Introduction

I would love to know the audience here who I'm talking to. So if you can raise your hands if you are a recruiter, an HR professional, a hiring manager, or who has hired before. If you can raise your hands.

One, two, three. Okay. A lot fewer than I thought.

And how many of you are actively job hunting right now?

Okay, a bunch more. So that means I'm gonna have to change the strategy and tweak what I wanna talk about.

I was gonna talk about recruitment generally and hiring and how AI is affecting recruitment as a space. Not just disrupting, but affecting and how that effect sort of plays out.

The Role of AI in Recruitment

From Resumes to Application Spamming

If you are a candidate, or if you have used any AI tools in your job hunting, you will know there are a bunch of AI tools out there. There are newer ones coming up every week that can help you create resumes, do keyword optimization for your ATS, write cover letters as well, as Stefano was talking about.

But that's not where it stops.

I hope you have seen some of those tools, but if you haven't, I can show you how to use ChatGPT for that. 1But there are also tools that can help you spam job applications.

So it will, on your behalf, look through the jobs that you're interested in, and it will send applications. It will tailor the resume for you and just send it out.

And you can also optimize your messages, LinkedIn messages that you can send to recruiters, as well as emails. So there's a lot of spam generally being generated from the candidate side.

Impact on Recruiters

Now, that's not really bad. I know a bunch of you are candidates here. So I'm going to have to use the right words here.

That's something that you need to know, you need to use anyway, AI in your hiring process, sorry, in your job hunt. Without it, it's very tedious, there are lots of job openings out there, and the more jobs you apply to, the better chances you have to apply anyway, right?

But that has an effect on the recruiters. If you think from the recruiter's perspective, now they're getting thousands of resumes per job opening. Before they would get maybe 100, 150,

Even a couple of weeks ago, I opened several roles, and for each of them, I got over 1,000 applicants. And I don't want to go through all the resumes and read the cover letters. And oftentimes, you have your GitHub, maybe LinkedIn in there, or some other portfolios, and I'm going to have to do more research.

So it's not really the... the ecosystem that we are generating is not really friendly for the recruiters. So we are kind of pushing recruiters to start using AI as well.

Now you can argue whether that's fair practice, a human should review all of our applications, but that's just not feasible anymore. So the more influx there is of applicants, the more there is a need for AI.

And that's where I think I'm going to sound more preachy, as Stefano was talking about. There's massive volumes of resumes that you have to screen.

You have to adjust the strategy of your recruitment. Now, maybe you had a hiring manager who speaks with five candidates, have quick interviews, and you move on. Now you need to add technology to automatically schedule these calls, make sure you're reaching out to a bunch of people, and there's email communication that is automated.

There's a bunch of stuff that you need to do to optimize your hiring process. Otherwise, it would take months to fill the roles.

AI Tools for Hiring Managers

So I'm gonna quickly show you, I might not discuss too much about this because I think it's not very relevant to the audience, but I'll show you a little bit of the AI tool that we have built. Okay, one second, never.

Okay, so this is like a very simple user friendly tool, but essentially what it does is recruiters can put their job description on this product and The AI sort of parses all the requirements that the companies are looking for. And you can adjust the weights of these criteria as well.

Like, hey, I'm looking for maybe CRM experience for my salespeople. And what AI would do then is screen through all the resumes and rank them. They quantify all the different criteria.

Like, hey, you have sales experience, you have customer success, maybe you know HubSpot, those kind of things. And it basically generates a more comprehensive report around how each of these candidates have performed on each of these categories, give them a score, and the end result that you get is basically a ranking of these candidates.

Now that might sound a little bit alarming. I'll maybe adjust my talk to talk more about the candidate experience side. But All it is doing is reducing the time that recruiters have to spend screening through a bunch of resumes that maybe don't qualify for certain reasons and just focus on the top candidates that it has pulled up.

Now, sometimes you might have workflows where a company is looking for someone who has sales experience selling maybe HR technology to SMBs in North America. That's not something that you can just look at the keywords and resumes and pull information from. It needs to be agentic in some sense. So what AI would do is search the companies that you have worked at before and figure out that information and use that to evaluate the candidates.

So that's like one of the tools that AI is enabling companies. And there are a couple other tools that companies are using as well that is similar to this. Essentially, what happens next is you would set up calls, interviews with these people.

Enhancing Recruitment Efficiency

You would have some note-taking mechanisms. There are a bunch of note-takers as well if you have had meetings. They sit in there, take notes. Those notes will also be processed in the same mechanism. And as you go through your hiring stage from one to another, you would always have AI evaluating these candidates and give you a more quantified score of how people have performed.

This reduces recruiter fatigue as well as recency bias.

Addressing Candidate Concerns

Now, you might think as a candidate this is a little bit unfair that my resume is being read by AI. And that's where the opportunity is that we need to improve our AIs.

A lot of the resumes that we are receiving from candidates are already built by AI, right? So they are optimized for the right keywords and that just bypasses the existing system that recruiters already have.

