🎧 Listen Now:

In this edition, I have the pleasure of speaking with Ashutosh Garg, the founder and CEO of Eightfold, a unicorn HR tech platform that brings AI to improve the entire talent lifecycle, companies, and the world of work. 

Ashutosh brings a wealth of experience from his early days pioneering personalization at Google to now helping companies like Ernst and Young, Bayer, Morgan Stanley and Starbucks use AI to revolutionize how talents are managed and nurtured in the workforce. 

Today, he'll share insights from his journey and how AI can play a pivotal role in enhancing our work lives.

In this episode, we gain the following insights from Ashutosh, here are the key takeaways:

1. Career Development and Employee Retention with AI: We've learned that AI's potential to tailor career paths to individual skills and aspirations can dramatically transform employee growth and satisfaction. By aligning opportunities closely with personal competencies, organizations can maximize employee potential, drive engagement, and keep the best people longer.
   
   
2. Human in the Loop: This new way of thinking about people also means sometimes a team or manager has to give up a great employee to benefit the greater good, central to the theme of Josh Bersin’s dynamic organizations at the beginning of the season. It also means that even the smartest AI platform will always have a human in the loop, to control its impact on an organization.
   
   
3. Reducing Bias with AI Tools: AI platforms can significantly reduce human biases that often influence hiring decisions. By standardizing processes and focusing on data-driven assessments, AI helps ensure that recruitment is based on merit and qualifications, not unconscious biases, leading to fairer employment practices. I love Ashutosh’ point that this starts even at job descriptions, where we need to be objective about even the kind of tasks done in the role we’re hiring for.
   
   
4. Creating a Better World of Work: Ashutosh's mission with Eightfold is to create a better world of work—a place where employment is more than a necessity but a fulfilling part of life. AI is poised to transform the workplace into a more satisfying and productive environment by matching individuals to careers that genuinely fit their skills and aspirations.

🔔 Available on:

You can find the full episode and transcript here:

Transcript:

Daan van Rossum: You're currently leading Eightfold, which we definitely want to learn more about, but maybe just a little bit of introduction about yourself. I know you moved to the US a long time ago. You've been in the Bay Area for quite a while. What has life been like up until starting Eightfold? 

Ashutosh Garg: Well, this is my second venture. So now I have been doing startups for almost 15 years. My background is that I grew up in India, in a small town near the north, and then moved to the US in 1998.

Fortunately, the timing worked out well. AI was still in its infancy back in those days, but I got an opportunity to do a master's and a PhD in machine learning in AI. And that has shaped my career ever since. I spent a year at IBM Research, followed by a short stay at Google Research, where I led all the personalization efforts, and since then have been doing startups since 2009.

Daan van Rossum: Wow. What did you say at Google about personalization?

Ashutosh Garg: No, I was on the research team responsible for most of the personalization efforts at Google. Whether you are reading Google Reader, Google News, watching videos, making search queries, or making sure that people are getting the results that are relevant to them based on what they have and what we know about them,.

Daan van Rossum: Wow, incredible. Now, that's stuff that we're so used to; it's very normal to us, but you were there when that was just getting started. 

Ashutosh Garg: Yes, we started the personalization team at Google.

Daan van Rossum: Yeah. How long were you there for? 

Ashutosh Garg: I was at Google for four years.

Daan van Rossum: You must have seen quite some shifts during that time.

And then, after Google, that's when you started your first company?

Ashutosh Garg: Correct.

Daan van Rossum: What did you do? Because obviously, like at Google, you're doing really interesting work. You're meeting a lot of really interesting people. Did some ideas kind of start percolating there, and did you have to just shoot off and do it on your own? How did you make the jump into entrepreneurship? 

Ashutosh Garg: So it was the year 2008, which is not the best time to start a company on one side; the market was tumbling down and everything was falling apart. But sometimes you feel like that is your calling; you have to do something.

More importantly, what you realize is that in a large enterprise like Google, as an individual, you can have only so much impact. The second is that you are bound by the constraints imposed by that company. So I was like, Can I go outside at that time? And with all the technologies around personalization and search that I'm building within Google, can I actually use the same thing to enable a much better e-commerce shopping experience?

