed skills. So CodeSignal has built this platform that helps job candidates validate their skills. But maybe even cooler,

They've also created a learning platform where candidates who don't land the jobs they want can boost their skills to improve their chances of landing the next open job. And all the content on the platform is created by the Code Signal team and is available for free to all job candidates. I talked to Tigran about how his team makes the economics work for that and how he's created a world-class educational team. And of course, the backdrop for all of this is a job market where hundreds of millions of people

are going to need new skills and retraining in the years ahead because of the AI revolution that's hitting virtually every sector of the economy. Code Signal has also been fortunate to have a front row seat to the generative AI revolution. Since one of the co-founders of OpenAI was one of the company's first seed investors a decade ago, and Code Signal's data helped train GPT 2.5 before ChatGPT had even launched.

We also talk about Tigran's background at MIT and Google, his contrarian advice for leaders and his recommendations for AI tools. You don't want to miss that. So here it is, our interview with Tigran Sloyan of Code Signal.

Jason Hiner (00:01.641)
All right, so for my use and for our editors as well, go ahead and pronounce your full name so that we have the exact version of it.

Tigran Sloyan (00:11.308)
For sure. Tigran, slow yet.

Jason Hiner (00:14.239)
Very good, all right. Well, Tigran, I wanted to start by having you tell us a little bit about what you do and your role and code signal. Tell us a little bit about the company and what is its place in the world.

Tigran Sloyan (00:29.612)
Yeah, absolutely. And thank you for having me, Jason. This is exciting and I look forward to the conversation. Well, I run a company called CodeSignal. We are on a mission to discover and develop the skills that will shape the future. And what that means is that we play in a few different worlds. The first and the most important one is skill assessments, because everything starts with understanding human skills, because you really can't...

teach people if you don't know what they already know. And you also cannot help companies go beyond resumes and really identify humans based on their skills if you can't measure those skills. You could reshuffle resumes, you can look at it five different ways, but at the end of the day, it's poor data in, poor data out, or garbage in and garbage out as people call it. When you go directly to the source, what can humans do? And what matters currently in the...

industry when it comes to the skills that we're focused on, that's our expertise, right? Really assessing skills, doing it in a way that is very, very high signal and helping individuals also build those skills on top of being able to measure them, which is the foundation for doing everything else.

Jason Hiner (01:46.011)
good. So how long has CodeSignal been around? How long have you been doing this?

Tigran Sloyan (01:51.305)
I've been at it for about a decade. It's a long journey, but sometimes important things take a long time to build. And if you look at the overall space of skill assessments, skills development, education as a whole, there's so many companies that have been added for quite some time because it's a very, very difficult problem, an important one, which is why, you know, people walk away from a lot more lucrative spaces to actually put their energy

and life towards fixing this because humanity and human skills is the most precious resource we have. But the way we are managing it, the way we are cultivating it is completely backwards right now. But it has so many layers of complexity that I'm sure we'll get into that we've been tackling them one by one by one. And we've reached a very exciting stage of the company's life where all of these pieces are all of a sudden coming together, which is really fun.

Jason Hiner (02:51.905)
there are a few things that have happened like timing. There's a lot of things that can go right for you when you're starting a company. And there's a lot of things that can go wrong too. But timing is the one you can't control. There are a lot of controllables you can control and you're always working on that.

Tigran Sloyan (02:58.764)
Right. Yep.

Tigran Sloyan (03:06.348)
Mm-hmm.

Jason Hiner (03:07.605)
but you can't control timing and you've had the benefit of some very fortuitous timing as I see it. The pandemic and a lot more sort of the rise of remote work and hybrid work, it was always there and it had been increasing, but obviously it exploded because of COVID. And then the AI revolution were now sort of the opportunity to...

to have help with things and to show, put a good foot forward. And when you're doing a hiring assessment or something like that has just been absolutely rocket fueled to where you've got tools at your disposal that were just never there before, much more powerful than just a Google search. so tell me a little bit about that. What does that meant? That's about sort of the second half of your decade journey here. And what does that meant for your company?

Tigran Sloyan (03:49.047)
And...

Tigran Sloyan (04:00.142)
Absolutely right. It's one of those things where you can do everything right and still get the timing wrong. And by timing, really, you're talking about, I call it getting lucky because you cannot control timing, right? You just have to be in the right place at the right time. But at the same time, you have to stay alive long enough, be even attempting to do something, right? Be attempting to climb a certain mountain and then get lucky on one of the turns that has been impossible historically because

If the timing is right and you're still sitting at the foot of the mountain, well, guess what? That also doesn't work, right? So you have to be in the game to even have a chance to capitalize on that timing. You can't just like sit and wait for it. So in our case, you're absolutely right. You know, if you go back to 2023, we were doing only technical assessments and we were only in the hiring side. We always wanted to do tech and non-tech.

Jason Hiner (04:34.753)
Yeah.

Tigran Sloyan (04:57.59)
right? Technical skills and non-technical skills. And we always wanted, that was part of the mission all along, to both discover and develop the skills that will shape the future. But neither one of those two other parts were possible without generative AI. If you think about non-technical assessments, right? The way we do assessments in general is simulation based. It's very hands-on. It's not a multiple choice question. It's not a quiz because

There's a reason you do a driving test when you're getting your license, right? If we gave everybody a license from the, you you completed this quiz, you know how to drive, correct? We all know that doesn't work. So when it comes to very high caliber skill assessments, you have to put people in the driving seat and see what they can actually do. Now, in our case, you know, we could do that for technical skills without Gen. AI. Why? Because technical skills are very deterministic.

