AI For Sourcing With Gal Almog of Talenya

Intro: This is the Recruiting live podcast, where we look at the strategies behind the world’s best talent acquisition teams. We talk recruiting, sourcing, and talent acquisition. Each week we take one overcomplicated topic and break it down so that your three year old can understand it. Make sense?

Are you ready to take your game to the next level? You’re at the right spot. You are now entering the mind of a hustle. Here’s your host, William Tincup. 

William Tincup: This is William Tincup, and you are listening to the Recruiting Daily Podcast. Today we have Gal on from Telenya, which was recently acquired by Paycore, and we’re gonna be talking all about AI and sourcing kind of that intersection of those two things.

And who better than to talk with than an expert in the subject. So Gal, would you do us a favor, and the audience a favor, and introduce yourself and tell us a little bit about Telenya?

Gal Almog: Uh, thank you for having [00:01:00] me, William. Sure. Uh, we’ve spoken in the past every time it’s a, it’s a new, uh, excitement for me. So appreciate it.

I’m a, I’m a serial entrepreneur who spent the last, uh, 25 years in the recruitment technology industry. Prior to starting tele, I started the PandoLogic, which was acquired in, uh, 2021. About, uh, five and a half years ago, I started tele Pan Logic was in the business of recruitment, uh, advertising programmatic.

We invented that market. And, uh, while I was with the Panda Logic, it really bothered me that the whole market out there of passive job seeker people who don’t go to job sites, to look for a job primitive, uh, methods like keywords on LinkedIn and this and that. and I felt that business is, uh, addressed by hundreds and thousands of, uh, recorded needs and better treatment in terms of, uh, technology.

So I started tele. We [00:02:00] worked independently for a couple of years, and last October we were acquired by peco, which is a public, uh, company in the human capital management market. Since then, we’ve been working on integrating our product into their talent, uh, acquisition, uh, suite of tools, and here we. 

William Tincup: Well, uh, lots to unpack.

Let’s do, first of all, I love Pan Logic, uh, for what it was doing and what it, what it did. Great acquisition as well. And, uh, I also love this marriage of Tel and Pacor because as we were talking pre show, Newton Software. They acquired a wonderful ATS a couple years ago, and I know you’re, you’re integrated with that as well.

I just think Paycor is being, is really progressive in kind of both what they build and also what they acquire and whom they partner with. So it’s just a really great company and they do great with partnerships and then also with acquisitions. So let’s start with kind of the basics with their solution and where you, where they saw the need to have tel in [00:03:00] their recruiting.

Gal Almog: I think that having an ATS without, uh, sourcing capability is like having a sales automation, uh, tool without the marketing leads. Right? Okay. It, it’s nice to have it automates. It makes everything more efficient, but if you don’t have anything to fit into that, it’s worthless. Right? So, uh, everyone who uses ATS knows that they need to go and get candidates into the ats.

It’s relatively easy to get active job seekers. Indeed the zip recorder will scrape your ads, uh, straight from, uh, the ATS and post them and get them back into you to apply to a job. But 85% of the tele pole is not actively looking for a job. It needs to be reached out and it goes back to poor recruiters will sit by computers, enter keywords if through listings, contact them, and that seems, uh, silly to me with all due respect.

Uh, to people who are spending their lifetime, you know, doing [00:04:00] that, we wanted to develop something that would integrate with an ATTs. So imagine you are in the ats, clicking on a button, go source for me. AI will go out, find candidates for you, prioritize them, engage with them, and bring them back into the ats.

They all passive job seekers are not looking for a job. So all of the existing tools out there in the market, LinkedIn, seek out, hire easy, require you to do manual work and therefore they cannot automate the process. But we have. Spend an enormous amount of effort and money on developing something that would do it automatically.

Uh, and that’s what we’re doing with Pcor. So if you buy Pcor at s product or you already have that, uh, every job that goes on, your at t s would automatically pass on to tele and we will reach out to the can. and bring them back into the, at ts so can have passive job [00:05:00] seekers already waiting for you for an interview in your at t s.

The process is really simple. I can talk about it because it is the essence of AI and sourcing. 

William Tincup: Yeah. Let’s, let’s unpack that cuz I think folks that are listening, they’re going to wonder, okay, is this, once we put a job together, let’s put a job, add job description together. Is it going out to the entire.

And then looking at, uh, you know, different databases and different ways to look for candidates. Is it looking at candidates specifically in our own at s and kind of, uh, re-engaging with people that might have applied before, et cetera? So what, take us into the process, so, so that they 

Gal Almog: understand, oh, yeah.

Uh, a job is posted on the ats. Mm-hmm. . Okay. We take the job description and we read it. When I say we, it’s ai right? , and creates a very granular. , which would’ve taken you hours and hours to do using keyword words, bullying strengths, and this and that, right? Okay. We know what’s a job title. We know every [00:06:00] possible synonym to that job title.

We know what’s important, what must have, what’s nice to have. This is all AI knowing from breathing your job description, but there about 30%. Of the requirements that are not listed on the job description, they’re in the head of the recorder, the head of the hiring manager. And if you just use that, you’re gonna go out and, uh, and get candidates who may not be qualified for the job.