So that was the genesis of my private company, and yes, on Google, we are providing you personalization around search, around news, and so on. But can we start providing a search and personalization experience when you go to, for example, Sears, Kmart, Staples, or any of these shopping websites? Needs a much better experience for consumers but also much better value for business owners. So that is what made me leave Google to start Bloomreach, my prior company.

Daan van Rossum: That's a big jump. Like you said, there's obviously limitations to working inside a company like Google where, you know, you cannot be as creative or entrepreneurial as you would be on your own, but obviously there's also some benefits to working in a large company like that, right?

There are some good perks, and there's obviously a lot of support behind you. So what was it like when you didn't first have to venture out on your own—the good and the bad? 

Ashutosh Garg: Yeah, suddenly, you realize that you're nobody. You have to start from scratch. You no longer have the excuse to say that this is what I like to do or what I don't like to do. This is what I can and cannot do. You have to do everything. But I think the key thing is that the common theme that I've seen with most founders, especially those who have done well, is that they never complain about what they don't have but are more focused on the best they can make out of what they have. To me, it was less about what I'm leaving behind.

It was the excitement of building something new.

Daan van Rossum: What part of that did you like the most? You were obviously research scientists before that, and you have been working with different teams. Then suddenly you're there, probably not alone. I'm sure you raised and built a team, but what was kind of like the part of the role that you liked the most in that new venture?

Ashutosh Garg: I think overall, I like everything.

Daan van Rossum: That's good. That makes life easy.

Ashutosh Garg: But the key thing is that in a startup, it's a full-circle experience. In a big company, you are touching small pieces of a big puzzle. At some point, no one is touching the full puzzle. Even at eightfold, now that we are almost 650–700 people, no individual is touching the whole thing, including me.

But when you have 3, 4, or 5 people in a garage, then it is ultimately you who are making everything happen at some level, or you and two people who are making end-to-end everything happen.

And all bugs stop at you; all accountability stops at you. When a customer buys it, they are not buying it because of the product; they are buying it because of you. When someone joins the company, they are joining because of you. When investors come in, they are coming in because of you. which has a lot of pressure.

Daan van Rossum: Yeah. I was going to say that it has a good side and a bad side too. That also means it's all on you. If it doesn't work, it's you. How did you deal with that? 

Ashutosh Garg: You have no excuses. It took me a while to internalize it, to be honest, but as I was internalizing it, what it meant for me is that I really have no excuse. If it's not working, I can't go and complain that it's not working because of someone else. Ultimately, I'm responsible. I need to make it happen. So you have to just analyze that. Yeah. Which was fun.

Daan van Rossum: Okay. I could see how it worked out. You clearly liked it. 

You did that for how many years were you in that company? So this is like e-commerce, personalization, and all the stuff that we're now used to when we go to places like Walmart or Amazon. You sell a lot more if you can make the recommendations personalized. And then somehow it all led to Eightfold. So tell us a bit about the origin story because you're going from personalization in Google to personalization in e-commerce. There's still some logical jump, but now suddenly you're doing a skills marketplace, talent, and recruitment. Very different. It seems, at least from the outside, what led to that founding? 

Ashutosh Garg: No, you're right. Actually, it seems very different from outside. In fact, traditionally, what people would say is that they stay in the same space.

I was in the marketing space. Keep building a company in the marketing or e-commerce space. Same buyers, same relationships. I think there were a few reasons to start Eightfold. The very first thing was realizing how important employment is for people. If we can provide everyone with a better career, a better job, and better employment, that impact on society will be very, very large. So that was the genesis of, like, let's do something over here.

But then the second thing, where it all comes together, is that the first half of my life was all about personalization and recommendations. Based on everything you have done to date and what you have bought, what is it that you're going to buy next?

What about in a career? If I can do the same thing based on everything that you have done to date, I can better understand what you are likely to do or what you can do next.

For example, you may not have ever done coding in Java, but if you have done coding in C++ and Python, what is your likelihood of learning Java quickly? If yes, now suddenly I have opened up more job opportunities for you. So the key thesis was that it's the same personalization. The last part of my life was spent building personalization models for users to predict their shopping behavior. Now let's use models to predict their skills.

Daan van Rossum: I'm sure it's a good business as well, but it sounds like it was a very mission-driven choice to again apply those same learnings and those same technologies to people and to give them better careers. So I may share a little bit about that. 