Another way to put it is there's a right and a wrong. You can say, here's a problem, right? Build a UI for an application or build this API or do this data analysis. And we know what right looks like. So we can deterministically say, you get to the ending of it? Here are some test cases that are going to validate if you haven't or haven't done it. And this is how we've built the foundation of the company in working with

tech and finance industries who hire a lot of engineers to help them go beyond resumes and find fantastic humans whose resume might not reflect what they do. But again, the dream has always been how can we bring this to the wider market because the same issue exists literally everywhere, right? People go through painstaking interviews and resume reviews and in many cases, incredible people who would have done really well at the job just

get dropped while others get hired who are simply not competent enough. They're just really good at interview processes. Then Gen.AI came along and it enabled a few different things. The most important one being the ability to create simulations of non-technical skills that are very, very real world like. So if you think about sales, right? What do salespeople typically do? They reach out to certain prospects. They

Jason Hiner (06:56.363)
Yes.

Tigran Sloyan (07:19.229)
have discovery conversations with them, right? Ask them the right questions, dig into the right areas. So skills like, you know, curiosity, active listening, probing for clarity. Then as they get later in the sales process, they might have negotiations, they might have some tough conversations, et cetera, et cetera. So if you're trying to assess whether a salesperson is competent or not, you have to simulate that entire cycle. Without Gen.AI,

we simply could not simulate that entire site. I mean, we could create the old school chatbots, which would be very like, ugh, what is this? This makes no sense. This is not our Clippy. Or we just have to wait and hope that things would work out. Well, so when I first saw the Gen.AI revolution starting, and we were at the roots of it because many of the, you know, actually one of the co-founders of OpenAI is a seed investor in a company. So we've actually

Jason Hiner (07:53.249)
Yeah.

Jason Hiner (08:15.379)
Okay.

Tigran Sloyan (08:16.013)
even helped train some of the early models of the GPT 2.5 because we have a lot of coding data from our skill assessments, anonymized coding data that we've obviously in their scale, it's a couple of drops in their bucket. But that partnership very early on in like 2021, 2022 has allowed us to have a front seat at like, oh my God, these models are getting really, really good and the possibilities are endless. So.

Jason Hiner (08:22.785)
Mmm.

Sure.

Jason Hiner (08:41.93)
Yeah.

Tigran Sloyan (08:45.281)
When we saw that, we started building those very early versions of like, we can start to simulate real email exchanges. We can start to simulate real voice conversations where it feels like, you know, in reality, you're talking to an AI agent, but it feels like you're talking to a real prospect or a real customer. And then on top of that, simulation is one piece. You have to be able to grade the outcome because you don't want people just sitting there and manually going through thousands of these simulations and saying, yes, no, yes, no.

And GNI has allowed us to create very high quality rubrics as well as grade them automatically to say, these are the folks that have done well, they have the skills and they have the capability to be successful. And these are the ones that haven't. And that's just a one layer of it, right? To say that GNI has really transformed our ability to realize this mission would be an understatement.

Jason Hiner (09:33.537)
Okay.

Jason Hiner (09:41.121)
I can only imagine. can only imagine. All right. You've also released some new research, Code Signal has, but before we, and I want to get to that, but before we do, I want to talk a little bit about your journey to start Code Signal, because you are the CEO, you're a co-founder. Where did it begin? And when you started the company, like where were you? And then how did you come to the...

the point where you founded and started this whole thing.

Tigran Sloyan (10:12.949)
The question is how far back do you want to go? But I'll start from the right before. I'll start from the right before and we'll take it wherever you want to go. So before Code Signal, was at Google and we were doing a lot of hiring. This is like early 2010s, right? So we're doing a lot of hiring and Google was growing like crazy. And I was interviewing a lot of people and I started realizing a few things that were kind of shocking for me. The first one was that...

Jason Hiner (10:15.775)
Okay, okay

Tigran Sloyan (10:41.805)
I had a lot of friends who went to competitions, programming and math competitions with me because that was my background. And I was like, these people are incredible. But I ended up going to a top tier school. I ended up going to MIT and we can get into the lucky circumstances, speaking of luck, that have brought that about.

Jason Hiner (11:02.305)
Sure.

Tigran Sloyan (11:03.661)
Many of them haven't, so they didn't have anything on their resume that really stood out that a recruiter was recognized. But I was like, I'm going to refer you, right? Like you're incredible and really looking for a great talent. I would refer that and they would never hear back. And then I'll go to the recruiters and be like, oh, we looked at the resume. It was like average. So we just moved on. And I was like, you're kidding, right? Like these people are one of the best engineers in the world. And you're telling me that the resume said, uh,

They went to an average college and you just passed. So that was like one of the core realizations that like, wow. Like even when people do get lucky enough and the timing and they actually get to build skills that are transformational for themselves and for the world, you really don't get noticed until I see how the resume. And I started to realize that part of the reason why I was there, even though I believed I was there because I worked hard.

Jason Hiner (11:36.969)
Yeah.

Tigran Sloyan (12:02.252)
I was smart, I was capable. Like, oh, shoot. If not for the resume, if you didn't say MIT, I wouldn't be, potentially wouldn't be sitting here because my resume will just get lost in that same shuffle. On the flip side of it, the way I got to MIT, the way I got to where I was, was many, lucky circumstances. I grew up in a small country called Armenia where I never haven't even heard of MIT, right? And the way I heard of it is,

Jason Hiner (12:29.259)
Sure.

