Transcript
Mark:
Welcome to PeopleTech, the podcast of the HCM Technology Report. I’m Mark Feffer.
My guest today is Mark Chaffey, the CEO of hackajob. The employers and candidates on their platform are all part of the technology business. Their candidates are tech professionals and the employers have some kind of skin in the technology game. It begs the question, are these folks interested in hackajob’s internal workings, or do they just want a good match, users, results, and advanced technology? That’s what we’ll talk about on this edition of PeopleTech.
Hi, Mark. It’s nice to meet you. Tell me a little bit about hackajob. What do you guys do?
Mark Chaffey:
We’ve built a full stack solution that enables predominantly enterprise organizations to improve every part of the hiring process. So, we have a two-sided marketplace that enables companies to directly source talent. And the real magic here is the level of engagement and relevancy of the candidates in the marketplace. So, companies get a 90% response rate to talent that they’re directly sourcing from our marketplace, which is just far higher than any other sourcing tool you’re going to find in the market. We then have a DE&I insights platform, which is really powerful because 80% of our users self-disclose D&I data. We’re then able to aggregate that data and play back your hiring funnel at every stage broken down by gender, ethnicity, neurodiversity, sexuality, disability, veteran status, et cetera.
And that’s really shining a light on companies’ processes that they often don’t have data to. I was speaking to one of the major tech businesses a couple of weeks ago, and they get about 7% of that application, self-disclosing D&I data. So, if you can’t measure it, very hard to improve it and we’re really enabling companies to measure that kind of D&I performance throughout that process. We then have an employer brand platform that talks to our wider community and enables companies to build more passive talent pipeline. So, we’re really trying to build this product stack that impacts every part of the hiring process.
Mark:
Okay. So, your community at hackajob is built mostly around tech professionals, and I’m wondering what’s their impression of AI and generative AI and all of that kind of stuff? And you’ve got it incorporated into your product, so what are you hearing from the users?
Mark Chaffey:
Yeah, it’s amazing. I literally just got off our company all-hands and it was one of the big topics we were discussing in all-hands today as well. I think there’s generally a level of excitement of what’s happening with generative AI right now. You do get the doomers in the market that will say that this could be terrible for humanity, it could automate all of our jobs, et cetera. But I believe if you’re any knowledge worker, not just working in tech, you need to embrace these tools. You need to really lean into these tools and these tools are going to give you leverage in what you do in your day to day.
So, it’s happening. Products are getting built and code is getting shipped with these solutions. And so, I think there’s just generally a level of excitement. We’ve partnered with an organization called Mindstone this year to host a number of generative AI hackathons and meetups. And honestly, I hadn’t ever seen this level of excitement for a new technology in the 10 years that I’ve been involved in the technology world. So, our users are really excited about it. And I think that any knowledge worker should be really leaning into these solutions.
Mark:
I’m flipping it to the other side of the equation. Could you talk a bit about the inroads that AI’s made into recruiting and talent acquisition technology and not just what hackajob has been up to, but what’s going on?
Mark Chaffey:
Yeah, I was at launch with a large customer yesterday and we were discussing this topic. I think where there’s still a lot of fear, and rightfully so, is any automated decision making happening through AI. I think that’s going to be the final hurdle and perhaps we’ll never even cross that hurdle, and I think that might actually be a good thing. If I’m a TA leader right now, I’d be very skeptical of any solution that’s going to take automated decision making, and that’s because of the potential bias in the training data sets for these models.
I think where are we’ve seen really positive impact across the industry, I think is a bit like I touched on for developers. How do we give internal recruiters more leverage? So, are we enabling the sourcing process to be much, much greater? And where we’re really excited is with LLMs, you fundamentally have a new way to match talent to jobs than what you’ve had before. If you think about so much of matching technology or searching technology and recruitment has been very keyword-based and booting-based. What a lot of these LLMs enable you to do is put context behind the keyword. Now, a computer can read natural text, communicate in natural text, natural language, and I think that gives a lot of opportunity.
So, I think you’re going to see companies really adopt it in the sourcing world, where they’re going to be able to identify talent that they wouldn’t do before. And then naturally, there’s also some really exciting opportunities in the world of content when it comes to recruitment and employer brand, both can you build models that learn from your existing employer brand content and make it far easier to produce new employer branding content? But also, can we use it to help with D&I from a bias perspective? Can we be using some of these LLMs to actually decode bias in job adverts, in branding materials, and then produce recommendations? So, the pace at iteration, the pace of innovation right now is incredibly exciting. I think they’re the kind of areas where we are seeing companies getting really excited in the space right now.