Ashutosh Garg: No, absolutely. It was 100% mission-driven. It was all if we could help people and individuals. See, normally when we think of let's do something good for society, we think of education and healthcare, but what I realized was that at the heart of each and every one of these things is employment.

See, we are making all these choices on a daily basis around jobs and careers. For some people, it is about feeding the family. For some people, it is about a bigger purpose, but the bottom line is, do I have the right job or career?

If that is the case, you are happy, and you are excited. Your day goes much better, right? If not, you scramble. If you're unhappy at work, you're unhappy at home. If you're unhappy at work, your health goes down. So the key thing was that today, when people are looking for a job, I mean, I'm sure you have friends and family members who apply for a job and never hear back.

I knew this, but the interviewers just didn't understand what I'd done. They didn't give me that opportunity. All of us have had that experience. So the key thing was that, through data and AI, can we help interviewers see past the candidates just one-page resumes to understand more about what they have done?

Similarly, on the other side, for an individual, if they can better understand what is the right role for them and the right job for them, we can do a much better job at connecting people to opportunities. We can make it a much more fluent workforce. And if yes, you don't have to do an amazing job, or even if you can make things 5% better, the impact is so large. So that was the thesis over here.

Daan van Rossum: Can be hugely impactful. Absolutely. Amazing. I love that you said that employment is so important. I talked to Josh Bersin at the beginning of the season, and he said, One thing I hate is when people tell me that they hate their job. They're not happy with what they do, and it's where we spend most of our time, outside of sleeping and maybe some family time. Some hobbies, like that's really where we spend most of our time. 

So making work better obviously makes life better, and like you said, those influence each other as well. If you come home from a good day at work, you're definitely different in your relationship, in your family, and in your community than if you're not getting meaning out of your job or if you're not getting something out of it beyond just the paycheck. 

Ashutosh Garg: Let's just flip it for a second. Why does it even work? You might say this is all financial. But what makes it a good business? And what makes it a good business? I mean, you are running your podcast as well. If you have employees who are excited about or see career growth in your company and who are the right fit, your business is going to thrive. Like it's not always that you need a lot more people, you need the people who are focused on the work, who are excited about the work, who know the work, who can do it.

What I like about Eightfold and this problem is that the best businesses are where there is such phenomenal synergy in the entire ecosystem. By working with enterprises and helping them better understand candidates, I can actually solve the problem for both.

Businesses are the ones who are employers. That is who we work with. But ultimately, the value we are able to drive is to both of them, and it's not about tricking an individual to take a lower-paying job or tricking an enterprise to get a poor candidate; the best outcome happens when it's a great fit between the two. It's good for both of them.

Daan van Rossum: You can make an impact on both sides. Then it starts the flywheel, because if you bring the right people in, then obviously you will hopefully hold on to them better. And that would, obviously, eventually really impact the company. 

Especially for people who don't know the platform, share a little bit about how Eightfold works, because I know there's sort of two sides to it. You'll be able to explain it better, but there is a recruitment side to it, and there's also a retention, learning, and development side to Eightfold. So maybe just share a little bit at a high level about how the platform works, because it's very intriguing. 

Ashutosh Garg: Absolutely. So high level, the way we describe it is that we have built a talent intelligence platform and a bad talent life cycle. In an organization, what is a talent life cycle? You're running a business. You have, let's say, a thousand employees or ten thousand employees, and suddenly you realize that over the next two years, new technology is coming in, new products are coming in, a new market is coming in, and you need a new skill set in your organization to grow or sustain.

At the same time, because of market dynamics and because of how things are internally, maybe there's some churn going on. Some people are leaving; some are attrition.

So what you want to do first is make sure you're able to retain the right people. Second, you want to make sure that people, if they're not in the right job, can be moved around so that they stay with you and deliver. If there are people who have the skill set but are lacking a few things here and there, you can up-skill or re-skill them so that they meet their future needs.

Then you look at your alumni and say, Do you have an ongoing relationship with them because they are maybe the best source of talent for you? They are the biggest brand ambassadors. And then, wherever there is a gap, can you actually identify who the right people are to hire and recruit them efficiently at scale.