Tigran Sloyan (12:31.146)
I got into mouth competitions at a pretty young age, even though I didn't like school and I was an average student, mouth competitions really sparked something in me because I'm competitive, disagreeable, like many founders are, right? Like you just don't agree with the way things are. So that is what you have to be, but that doesn't work well in school, right? Because in schools, traditional schools, they're designed, you've got to follow rules, you've got to do what you're told, you've got to, you know, just sit and be quiet.

Jason Hiner (12:42.785)
Okay.

You

Jason Hiner (12:53.547)
Sure. yeah.

Jason Hiner (13:01.119)
Yeah. The ones who do what they're told the best do the best, right? Have the best grades.

Tigran Sloyan (13:01.13)
Those are the attributes that actually get you the A's.

Tigran Sloyan (13:08.268)
Exactly. So for me, like that reflection on my past and realizing that if not for the math competitions where I accidentally some kid, some random kid from Bulgaria was like, I'm going to MIT next year. And I was like, what's MIT? And he was like, which, which planet are you from? But he grabbed a sticky note, right? He was like MIT.edu go apply. they love nerdy international math kids. So like I went applied, got in.

Jason Hiner (13:28.219)
Hahaha

Tigran Sloyan (13:38.049)
got a full scholarship and moved to Boston. But if you look at that journey, so many people that I grew up with never even heard of it. Never were there, never in that moment, never had a chance to even build up their skills. But then even those who did, if you never managed to get it on your resume, you're still getting blocked behind it and not getting that opportunity.

Jason Hiner (13:47.713)
Yeah.

Tigran Sloyan (14:04.712)
In short, that realization that like, A, most people never get a chance, never get lucky. just get, the, your teacher says you're never going to be good at this and then you buy it. Especially if you tell that to a middle school, which often happens, even those who do, unless it's reflected on their resume, you might still not get a chance to do what you want. And then third, which is what we started with, you know, talent is the most precious resource humanity has because everything that we're

is around us right now has been created by skilled and talented humans. This application that we're currently using, the microphone, the electricity, the computers, right? Without human skills and talent, none of this would be here. So that realization that, you know, this is one of the most important things is the meta level problem that transcends all other problems. And how broken the current system is was what gave me the fire to say, you know what?

I it's gonna be hard. I didn't know it was gonna be this hard, to be fair. I like, I know it's gonna be hard, right?

Jason Hiner (15:07.649)
Yes, as is often the case. We wouldn't do it if we knew it was as hard as it was.

Tigran Sloyan (15:15.28)
Exactly. Exactly. I always say that you've got to be naive enough to believe you can do it to attempt very hard things. And I wouldn't change anything about it, right? People sometimes say, what would you say if you went back to your, you know, 10 years ago self? I'm like, I wouldn't say how hard it was because I knew he would chicken out and not try it. I'll be like, go for it. It's all going to work out. But anyways, that combination is what sparked the fire of, you know, what were

We're going to go figure out how we're going to discover and develop the skills that are going to shape the future and how we're going to create a new standard for resumes. Because current standard for resumes is LinkedIn profiles. And that is very broken and it hasn't changed for many, many years. LinkedIn is only 30 years old, maybe 25, but resumes came about a hundred years or so ago. Some people say Leonardo da Vinci wrote the first resume, not really.

But in reality, resumes came about in early 20th century when the industrial revolution and people were trying to figure out how to represent themselves in white collar jobs. In the beginning, they were written in a piece of paper. Then we started writing them in a PDF. Then we put them into LinkedIn profiles. So the content hasn't changed for like a hundred years. The medium in which we put it has changed.

And these days, when AI is transforming the job market, the skills that matter, the educationals, like everything is changing, relying on a piece of paper and some keywords on it that describe your skills and capabilities and what you can do is just more broken than ever before. And I believe, speaking of timing, timing is right to create a new type of standard that is all about skills.

and not about where you've been or what you've done.

Jason Hiner (17:12.993)
Boy, I love that there's so much to unpack. I would love to talk more about some of those things. And I think we should talk a little bit about the research that your team has announced that I think will help us unpack some of these things a little bit further as well. So you've announced some some new research that that finds a few things that are interesting, maybe a little disturbing.

Tigran Sloyan (17:28.876)
Mm-hmm. Mm-hmm.

Jason Hiner (17:40.865)
but also get to some of the larger problems of this whole hiring moment that we're in with the technologies we have and with the things that really aren't working for many people. And so I'm gonna let you describe the research that your team has released.

Tigran Sloyan (18:00.649)
Absolutely. So the new research that came out is about cheating and fraud, right? That the dark side, I guess, if you put it, of generative AI for us and for the hiring industry in general has been that cheating and fraud has gone through the roof. I mean, that's always been part of hiring, right? Like there's, it's nothing new.

Jason Hiner (18:10.911)
Yeah. Yeah.

Tigran Sloyan (18:24.671)
But historically, let's say you had an assessment and if you wanted to kind of find a way around it, you would have to go find somebody who knew enough to help you out or you would have to like go searching on the internet for many hours. Now you have intelligence at your fingertips. Now, when we talk about this, I want to start off by saying we in no way are saying that, you you should not use AI at work. It's the opposite.