Mark:
Have you seen behaviors change at all among TA teams in the last year, which is when the whole AI rush seems to have really started? Are they approaching recruiting and their challenges differently?
Mark Chaffey:
I think they’re fundamentally challenging the way things are being done, and I think people are looking for insight, and hopefully answers. But I think everybody is approaching this in this very fast-paced moving world. So, really now, every conversation I have with a global head of TA, AI is the number one topic that they want to talk about. “How’s it going to impact my role? How’s it going to impact my team? How we should be leveraging it?” I don’t think there are clear answers to that yet. I think there’s a lot of experiments going on, a lot of testing going on. I don’t think there is clearly like this is what a TA team is going to look like that is supercharged with AI.
And then, there’s also the big focus on compliance. We need to bear in mind the jobs that we all do working in recruitment are very, very important and we need to make sure that we are building AI safety around these models and making sure that we’re not going to end up exaggerating preconceived or preexisting bias in the hiring process. So, I think there is also just a lot of work being done by TA leaders, procurement teams, to really understand the AI safety behind these models. And I think that might be one of the things that potentially slow us down a little bit in the adoption of some of these tools, which again, given the segment and the sector that we play, and I don’t think is necessarily a bad thing.
Mark:
Is this a situation where AI happened to come along at the right time, and so, it’s able to address a lot of the pressing concerns of recruiters? Or has this AI appeared and people said, “Oh, it can do this and it can do that, so let’s do this and that”? In other words, is AI driving all of this or is there a real need out there that’s driving all of this?
Mark Chaffey:
Phenomenal question. So, I want to say a couple of things. Firstly, AI has been around for way longer than this last year, where we’ve got really, really excited. We’ve been using machine learning algorithms and recommendation systems almost since day one of hackajob. And what really happened was there was a paper that got released around the transformer model that then created a lot of these LLMs. And really, it’s this LLM innovation that’s happened over the last year that’s got people really excited. I think there’s an amazing Steve Jobs quote, which is often how I think about this, which is, “The end user of your product does not care about the technology you are using to solve their problem. They just care that you solve their problem 10 times better than what they did before.”
And I think if we think about the last hype cycle was a lot of blockchain and Web 3.0, and I personally never really saw, in our world of TA and recruitment, how that was going to fundamentally make a recruiter’s life 10 times easier. I do think what we’re seeing in AI is it is being driven by AI. I don’t think 18 months ago recruiters were turning around and saying, “Oh, my God, I wish this solution existed.” I think what has happened is the technology leap that has been made over the last 12 months is enabling new solutions to be built, which will make recruiters 10 times more effective. So, I think that AI is being applied to solve preconceived problems, but I don’t think the problems came first. I think it really is the step change with the innovation in AI that’s enabling these problems to be solved in completely new ways.
Mark:
Let me ask the same question about candidates. Mostly, I’m interested in have they changed the way they approach their job search now because they have these new capabilities?
Mark Chaffey:
Yeah, and there’s going to be a lot of challenges here in my opinion. So, we are definitely seeing, and everyone’s seeing this in the industry, there’s probably two different things that are happening from the job search perspective. There are now tools that will build you a CV based off a job description. So, you used to have one CV, and then you might write a cover letter to make it personalized for the job. I think we’re going to move into a world where you’ll have a unique CV for every single job that you apply to, and that will be generated by generative AI. And then, obviously the lazy apply kind of thing that’s gone very viral, of people applying to 5,000 jobs with one click is also a dynamic in this world. And what does that mean for companies? It means that they’re getting way more inbound applications than what they weren’t doing before, but the validity of the data is now really questions and skills assessments are now going to be really a pivotal part of this process.
But how do you ensure your skills assessments are valid? If a candidate is submitting a different CV based on your job description, how do you really know what that candidate is good at? And that’s really exciting because I think that that just challenges employers to think differently about how we assess skills, which I think is only a good thing. So, we are definitely seeing these generative AI tools make the candidate job search process kind of easier, in inverted comments, for the candidate. I think we’re still waiting to see what that ends up meaning for the overall candidate experience and the ultimate quality of hire that we end up producing.
Mark:
Not long ago, you launched a tool that uses AI to scan job descriptions for bias. And I’m wondering how should users, how should employers oversee that tool to make sure that it’s accurate and that some other unforeseen kind of bias or issues creeping in?