Daan van Rossum: The platform helps with that. What's the technology behind it?

Ashutosh Garg: An entire end-to-end platform for helping you manage your current employees, understand their skills, develop their skills, up-skill or re-skill them, provide a talent marketplace, and think of success planning for your key leaders or every employee in the company.

Then there's CRM functionality to engage with every talent who's out in the marketplace and an entire recruiting workflow system to help you hire the right talent efficiently.

Daan van Rossum: Besides AI, there must be a lot of data involved in terms of understanding people. Going back to that recommendation engine kind of idea, you need to understand a lot about that person beyond just the one-page resume, which I'm sure is a terrible indicator for who this person really is, what they're good at, and what their passion is about what they want to do, and then you need to understand what is needed in a company.

So I'm just assuming that this really becomes an operating system for a company, not just a tool that you use here and there. If your customers want to use it successfully, I'm sure it's implemented throughout their business. So this has really become, like, an OS for the company, and what kind of data goes into that?

Ashutosh Garg: Data is the new lifeblood of all these things. Data is available from thousands of different public data sources. You might be updating your profile in all kinds of public places; data is sitting over there. Data is sitting in internal HR systems. Data is in the enterprise flow of work systems. So you're using Google Docs, Confluence, Jira, Asana, GitHub, Bitbucket, Salesforce CRM, and Microsoft Dynamics CRM; you're using hundreds of internal systems; details are all over the place. You're writing papers, so they're sitting in publication databases outside. You're filing patterns, sitting with USB to your other pattern databases.

So what it will do is that each person brings the data from all these places into one single place. But then what the second thing it does is that it looks at the data of each and every other person or similar to this person to better understand the data of this person and uses that to build models around each person and each role in the world. For example, what does it mean to be a senior software engineer at Google on a certain team? What does it mean to have a certain skill? And what experiences do you have that might relate to a certain job?

Basically, we say that, as an AI company, you really need to excel in algorithms and data collection. And you have to do both.

The third thing that you actually have to do is really excel at understanding user feedback and recommendation systems with AI. The good thing about humans is that they are very adaptable. If I say something to you, you will listen to me; you will react, and that is how AI systems have to be designed. Every interaction is also producing data. And can we capture that data, and can we take the data as a feedback loop that makes a system much better over time?

Daan van Rossum: And that would be, for example, if my company uses Eightfold, and Eightfold learns a lot about me because it shows me different opportunities, maybe for new roles or new skills to develop, based on what I'm doing within the company.

Ashutosh Garg: Every enterprise's data is their own data. See, in a company, there are courses that you might take. There are mentors and coaches that you can work with. There are full-time job opportunities within the company. How do I understand each person? Everything they have done says that these are the things that you can do. These are the things you can learn and do today, but this is the path you need to be on to get there over time. Who can mentor you to get there? What courses can you take to get there? What skills do you need to develop? And then let's say I show a job to you, which is about product management, but I see that you're not excited about that. You're more excited about marketing, so how can you now update the system to automatically guide you around marketing? So the constant feedback from your interaction with the system.

Daan van Rossum: Then I'm very curious. In that case, if the system realizes that somehow it's seeing something in my data—maybe my conversations, maybe any data source that it has access to—so it's seeing that actually, I think you would probably fit the marketing track better, and I know that those positions are open in the company.

Is there still any human involved in saying, Yeah, but I don't want that person to move to a marketing role because sometimes those can be conflicting? Maybe at a company level, we want people to be in the right roles, but maybe at a team level, my manager would want me to stay in my position because it's hard to replace me. Is there an opportunity to override or to set a temperature meter, like you do in AI, on how much we recommend versus how much we dictate? 

Ashutosh Garg: And companies can set their own policies. But yes, change management is a big part of what we have to go through. I may not want my team to disperse at all. I may want to have talent. We're at the company level. If people don't move around, my culture will not grow. I will not learn from people.

In fact, these people will eventually leave if they don't get another option. Sometimes what companies do is set up the policy that, okay, for the first 18 months, you cannot move, but after you are in the company for 18 months, and if your performance is, let's say, above a certain threshold, then you can move within the company.

Daan van Rossum: I'm very curious about this—almost competition between individual teams and their leaders and then the company as a whole—because I'm sure that, like before using a platform like Eightfold, this would have been invisible to company leadership. 