I'm the number one believer that the diffused state of humans working with AI, delegating to AI to get things done is where the future lies. But many organizations are still in that transition phase where, you know, yes, AI is there, but many of them haven't introduced it in their workflows and in their work, or they're just aiming to assess your foundational skills, not your ability to collaborate with AI. Because believe it or not, you still need to understand.

what it is that you're asking AI and be able to do it yourself. Right. I often compare it to being an effective manager. We've all at some point interacted with managers who didn't know the work, couldn't do the work themselves. And we know they're not competent. Right. If you cannot understand, if you can't do it yourself, just being able to ask somebody else to do it without the ability to understand the nuance or give feedback or be even clear in what you're asking.

only comes from when you can do it yourself. So in many cases, our partners and our customers choose to prohibit the use of AI in these assessments. Now, when that happens, it's on us to keep a fair bar, right? Because if the rules say you cannot use it and a certain percent of people do use it, they gain an unfair advantage. And now you have people who are making it through the hiring process who shouldn't be making, and you have people who...

stayed true and stayed fair and stayed honest who are getting dinged because of that, because other people found a way to game the system. So we take it very seriously to ensure that whatever rules exist, if the rule says use it, we'll make sure that you have access to it and all the opportunity to use it. If the rule says you're not allowed to use it in this assessment, then it's on us to ensure that that happens as well. Now, like I said, with the rise of Gen.AI,

Tigran Sloyan (20:47.189)
people attempting to do it has gotten almost exponentially, almost three times as many people attempt cheating and fraud as they used to before just because it's gotten so easy. And our goal has been with the transparency, part of the reason we're releasing the research is to say, don't do it because A, you can't game the system. We see it, we do this, you do it, you try to do it once. We do this every day, thousands of times a day.

Jason Hiner (21:05.931)
Okay.

Tigran Sloyan (21:16.275)
And there are ways. Sometimes people mistakenly believe the advertising of certain tools that like, this is an undetectable thing. There's nothing undetectable about it. And you have to understand that even if somehow, somehow, which we have best in class detection and prevention mechanism, you were to gain the system, you're still not going to be able to do the job at the end of the day because the assessment measures your ability to do the job.

Jason Hiner (21:23.681)
Right.

Tigran Sloyan (21:44.17)
And if you somehow manage to get through, you're still going to be out of that job in a couple of months because you're just not able to do it. And this is your sort of early signal of, you qualified for this job? Are you going to be successful when you get in? And whatever it measures, we've made sure that it measures what's relevant on the job.

Jason Hiner (22:04.705)
Very good. Tigran, I'm going to pause for one second for my follow up question because I'm not sure why, but the light went off in my background light. I'm going to try to figure out what it's doing. I tried to fix it on my phone, but it, it,

Tigran Sloyan (22:10.505)
I noticed. I noticed.

Jason Hiner (22:30.849)
There we go, okay.

Tigran Sloyan (22:33.971)
Alright, much better.

Jason Hiner (22:37.257)
So the way that this process works, so CodeSignal does what would be maybe used to be called like proctored exams. Like it's making sure that it's verifying that the person taking this exam is the person applying for the job. Am I understand that right? And then it also is making sure that if they have the ability to use the tools that everybody has access to the same tools.

and then if they aren't supposed to have access to stuff that they're not using things that they're not supposed to be able to access. I'm probably oversimplifying it, but that's what you're doing. then this assessment found that people are still trying to override those things. And that's what you've discovered is that there was a lot of people that now just want to sort of cheat to get through the system, whatever these tests are.

Tigran Sloyan (23:28.459)
So the percentage of people who attempt to do it, not succeed, attempt to do it has gone up almost 3x is the kind of key finding of it. And again, that's, think in many ways, you know, it also uncovers other interesting things in the market because there is, especially for young people, many young people are getting told that like, well, why do you need to know how to do this if you can ask AI?

do it. And I think that it's a fundamentally broken premise, right? Like part of the reason why people are still kind of like fighting it a little bit when they're asked to do a skills assessment that focuses on their own ability to do it versus using and leveraging AI to do it. They're like, well, why? And I keep going back to our ability to do it makes us more effective at delegating to AI to do it. And those who can't do it,

are not, even though again, for simple things you might ask and AI will, as they call it, one shot, get it perfect and you'll be like, see, I knew nothing about this. I just one shot asked and got it right. In real world, the complexity of work gets high enough. The nuances get complex enough where those who cannot tell the difference between good and bad and cannot do it themselves if they had to struggle. So which is why

Jason Hiner (24:48.961)
Sure.

Tigran Sloyan (24:49.61)
I believe we're going to continue living in a world in which there will be AI assisted assessments that measure your ability to effectively work and delegate with AI. And we have many of those. It's our fastest growing category, but assessments that are prohibiting the use of AI that are aiming to see your foundational skills will still exist. And this is nothing new. mean, if you look at, you know, math exams, right? You don't, it's just because we've got calculators. doesn't mean that, you know,

Jason Hiner (25:02.017)
Okay.

Tigran Sloyan (25:19.626)
Third graders, fourth graders don't learn how to do basic math just because calculators can do it. And calculators get prohibited at that stage because we're trying to teach you the foundations. And once you get to later stage, like, yeah, of course, you use your calculator to do this faster. But now you know what you're inputting into the calculator. You kind of have a sense of whatever what you've got back is right. Or did you mistakenly put a zero there? And like, you can't tell that you did because you don't know how to do it yourself.