And that puts other candidates at disadvantage that are being more maybe truthful or they haven't disclosed much information on their resumes. So how do we build an AI that is that can do a lot more extensive research, can actually go through your maybe LinkedIn, other portfolios, and actually scans through and takes in a lot more information.

So instead of a recruiter just skimming through your resume, which is what traditionally happened, Now AI can do a much deeper research on every candidate and surface the ones that actually are right suited.

So instead of seeing AI as maybe negative for your experience, think of it as a positive. It is gathering a lot more information for the recruiters and helping you with that.

Transparency in AI Processes

Let's see. I think I would maybe pause here and do a quick sanity check if you guys have any questions so far.

Yes.

Okay, let me go back.

So behind all of these, like it says proficiency in WordPress, we have like a really extensive rubric that we automatically generate behind the scenes. And that rubric basically tells the AI how to read a resume, what pieces of information to take. So we anonymize the resume fully and we look through these keywords here, basically.

But what it might do is also figure out some adjacent skills. So instead of WordPress, it might find Wix or Squarespace or Webflow, and it would be able to give you some scoring on that as well, which is something that you wouldn't find in a typical ATS. So there is a little bit more rubric behind it, and we are still working on showcasing that and giving you more control over that.

You can always go back and tell the AI, like, hey, I actually don't care about certain things. I want this category to be a little bit more weighted. And you can let the AI know, and the AI will rank the candidates on the fly and give you the next top candidates.

So the final decision is still being made by you, which just makes the process a lot more efficient. And now you can spend more time focusing on your strategy and how to interview the people.

Yes?

Keyword Filtering and Resume Optimization

So I mean is it fair to say that you know AI is kind of improving on what's kind of Yeah, you can think of it as like a 2.0 version of it, but a lot of recruiters don't use those keyword filtering anyway. So while there is a lot of, you know, thought leadership around like, oh, add keywords in your resumes, recruiters don't use that functionality.

Because if, for example, WordPress, right? If a candidate hasn't mentioned WordPress in their resume, that doesn't mean they don't know. They might have mentioned an alternative tool for this.

And keyword search is not really good. You might misspell the word, or maybe you have an alternative. Instead of saying JavaScript, you have JS or something like that.

So while it is used in some sense, it's not typically used a lot.

yes either of you I was just going to say that.

So the feedback you would have gotten before is just stick to one page. If you go beyond one page, that's bad. I think that only applies to a human reviewer.

Companies have not adopted this AI as rapidly yet, but it will eventually get there. So be mindful of that.

I think going to two pages, but be more descriptive around what you have done and explaining a little bit more and explaining your portfolios more as well will definitely help you more.

Yes. . I think there are already I was building my resume for searching for a job. I was practically building my resume for each job. Especially if I was interested in a company, I would stay early in my resume using AI to be at better words applied to that specific job.

Yeah, there are a bunch of tools already that let you optimize at scale or tailor at scale and then apply. If you're using those tools, I think you'll be a little bit at advantage today.

Maybe not always, but there are a bunch of candidates that have only one resume that apply to all the jobs. If you have that, you might be slightly disadvantaged, but that's just, I think, luck at this point.

Challenges and Ethical Considerations

Yeah. Do you have a question? Yeah.

Yeah, I think that there are some things that cannot be quantified. So for example, I'm looking for somebody who is a go-getter. How do you evaluate that?

But there are some others. For example, you have led a sales team for three years. I think that's quantifiable. When it comes to those attributes, AI does a very consistent job.

So that's one. Second, I think you are very technical. We are using the highest model possible. And we are evaluating every criteria as a separate API call. And even then, there are inconsistencies. So AI is not there yet that can do a great job altogether.

We are currently training our own model, and we are hoping that that will produce more consistency. Because there's a question that you have to ask is, do you need a reasoning model for this, or do you just want a text processor? So we're still doing that, but what we have built so far, we run several batches. It's very consistent, and we do let the recruiter know as well that there's going to be some level of error in there.

Any other questions? Yes.

Addressing Bias in AI

What are you doing for convoy bias training? um sorry are we using their data to train ai that they only have one specific type of person that they were selecting using the ai for that

So right now there are only like two main strategies. Anonymization, first of all, and not just anonymizing the name and maybe email, which you wouldn't send to the AI anyway, but also anonymizing the kind of companies that you have worked with or maybe some location. So the level of anonymization is something that we're still playing with, but we do at a basic level at least name, location, and those kind of things.

And the second is... variation of data. So you would have three records of the same type with different variations and see how AI performs, and then you would do an average of that.

So there are a couple of strategies like that, but I think there's still more innovation happening to figure out how to test this and how to get data around this. Do you think it's possible to make this smaller to be able to look at both of the kinds of models? Yeah, that comes up a lot as well. No, actually, you'd be surprised.

There are a lot of people who are truthful on their resumes. And they do sometimes score lower. And that's something that we are building.