They would not really know that people are stuck in the wrong role, so a manager is preventing them from moving on. But the moment you make that a machine, you make that a platform. Now these insights do maybe go to the top, and maybe we see where people are being blocked, and maybe we see some reasons why there's attrition or why there is a lack of retention in teams. 

Ashutosh Garg: Absolutely. And I would say that things are changing. Five years ago, when we were talking about internal mobility, this was the number-one question people asked. That's what you're asking now. Actually, over the last few years, I think that noise has gone down dramatically.

I think as COVID happened, people realized how important it is to focus on their own employees and teams. They also realize during this time of great resignation that people are going to move. They need to be given the opportunity. And as part of that, even managers realize that that's a new world we are living in.

One thing that is interesting, and we don't talk about it, is that most managers themselves have been in the job for less than a year or two. Sometimes, once you internalize it, if I'm switching and that is what I want, I can't stop my team from doing the same thing. So I think now people are a lot more okay with this thing.

Daan van Rossum: On the recruiting side, you're getting a lot of data. Is there any issue around bias in terms of the platform maybe recommending certain people for certain roles that would maybe not be in line with how a human recruiter would do it? Not to say that human recruiters don't have bias; you know the topic.

Ashutosh Garg: Absolutely. And I think that's a great question. The key thing is using data to bring transparency to the whole process. One is the way the system has been designed; humans are in full control of what is happening over there. A very simple example is when you are creating requirements for a role. Today, we don't even understand the implications or diversity because of that. For example, I may just say that I'm looking for someone with these 20 skills.

Now, I may just not realize that all these skills are not really needed, and some of these skills are more biased against women. At each step in the process, the way the system has been built, we provide transparency and have humans involved. So humans modify, adopt, and adjust things based on their organizational needs.

The second thing is that we have built the way the AI has been designed, and that is a key thing over here. With the biased elimination in mind or reduction in bias in mind, So I think the more important question could be that if there is bias in the data, are you going to perpetuate bias or are you going to reduce bias.

The way our system has been designed, we call these equal employment opportunity algorithms. That is, specifically, taking into account how our model's performance is going to be for each gender class, each race, each ethnicity class, and so on. And then we provide rich analytics so that, as an enterprise, you can monitor each step in the interview process. Is bias creeping in?

Let's say 100 males and 100 females apply for the same job. And so at the top of the funnel, the ratio is 50:50. When you're doing the phone screen, are you now inviting more men than women? Yes, the system will flag and then say that there is some bias going on over here. So, actually, it helps. Not only are the algorithms well designed, but they also help reduce human bias.

One very simple example of what we do is that when you're looking at a resume or profile of a candidate, we can mask all the things like name, gender, race, etc.

Daan van Rossum: Everything that gives it away.

Ashutosh Garg: So that you can only focus on what's relevant for the role.

Daan van Rossum: But there's something so interesting in what you said that if it goes wrong at the level of writing the job descriptions or requirements, then it's almost like it doesn't matter what happens in the next stages.

And maybe we write job descriptions almost already, thinking about what the ideal candidate looks like. And there may be bias in how we write that JD. So already, before we even get the platform involved, before we even speak to the first person, we've already adjusted the funnel to make it work for what we think is the right person.

So a system like this could then detect that and maybe give recommendations too, because probably you know better; maybe the platform would know better who to hire or what the right profile is than the hiring manager.

Ashutosh Garg: Let's say hypothetically that you are a recruiter. I'm a hiring manager. I told Daan that what I'm really looking for is someone who knows Java, Python, or C++. Sounds simple. You're like, okay, I'm just a hiring manager. He knows what he is looking for, and I'm looking for a data scientist role. So I'm looking for a data scientist who knows Java, C++, or Python. What if you come and tell me that, Ashu, most people who know Python are a lot more men than women? But if you need to find diversity balance, then maybe you should look for language R as well, which has a lot more female population than men, and you will have a much more balanced set of candidates coming in.

In general, in that case, my reaction will be, Oh, I didn't know that, sure, absolutely. It's all the same for me. The first time, I just didn't know, and I may just bias myself against one class over the other. It has information on how the data is distributed.