Jason Hiner (25:30.347)
Yes.

Jason Hiner (25:43.391)
Yes.

Tigran Sloyan (25:48.459)
It makes a difference. And I think the kind of confusion around like, the humans need to sort of still learn, still study anymore. I'm like, of course it's the opposite. We need to study even more because the bar for being effective has gone up. Like you actually now many historically as a new grad, for example, in software engineering, you came into an organization and you knew how to do the basics, right? Like algorithms and data structures and

Jason Hiner (26:01.185)
you

Jason Hiner (26:04.961)
Okay.

Tigran Sloyan (26:18.558)
They would be like, Hey, go fix this couple of bugs. And then they will give you like five, six years on the job to be trained on true software engineering, the software engineering principles on how to build complex systems. Now, what used to be a job is now becoming a task. And this is also something that has happened throughout history over and over again, where the jobs of yesterday have become tasks of today. And the jobs of today have

Jason Hiner (26:44.417)
Okay.

Tigran Sloyan (26:46.504)
become more abstract and more complex. To give you an example, document editing and formatting used to be a job, right? You've basically was like, here's a bunch of text written horribly. I need you to like spend a lot of time, figure out how to like properly format this, edit this, the headers, how they should go, how they should be positioned. Now you just like plug it into Google docs and document editing is no longer somebody's job. It's just part of a task.

that we do to make things look good. So same thing is happening, starting with software engineering. Why? Because who integrates AI into our work? Engineers do. Well, engineers, like many humans, are being a little bit selfish. And we started by integrating it first into the engineering workflow. So essentially, when you look at, there is intelligence layer, and then there's an integration layer.

Jason Hiner (27:14.987)
True.

Tigran Sloyan (27:40.245)
Things can be smart and then they can be integrated into our day-to-day work so that they can actually help us contextually do things better. With tools like Cursor and Cloud Code, the sort of integration of that intelligence into work has first come to engineering, which is why there you're seeing first and foremost a very fast transition from things that used to be jobs, such as fix a couple of small bugs and just like do some maintenance work, are becoming tasks and...

Entry level engineers are expected to do even more because that simple task can already be done with AI. Now,

Jason Hiner (28:16.961)
That's a great insight. That is a great insight. Yeah.

Tigran Sloyan (28:19.762)
Right. Now the problem though, the big problem, dark side of this is that companies at least so far are unwilling to be the educators to get you to that next level. So educators have to step up. the expectation from what people should be coming out of their secondary education is dramatically higher today, which is what's driving a drop in employment for early talent. Because companies that are just saying like, Hey, I don't want to

Some companies, actually, some of our partners are taking it upon themselves to say, hey, you know what, we are going to be that next year of education. We're going to help you close that gap because they realized that they can't just stop hiring younger people. That's going to backfire at some point, right? But many others are like, you know what, our more senior engineers are already there. They're just going to kind of delegate to AI. And believe me, this is coming to every other job as well. It's just missing that integration layer.

Jason Hiner (29:05.003)
Yeah.

Tigran Sloyan (29:19.786)
where most other professions are still copy pasting things from ChatGPT and Cloud back into their system, that is a disconnected intelligence in the work system. Eventually, it will be in every part of your workflow.

Jason Hiner (29:36.258)
Okay, that is such a powerful insight and clearly you and your team do a lot of thinking around this and obviously working with organizations, thinking about this sort of at the leading edge of the ways work is changing. At the DeepView, we just did a story with IBM this week where they were...

saying that they are tripling their hiring of entry-level roles, but what they're doing is they're doing similar to what you said. They're taking on that we're going to train them in terms of what work looks like now and what are the things that are uniquely human and let's them focus on those things that humans can do and have to do and not the things that could be done.

Tigran Sloyan (30:10.513)
Yep. Yep.

Jason Hiner (30:20.705)
by by algorithms or AI so or agents so that's that's really interesting. So how does this tie in any other actually before we wrap the the the research itself any other insights from the the research?

or any other data points that would be helpful for our audience in getting their heads around this. I know one that I'll share and maybe you have others that sort of feed off of it, that 80 % of Gen Z reportedly are using in their AI in their daily lives. And so then the entry level assessment cheating jumped from 15 % to 40%. And I'm assuming...

you all making the connection between the two that a lot of those were Gen Z who are, you know, leaving college and applying for their first jobs.

Tigran Sloyan (31:10.299)
And there's also this kind of mistaken belief that everybody is doing it. So I have to do it as well to get ahead. And we've, we work very hard to combat that wrong impression, right? Like, yes, there's, there's tools out there that are really poor are a detection and prevention whenever the rules prohibit, which means yes, everybody's going to be doing it. And that means you have to do it as well to get ahead.

Jason Hiner (31:18.673)
okay.

Tigran Sloyan (31:39.966)
But what we see is that we go out of our way to explain to our assessment takers that we will ensure integrity. We will ensure that others, because again, you got to realize that like this is a prisoners, classic prisoners dilemma, right? Like if others are going to do it, I got to do it too, because that's the only way, right? So the way you explain, because most people are well intentioned, right?

And you have to be very clear upfront that CodeSignal has best-in-class systems to ensure that others will keep the integrity and they will not cheat through this assessment. So you should do the same. And that is key in that transition, right? That like, hey, don't believe what you're told in various different marketing websites of cheating tools. We can detect that.

and we will ensure that others keep the integrity. So you should do the same.