So AI can also detect whether AI is being used and how the resume is structured. So there are certain things that we are building on top of it. But this is just the first pass.

AI in Interviews and Candidate Evaluation

I think the expectation would be to have interview process as well. There are new AI interviewers coming out as well in the market. So I think if you're interviewing

10 people, a human can do a good job at it. But if you are interviewing 500, then you would need some kind of AI. And I think AI can detect lying and bad behavior on that step.

So I have a question about those ratings. Do you compare it on other applicants? So meaning like, you know, you're like, you know, top candidates among the applicants.

Um, not yet. Uh, we are still exploring if that is the right, uh, thing to do. Um, because like some candidates can influence other candidates. I, I, I'm not sure how to quantify that.

Uh, but the approach we have taken so far is just very data driven. We just present you the data and you decide at the end of the day and it's very individual, uh, individual based and individual criterias.

And there are also like some more optimizations we have made, like have you provided evidence of this or are you just throwing keywords? Because a lot of people have like the skill section and they just write random things in there even though they don't have the skills. So the AI tries to map the things like, hey, what are you trying to do in your resume, basically?

So to his point, I think if you have more information and if you are able to describe more accurately, but if you can provide more description in your resumes, I think that can go a lot further.

AI Application and Adaptation

Let's look at this problem, not from the view of the company. Can you mention any studies or researches that show that it's really effective? Because when talking about sorting the people, if you want the best candidate, you need to look at the type of . And many things, and even if you just sort it by some scores,

So AI is very new, right? So there isn't a lot of research done just naturally on this.

But the way we onboard companies to use our product, I don't know about other companies, we try to run a few batches of their previous hires. So they already have that data of how they performed, which people they thought were suitable to interview at the first pass. and who ended up being hired for those roles.

So we run a couple batches there and figure out, okay, is our AI giving the same indication as what happened before? If that's the case, it's very likely going to perform in a similar manner later on.

But there is a caveat of AI not doing a good job. So we do encourage our recruiters to be mindful that this is still a beta phase and just use it or take these numbers for a grain of salt.

Customizing AI for Recruitment Needs

The couple of things that we are adding is recruiter feedback as well.

They can upload an ideal candidate profile, like this is the kind of person I'm looking for, so AI will try to adjust its evaluation according to that. That's one. And then if they think, hey, Photoshop, this should not be as high of a score, it should be lower, the recruiter can give that feedback and it will re-evaluate the entire list.

AI is choosing candidates who made their resume with AI. So it's also your proficiency in AI and in AI learning and cooperating with it. But if you want the person on some particular position, it's not about how he's dealing with AI.

It's sometimes a little bit broader Yeah, so we are building those mechanisms to check whether AI was used and how optimized a certain resume is. But a lot of times, when you interview those people, they do have some baseline skills on that.

You can always pick more people and interview them, but it still gives you some reference of who might be worthy of interviewing. So that's one point. And as I said earlier, candidates are going to use AI anyway.

Scaling Recruitment Efforts

So it's like an ecosystem problem. You're going to get thousands of applications for a role that you need to hire within two months, right? Ideally, a couple of weeks, but it takes time. To combat that, you have to use some kind of tool to perform at scale. It might not do 100% job, but even like 80% can get you some decent candidates.

All right, you got a question? Yeah. Yeah.

Future Opportunities with AI

I was going to talk about all those recruiter things, but not a lot of recruiters are here. But comp analysis is also a great one.

Where you're hiring and what that comp structure looks like, how competitive this space is, how difficult it was to find this candidate, how much you should pay. So there is AI being built around those things. Right now, people are using Pave.

There's some data tool, but AI is coming into that place. So right after this interview, you would get into negotiations where there's some kind of transparency where candidates can also let the company know, like, hey, I'm also looking at three other offers. And the companies can also figure out like, okay, these are the few candidates that I can potentially provide offer to.

This is the compensation and sort of help bridge the gap of negotiations, right? So I think there's a lot of opportunity to improve the overall experience on both candidate and recruiter side with AI.

But yeah, so far there is nothing.

Yeah, I think I might take one more and then I'll have to pass the baton, yeah.

As in the competitive advantage? Competitive advantage?

Competitive Advantage

So our idea is to build AI tools to help with the entire recruiting ecosystem, not just for recruiters, but also candidates.

So we can provide feedback to candidates as well, like how they can improve certain things, provide them resources around, like, this is where you need to upskill yourself, and this is how you can present yourself during interviews. So this is going to be a lot more broader.

Improving Candidate and Recruiter Experience

Conclusion

But this specifically, the candidate screening, competes with your traditional ATS tools that have basic keyword systems and they are like the previous generation of technology so it's just like this is new tech there's really no competitor in this space at the moment okay thanks a lot you've been a great audience and I will log out from here

Yeah, awesome. Thank you.

And by the way, my name is Ishan. I didn't introduce myself.

Thanks.