Daan van Rossum: Then those people get hired. You said maybe one of the data sources could be the actual collaboration platforms they're using. Maybe some data sources could be the HRIS. So you could see things around again. How long do people stay in a role? What are their performance reviews? Maybe even how do they work day to day? Does that also allow you to almost go back and create a sharper profile for who the next person in that role should be because you collect so much data? 

Ashutosh Garg: Absolutely. I think that the way we have designed the system even on day one, more often than not, instead of asking you what you're looking for, we just ask you who are the top five performers in your team so that we can build a profile based on that. There are quite a few times when, when you describe it in words, you bring all the bias. Well, I'm really looking for someone who's outgoing, who's very talkative, who can do this thing, who is... no, just tell me what the five people are. Let me look at what experiences they have and what skillsets they have.

And you realize that, no, actually, half the things you're talking about are not even relevant. And you are missing out on the relevant stuff. Having that ideal candidate profile actually helps a lot in building the right models.

Daan van Rossum: But it seems, on the topic of bias, that there is a lot of tension between people and the system. Somehow the systems get called out for inducing bias or perpetuating bias and there's a lot of regulation now coming in around. We need to do bias audits and these things, but obviously humans are incredibly biased and incredibly flawed generally. And it seems like maybe there's some opportunity here to correct people. When you're saying that I need to hire this and this person, if you say from the get-go that it has to be someone really outgoing and with these qualities, maybe again, you're already skewing the JD and therefore the interview process and therefore the final candidates towards a certain person that you have in mind. 

Ashutosh Garg: Absolutely. The way to think about it is as follows: there's a conscious bias, and there's a subconscious bias. Quite a few times. The conscious bias comes from the fact that if there's someone on your team who's actually biased, true analytics will capture it and show it back to you. But more importantly, I think 80–90% of people are good. They want to do the right thing.

The bias comes in because we don't have all the information. Suppose I didn't know that Python and R could have a different demographic profile. Like I told you at the beginning of this podcast, I was at Google. You're like, Oh yeah, you were at Google. So you must know all these things. If I had taken the name of a much smaller company that is not as successful, you would have completely ignored it. Even though that company may have really high-quality talent who might have worked really well,.

If I say I went to Stanford, you'll be like, Oh, Ashu, that is awesome. It's a great institute. If I say I went to BITS Planning in India, you're like, What? Where did you go? I don't know if that is cool. Is that…? So there may be a lack of knowledge, or one recruiter may know one thing, and another recruiter may know another thing.

Through AI, what we are able to do is democratize that information for everyone. Everyone has access to the same assistant, and that helps a lot.

Daan van Rossum: I can imagine what that is. It now speaks to people probably way more than even two years ago, because now that people have seen that, oh, I can talk to this bot who somehow has all the knowledge in the world, maybe now this idea comes to life more. Has it become easier to sell this kind of solution now that decision-makers are maybe a little bit more understanding of AI in their own daily lives, because I'm assuming it was hard early on?

Ashutosh Garg: Yes, till two years ago, the conversation at times would have been, Do I need AI or not? Now, no one is questioning that. People are like, Yes, of course, I need AI. So in that sense, it has become a lot easier for us.

Seven years ago, when we started the company, people were like, Why are you even putting AI in your name? Like, why are you so fascinated by AI? Now, seven years later, it looks like you made the right call back in those days.

Daan van Rossum: Definitely. So you must be quite thankful to OpenAI and ChatGPT for somehow educating the masses on what AI can do. And where do you see AI's biggest power within what you do? So you say about unlocking maybe information people don't even know that helps them make much better hiring decisions or better retention programs? What are some ways in which AI really helps? 

Ashutosh Garg: I think, as of today, it really helps on the understanding side, inferencing, but I would say that over time, it will enable us to do a lot more things, a lot more efficiently. Whether how do we upskill the talent, how do we teach and guide our employees, how we craft the projects, how we take various actions? There are so many manual processes that happen in that challenging space, especially the ones that are very tedious and cumbersome and that no one likes to do. You can start automating some of those as well, and then the scope is unlimited. The way to think about this is that AI is no longer that. What is this new piece of technology?