Jason Hiner (32:44.545)
Very good, so I am gonna just flag a couple of the other findings in the study that I really like to see if you have any thoughts on these that, 35 % included some frequent off-screen referencing, so during their sessions where they're doing these assessments, 23 % showed unusually linear typing patterns where complex solutions were produced with minimal pauses or debugging, and then 15 %

Tigran Sloyan (33:03.977)
Mm-hmm.

Jason Hiner (33:13.823)
demonstrated elevated similarity to known answers or leaked content. So you found your tools are strong enough to figure these things out and that's one of the things that people who are younger people out there who are applying and are doing these assessments, for those that are listening to this, we want them to know the tools are very good. They will understand if you are breaking the rules and so.

it should be on them to prepare properly and to make sure that they're doing and applying for jobs that they are a good fit for. Because if they think that they're going to use AI to sort of cheat their way through it, it's not going to be good. And even if they did get the job, that the assessment is going to tell them that they're not going to be in a good place to succeed in the role. Is that summed up pretty well?

Tigran Sloyan (34:05.659)
Exactly. Exactly. Right. Exactly. I think at the end of the day, you know, it's, always tell people focus on your skills because the world is moving towards a place where skilled humans never go unnoticed. And if you just focus on building your skills and obviously we also help educate people on what the right skills are. Cause that's even that's kind of off for debate right now. Right. What are the skills that matter?

Jason Hiner (34:32.939)
Sure, sure.

Tigran Sloyan (34:33.661)
But focusing on your skills is the best thing that you can do for your career.

Jason Hiner (34:39.169)
So you had your founding thesis of the company was you wanted to take people who were skilled and weren't getting a fair shake because they didn't have the right pieces of paper or the right lines on their resume. Are you accomplishing that mission? What does that look like 10 years later?

Tigran Sloyan (34:48.988)
Mm-hmm.

Tigran Sloyan (34:58.023)
Yeah, I'm happy to report that. Yes, we are. So if you look at the industry, I'll give you a couple more stats, right? So in software engineering, more than 80 % of all computer science graduates from US-based universities every year go through our certified assessments. And our certified assessments are assessments that we design and maintain, and not just kind of every company doing their own thing.

So it's more standardized in nature and we see the cross-section. And additionally, most of the Fortune 100s are using skills-based hiring approaches. And it is very quickly, like I said, this has not been possible before 2023. It's only been barely three years, but it's very quickly expanding to other job categories and other job domains.

Our skill assessments and non-technical skill assessments have been the fastest growing category in our business where other organizations and other groups within our customers, right? Like the go-to-market teams, the support teams, the sales teams, the finance teams are saying, I'd love to understand the skills of the people that I hire upfront versus spending hours in pointless interviews and give people a chance that

If you have the skills that you get through, because traditionally what happens is only five to 10 % ever hear back or ever enter the funnel. 90 % just get locked behind the resumes. And that's a massive, massive opportunity for companies to find great people and then for individuals to really shine and put their skills to use to make the world a better place.

Jason Hiner (36:30.401)
Yeah.

Jason Hiner (36:49.141)
Are there, have you enabled companies to get beyond that sort of five to 10 % to find folks that would normally have been missed and that could be potential candidates?

Tigran Sloyan (36:57.075)
Mm-hmm.

Tigran Sloyan (37:01.993)
Yeah, absolutely. Like if you look at, again, our customers, we have some of the biggest employers in the world, right? So from Netflix and Meta and Amazon to Capital One and MasterCard. So like some of the largest, largest employers. And when they work with us, the equation is actually flipped. 90 % get the opportunity, 10 % don't.

most cases because they don't pass basic qualifications such as authorized to work in the country or so there's still some resume filtration that happens mostly on like, okay, do you even have permission to work here? Are you located in a place if it's location specific? But beyond those basic qualifications, almost everybody gets a chance to say, I have the skills to do this job. And we become the arbiter to decide.

Jason Hiner (37:29.857)
Okay.

Jason Hiner (37:39.136)
Okay.

Tigran Sloyan (37:59.111)
do you or do you not? And if you do not, this is very key to our mission. It doesn't stop there. We don't just say, tough luck, you don't have the skills. We say, here are thousands of courses, hands-on learning courses in CodeSignal Learn Platform that you can utilize to actually build those skills. Because you don't have them today, doesn't mean you can't have them tomorrow. And you need to be given the chance to build those skills versus just being told, sorry.

You don't have it.

Jason Hiner (38:30.529)
So your customers are the big employers, but they are also people looking for roles as well. And is there a price to pay to the courses? Is there a subscription or you pay per course or how does that work?

Tigran Sloyan (38:36.808)
Mm-hmm.

Tigran Sloyan (38:45.18)
Yeah, so first of all, there's a free tier, which is very generous in terms of allowing you to learn continuously and no courses are gated behind premium content. All of our courses are freely available. We do have an AI tutor and mentor, which is another part of how Gen.AI Revolution enabled us to keep climbing this mountain, is that great learning happens in form of tutoring, right? So if you're not actually...

personalizing the learning experience, people get stuck in things for very different reasons. Some people get stuck because they forgot what you taught them three courses ago, and others get stuck because they forgot some fact from eighth grade, right? And the only way you can figure that out is to continuously evaluate their skills, but also have one-on-one tutoring. This...