Mark Chaffey:
We’ve actually built this into the product. So, what’s really special about what we’ve been able to build is it combines two different elements of hackajob. As I touched on, 80% of our candidates self-disclose their D&I data. So, we actually know somebody’s D&I profile. We then use generative AI to decode bias in a job description and make a recommendation to change. Now, it is on the end user, and there’s a lot of controls a company can put in place on that end user, to approve or decline those changes. But then, what we will automatically do after a 30-day period is use the self-disclosed data to see actually what changes did this literally have on your acceptance rate from different demographics.
So, we’ve already built the safety into the feature where a company, after 30 days, will be able to see it had a positive impact on this demographic, a negative impact on this demographic, it had no impact whatsoever. And is that difference statistically significant? And AI safety is so important in our world. Every company I speak to still has really passionate and strong belief in D&I objectives and making the workforce more diverse and inclusive. I think there’s a really exciting opportunity with AI to do that, but we have to be very, very deliberate on what we’re building. And joining those two dots together means that companies have a very controlled environment to be able to test this in.
Mark:
Can you take us behind the curtain a little bit for a peek? Just how’s your platform doing all of this with AI? Are there certain distinct approaches? Are you just licensing something like ChatGPT? How are you making it happen?
Mark Chaffey:
So, we’re definitely using a combination of the LLMs that are already in the market. What’s really interesting is different LLMs have different strengths, and therefore, we can apply and take the best of each of them. But something that’s always been very unique about hackajob is we collect all of our user’s data as our own first-party data. So, we’re not scraping external data sources, like a number of the solutions out there. This is all data that candidates give us. And we’ve now got hundreds of millions of data points on our users where we’re able to train our own models. And that’s really exciting, because actually, if we’re just building a wrapper around ChatGPT, there’s very little differentiation there, there’s very little defensibility in what we’re doing.
Being able to build our own models based on our own dataset is really where our companies, our customers and our candidates are going to get the most value from what we’re building. And we actually took the decision to rewrite our core matching engine based on the step forward that a lot of these LLMs and AI technologies enable us to do. And we’ve already got customers testing the new products, and we’ll be releasing that in the new year in a very, very exciting way, and I think people will be very impressed by what’s possible now. So, it’s definitely a combination of using some of the open source LLMs that are out there, licensing some of the major LLMs, but then also using our own data set to train our own models, which will give us a really unique edge.
Mark:
My last question actually just came to me, and we’re going back to talk about hackajob specifically. I spent eight years as the managing editor of Dice. And so, I’m wondering what do you think is the difference between hackajob and Dice? You’re both looking for technology candidates, that’s your expertise, that’s the community you’re building. What’s the difference?
Mark Chaffey:
I really think it comes down to the candidate journey. So, we built hackajob in a moment where if you are a software engineer, you get spammed nonstop about mostly irrelevant opportunities. And actually, often when you go and apply for a job directly, you’ll never hear back. That company just won’t ever respond to you. And so, what we were able to do with a completely kind of green space or greenfield architecture is think about how do you place the candidate experience at the heart of what we do and really make what is quite an opaque process for a candidate far more transparent. So, how we do that is a candidate onboards onto hackajob. We capture their job-fit data, so stuff like their salary expectations, seniority, visa status, location preferences. They go through a tech screen, so we understand what they’re really good at and what technologies they want to be using.
And that data will then determine are they relevant for the jobs in our marketplace? If they are, we flip the model. So, rather than a candidate applying to a job, the company applies to the engineer, which creates this really, really magical experience. And engineers end up accepting about 70% of the requests that we get. So, we’ve completely solved the spam problem for the engineer by actually flipping the model and putting them at the heart of the process, rather than relying on them either just getting inundated by companies that don’t meet those requirements or applying for jobs and never hearing back from those companies.
Mark:
Mark, thank you very much. It has been great to talk to you. I’m glad to meet you and I hope we’ll talk again soon.
Mark Chaffey:
Likewise, Mark, it’s been an absolute pleasure. Thank you so much.
Mark:
My guest today has been Mark Chaffey, the CEO of hackajob. And this has been PeopleTech, the podcast of the HCM Technology Report. We’re a publication of RecruitingDaily. We’re also a part of Evergreen Podcasts. To see all of their programs, visit www.evergreenpodcast.com. And to keep up with HR technology, visit the HCM Technology Report every day. We’re the most trusted source of news in the HR tech industry. Find us at www.hcmtechnologyreport.com. I’m Mark Feffer.
Image: iStock