Another simpler way I would say it is that, as a HR person, I already have a team of 20 people who are doing great work. But they're overworked. Suddenly, I have an army of another 20 people who are comfortable doing any amount of work for me. What more can I do? How can I make my lot happier and more efficient? So I think that's how I'm thinking about AI.

Daan van Rossum: So it's really freezing up to do the work that we're good at as humans, and that's hopefully a lot more meaningful than a lot of the admin work that we were doing and sitting in endless meetings and messaging back and forth.

If AI can take over a bunch of that stuff, then hopefully we can do the things that are meaningful to us and then we enjoy doing, which I think is much more important to your point about; there's something very important to be done in employment and at work.

I guess one final question is around regulation, because you're in the AI space. I'm sure you're keenly aware of a lot of regulations being either proposed or being implemented, and in California, there is an act now being implemented, and maybe in other states not. You said you have about 700 people, and there are quite a few people in that team tracking all these regulations, and in all the markets that you're working in, how do you even keep up with everything that's being proposed that may affect the way that the platform would work?

Ashutosh Garg: One, regulations are going to be here. They're already here. More and more are coming. The second is that instead of fighting the regulations, you have to just embrace them. I think many of those regulations are designed with the right purpose in mind. The third is that it's not about whether one state has a regulation and another does not. See, if there's a state that does not have regulations today, they will have regulations in a year or two years from now.

Daan van Rossum: You're trying to anticipate what that's going to be?

Ashutosh Garg: So we have been worrying about bias from day one. When New York law came in, we were like, Sure, yeah, let's do the audit. Who cares? We'll be fine. EU AI Act—sure, we are not worried about it. We have gotten our audit done multiple times. Our audit reports are on our website, among other things. You have to be ahead of those regulations. You have to design the systems with all those things in mind, and if you are following the law, just part of the law, you'll be fine. You don't have to worry about whether there's a lightweight regulation in other states. But if you're barely trying to get by some of these laws, then you will always struggle. So, this has not been an issue for us at all.

Daan van Rossum: Okay. But I'm sure there must be someone on your team who is pretty stressed when they hear about a new.

Ashutosh Garg: The good thing is that my general counsel and I did our PhD together in machine learning. He knows law, and he knows AI. So I'm like, please take care of all, making sure that. In fact, I'll give you one example. It's not a regulation yet in India. I think almost three years ago, we started an AI ethics council at Eightfold. We got the ex-commission of EOC, the ex-director of OFCCP, and a few other people and said, Let's have an AI ethics council. This council will meet once a quarter to review all the AI work we are doing and make sure that not only are we following all the rules, regulations, and laws as they are coming in, so we are up to date with all that stuff, but we are also putting all the checks and balances internally. So that has been very effective for us.

Daan van Rossum: What does ethical AI even mean in that context? What would they test it against? 

Ashutosh Garg: The way to think about ethical AI is that it meets all the laws of the land. Second, it is very transparent. So that you understand what it is doing. It keeps the human in loop at the right time in the right way, and under the standard definitions of bias, it is able to ensure that the way the system is working, the bias is not creeping in. In fact, it is being reduced and eliminated.

Now they are, of course, very complex definitions of ethics, and maybe there will be a time and day to discuss that as well.

Daan van Rossum: Definitely not today because I see we're getting to the end of the time. So I'll close on something a lot lighter, which is maybe one wish that you have for the future of work. 

Ashutosh Garg: I think first of all, you're doing a fantastic job capturing these topics. I wish great success to you and the Future of Work podcast. Hopefully, it will reach every person in the world. I think work is materially important for all of us. So thank you for sharing the future with us.

Daan van Rossum: My pleasure. Then we definitely have one minute left. So then I also want to ask you one piece of career advice that you would give someone because you're really in a position that a lot of people would aspire to. What is something that you would recommend someone do in their own career? 

Ashutosh Garg: Follow your heart, but stay true to yourself. If you want to do something, actually do it, commit to it, and have a big impact. We are living in amazing times. Who would have thought technology would have come so far? This is the best time ever. The world is our oyster, and technology is making everything possible. So I would say to follow your heart, follow your passion, do something that is good for humanity and our society, and everything else will follow.

Daan van Rossum: Amazing. Make a beautiful impact. I could not have asked for better closing words. Ashutosh, thank you so much for being on today. 

Ashutosh Garg: Thank you.