Jason Hiner (39:26.177)
You

Tigran Sloyan (39:38.375)
This is one of the best known educational psychology research outcomes. It's called Bloom's Two Sigma Problem, which has been conducted more than 40 years ago by Benjamin Bloom, which says that when you take a group of people who are learning something and you split them randomly into two parts, and you have one part go through traditional kind of classroom style, one too many instruction, and then you give the other part a one-on-one tutor for everybody.

You see two standard deviations above educational outcomes. To put it more simply, people who would normally get a C get an A. People who would normally fail get a B. And this finding, right, it's not called the two sigma revelation or two sigma solution. It's called the two sigma problem. Why? Because they were like, okay, great, we found something that works. Can't be scaled. How are you going to get a tutor for everybody in the world? Right? It's just a

Jason Hiner (40:33.057)
Mmm.

Tigran Sloyan (40:36.36)
Okay, great, once something works, had to be done. But that was before Gen.AI. Gen.AI actually transformed that, which is why I've been talking about Bloom's to Sigma problem, because it's no longer a problem, it's the solution. So with our learning platform, we've built a very capable AI agent, Cosmo, who is continuously helping you as you go through the platform, as you learn.

Jason Hiner (40:42.209)
Okay.

Tigran Sloyan (41:04.104)
Now we started this conversation, this piece of the conversation from the pricing piece, Cosmo has an energy level and on the free tier he gets tired and he says, come back tomorrow to keep learning. That is literally the only gating factor because obviously we've got to pay our AI bills and we've had to put some gating factor. Now here are a couple of things that happen. First, we constantly give generous scholarships to...

Jason Hiner (41:15.297)
Okay.

Jason Hiner (41:22.475)
Sure.

Tigran Sloyan (41:31.273)
people who cannot afford this to make sure that they have access to free world-class education. Second, we work with our employer partners who both buy the solution for their employees for GEN.AI literacy. For example, one of our partners Dropbox runs an awesome GEN.AI Academy powered by code signal for all of their employees to actually build up their GEN.AI literacy skills and the GEN.AI awareness skills, GEN.AI integration, or even GEN.AI research.

skills with Cosmo, with CodeSignal Learn. And in that case, of course, the company pays, the employees do not pay. And then lastly, many of our partners are actually choosing to buy licenses and give it to their candidates. So they're saying, hey, thank you for applying here. Sorry, not now. And we don't have an internal university, kind of like you mentioned some employers do to kind of take you through.

Jason Hiner (42:18.049)
Mmm.

Tigran Sloyan (42:28.808)
But we've partnered with CodeSignal and we're paying essentially for your education to keep on learning in the hopes that no strings attached, but in the hopes that you will one day when you are ready, you will come and try applying and working here again.

Jason Hiner (42:44.513)
That's great. That's really encouraging to hear and I hadn't heard it yet. And then also, you're basically using AI to help solve some of the problems with AI that AI is creating with jobs. That's, yeah.

Tigran Sloyan (43:02.376)
That is the only way. That is the only way because it is, I mean, and this is true always, right? Any advanced technology, any powerful technology creates opportunity and creates distraction. And AI is no different. There's going to be a lot of job displacement. There's going to be a lot of chaos. There's going to be a lot of people who get caught in this transition. And the only way really to...

Jason Hiner (43:16.565)
Yep. Yep.

Tigran Sloyan (43:31.942)
make it less rocky is to leverage that same technology to make this transition more smooth. And I personally believe that education is the way because everybody will talk to any AI expert. say the biggest disruption is going to come in job loss and job automation. Now, this doesn't mean there's nothing to do. This means that, you know, today's jobs became tomorrow's tasks.

And there's new types of jobs to be done, but they require new types of skills. And we have to provide opportunities for humans to be re-skilled in an AI native world and not in the world that was created before. Cause the world that's been created before is built to educate 20 year olds only really, right? Like it's like, there's a, there's a funnel. Good school, go to college, get a job and then you're set. That world is gone.

The world we live in today is that in a couple of years, all truck driving is going to get displaced with autonomous vehicles. All Uber driving, all delivery is going to get displaced. mean, you come to San Francisco every day, there's more waymo's than there were yesterday. You can just see it happening. Now we can't just sit back and say, well, tough luck. What you used to do to feed your family is no longer possible.

We have to say, here is how you can learn new skills to continue providing for your family, to be a contributing member of society. And just saying, you know what, well, let's just do universal welfare, universal basic income. That's not going to solve it because humans still need purpose. You need to do something to feel useful in accomplishing something and not just get a check every day and say, go do whatever.

Right? Like we need something to fill that sense of purpose.

Jason Hiner (45:27.615)
Yeah. Yeah.

Jason Hiner (45:34.06)
Quality education materials is quite a challenge too. How do you handle that? Because it's great that you're doing it. You have your own academy. Do you have partners with that? Do you have your own educational department? How do you make sure? And of course, as you know, I'm sure it could tell much better than me, educational materials when it comes to tech are almost outdated, but the moment you finish creating them. So how do you manage that challenge?

Tigran Sloyan (45:43.429)
Mm-hmm.

Tigran Sloyan (46:02.599)
Absolutely, it's a fantastic question. And when we started on the learning journey, we wanted to go partner to kind of solve that exact challenge. But the challenge with partnering is you simply do not have hands-on learning content available almost anywhere. Like we looked, right? Most of existing learning courses and content is videos and passive learning experience. And we knew that's not an effective way to teach. Effective way is hands-on.

and one-on-one personalized. You have to get both of those things right in order for you to even have a chance of engaging effective learning experiences. So there was no other way we had to build it ourselves. So we've built up a large course production team that is continuously building and staying on top of all the new changes. But again, we couldn't have done it without help from Gen.ai. We've built many tools that make this process simpler for us, such as...

All of the video lessons that we produce aren't filmed. They are generated from human curated lessons. So we human curate the lessons and then instead of spending days recording videos, we actually within minutes go from that to a video lesson that would have taken us, you know, three times more, if not longer to actually accomplish.

Jason Hiner (47:22.635)
Sure. Yeah. Yeah.

Tigran Sloyan (47:25.997)
And then there's other elements to it, such as review and catching mistakes in this, right? Like historically you had course designers, then you had reviewers, then had this QA iteration cycle. Now we go from, here's a lesson that we've produced to, there's issues in there. And here's quick fixes to it, again, within seconds. So all of these advancements in generative AI have allowed us to go from, know, hey, we just...

enter the education space a few years ago to being one of the leading providers with one of the largest content libraries available and content library that's not just passive learning content library that is all about practice, practice, practice. If you're an engineer, you've got an IDE where you write code and then Cosmo tells you if you're doing it right or wrong. If you are a, actually a student who's trying to understand, you know,

How do I, we have this thing called career readiness Academy, right? Cause many students graduate, they've never worked in the industry before. What's professionalism look like in a meeting? Well, we have simulated environments where there's an AI agent like your, tends to be your boss who's giving you feedback. Like, how do you react to that? How do you, these are skills that people normally learn on the job, right? And the more you can practice it in a simulated environment in advance in a safe environment where you can mess up a hundred times.

Jason Hiner (48:42.016)
Yes.

Tigran Sloyan (48:49.211)
There's no implications. There's nobody's going to be like, this guy, right? Mess up. It's okay. You can just go again and Cosmo will be like, look, this is what you're doing wrong. Let's just try it slightly different this time.

Jason Hiner (48:54.133)
Yes, yes.

Jason Hiner (49:04.075)
Very good, so you've created a whole content team that's trying to create sort of the most sort of cutting edge, best content in the world for learning how to develop your skills.

Tigran Sloyan (49:16.293)
Yeah, the only way. The only way. I wish there was an easier way.

Jason Hiner (49:18.721)
That's, that is, I was gonna say that is a formidable, a formidable thing to do. But, but very cool. All right, Tigran, I wanna ask you two questions I ask at the end of these that I'd like to ask all of our guests. One is you're a CEO and in the world that we're in, you know, the most valuable commodity we have and maybe the real currency is time.

And so how do you spend your time and what's perhaps your best tip for other leaders, CEOs and leaders that are trying to maximize their time, maximize their leverage?

Tigran Sloyan (50:01.85)
Yeah, it's a great question. And I'm a contrarian in this because every business school course you look at, every book you read, they encourage you to sort of find leverage in delegation and abstraction. And almost they encourage you to kind of separate yourself from the work itself. I believe that the opposite is true. I believe great leaders lead from the front and they understand the details.

Jason Hiner (50:22.347)
Yeah.

Tigran Sloyan (50:30.038)
You don't have to understand the details of everything, right? But you want to narrow down what are the biggest challenges facing us right now? What is the thing that is truly getting in the way? And once you've kind pinpointed them, you don't just say, hey, you go figure it out. Because guess what? The hardest problems, if your team could have figured it out, they would have already figured it out. So what I try to do is find the bottlenecks in achieving what we need to achieve and then go very deep into it.

Not alone with the team to say like, okay, let's think from first principles. Why is this getting in the way and what can we do about it? Sometimes it's about how on earth are we going to build thousands of courses in a matter of months to be able to help our customers? And sometimes it's, look, cheating is on the rise. What are we going to do about that to continue be serving our customers? Whatever is the hardest problem at that moment, whatever is the bottleneck.

I get very deep, get very hands on and I work with the team in the frontiers to figure out how to solve it.

Jason Hiner (51:35.289)
As a journalist, I love a good contrarian, so I appreciate that. And then the other one is, the last question is, what are the AI tool, what's an AI tool that you're using regularly these days that's really helping you and that you think others could benefit from?

Tigran Sloyan (51:38.131)
I'm

Tigran Sloyan (51:43.462)
Thanks.

Tigran Sloyan (51:54.299)
Yeah. So that second part changes things because my favorite one will be Cursor, which is an AI agent for writing code and understanding code bases, which always helps me, especially when I'm in the frontier, trying to understand what's going on. But a tool that actually I think a lot of people can use that I've found very valuable is Claude. Obviously everybody knows Claude, but Claude's Chrome extension, which I feel like many people don't know about.

Jason Hiner (52:04.587)
Sure.

Tigran Sloyan (52:23.346)
Claude has this Chrome extension that I can actually take actions for you in the browser, such as, hey, fill out this random forms that, you know, given this context, I don't want to click into 50 places to do this myself, or go make a quick reservation for me at this place. Essentially anything you can do yourself in the browser with enough context, with proper prompting, you can just hand it off to Claude browser extension and walk away.

Jason Hiner (52:51.394)
Very good. Tigran, thank you so much for being here. Great conversation. Thanks for the work that you're doing. Clearly it has a lot of meaning.

Tigran Sloyan (53:11.206)
Thank you for having me, Jason.