Right. So we’ve developed a process under which as soon as the job is, Posted with a job description on the ats. We asked the recorder to rate seven profiles. And when I say rate it thumbs up and thumbs down, uh, with, uh, one reason why you thumbed up or thumbs down that profile. 

William Tincup: And so then it gets smarter.

Gal Almog: it gets smarter. Uh, we also know what candidate, uh, have been, uh, submitted and who got advanced in the hiring fund so we can learn. It’s machine learning, learning about your specific preferences and priorities, [00:07:00] right? And about your company. So when you finish your seven profit rating, you have an option.

You can continue select candidates one by one like you would right on LinkedIn, except no keyword at all. Thumb up. Thumb down. or you can say, okay, I got the, uh, idea. Go and fetch candidates for me, and we’ll take and curate that list based on your feedback and the job description. Prioritize them on quality and on propensity to change jobs so we can reach the candidate, right candidate at the right timing.

and then we reach out to them primarily via email, but also, uh, via LinkedIn and show them the job description and invite them to apply, but we don’t reach. 20 of 30 candidates like they would’ve reached if they did it, uh, manually. Right? We reached 200, 300 qualified candidates. Right. It’s a massive reach out and then they say, I’m interested.

We pushed them back into the ats, 

William Tincup: but it’s cur. What I love about that is [00:08:00] yes, it’s, it’s gonna be whatever the number is, but it’s curated in a sense of. It isn’t just a thumbs up, thumbs down, which I, I I like that you’re asking them why. So it could be a thumbs up. And again, that’s great. If it’s a thumbs down, you the, it’s actually learning.

Cuz you and I, we could be in separate companies hiring the exact same position. We’re hiring, let’s say a, a Java developer, but I am looking at taja de developer differently than you are. And so for that, that personalized, that’s, uh, it’s uh, sourcing at. In the sense of understanding my preferences around that Java developer and your preferences around their hiring manager’s preferences around that Java developer.

So I love the thumbs up thumbs down and why the contextualizing why they like that and, and obviously it’s a learning. The more it learns, the better it gets sharper it gets for the hiring managers and the recruiters. So the, you know, quite one question I, I, I have was when they, when they look at a profile, are [00:09:00] they looking at a unique profile that you’ve built for them?

Are they looking at a u a LinkedIn profile? Like what, what are they, what profile are they looking at? , 

Gal Almog: great question. So it’s a profit profile we build mm-hmm. from multiple sources. Right? So LinkedIn profile may be there, right. But if they have a profile on the GitHub or stakeholder flow, or, or we would take that, you know, digital footprint of different places and put it in.

And, uh, the nice thing, William, is that our search is thinking like a recorder. It’s not you. , put the keyword, delete the keyword, add you know, string. We know what recruiters look at and you know, they look at specific things within the profile and they only spend five to 10 seconds and they typically value the job title more than the value, let’s say the education, right?

And there’s some skill that they like more, uh, if they see the skill within the. of the candidate self description, they value it more than it would value [00:10:00] if, you know it had it on, on the skill in LinkedIn, right? So, uh, the, the most. Effort has been invested in creating AI that thinks like a recorder, and that’s reflected in how we create that, uh, prioritized list of candidates that we go after.

And the nice thing about the ATS integration that, you know, you get a candidates and you, uh, reject them or accept them or move them, and they ended up getting hired. We know which candidate you. And it, we close the circle, so next time we don’t have to ask you all these questions. That’s right. If you come up with the same job, we know who you like, and then we refine the search and it gets smarter every time.

What’s great 

William Tincup: about that is it’s not always just getting smarter, it’s getting faster. getting that. The idea of, again, if it’s a Java, Java developer and you’re fulfilling your hiring another 10 of those, it’s getting smarter that each time you hire one, it’s getting smarter and, and I guess it serves it up and says, what I love about this is it’s taking some of the low value task away [00:11:00] from recruiters and then saying, Hey, listen, if you’re carrying 40 wrecks, You can’t do all of this stuff.

You should. First of all, let’s question whether or not you should carry 40 recs. Put that aside. Okay. But let’s say you’re carrying 40 recs, you can’t do all this stuff, nor should you, uh, I know the audience has a, will have a question around the, the messaging. So once we found the, let’s say 200, uh, that are really, really a great fit for this specific job, you said you can go out through email or LinkedIn.

Is that something that the recruiter pick? Or do you have some probability or analysis there that says this is gonna be a better way to kind of reach these candidates, et cetera? It’s all 

Gal Almog: ai. Yeah. Yeah. So think, think about the share G P T built into a recruiting software. Okay. Say create a message for Java developer, uh, refer to his previous experience to his current job title, right?

Everything that recruiters would not spend hours personalizing the message. Right. Uh, it, it’s all done [00:12:00] automatically. People look at it, they don’t know AI created it. Uh, it’s so personalized that the response rate is much higher, and it’s not one message if you don’t respond. We follow up with a second message, then a third message, right?

We know who clicked. You have an option to say, I’m not interested, and we don’t continue with the thread. So, uh, it is something that, you know, and, and the, in the. You know, how many messages can you craft for a job and how much time do you need to spend writing it and following up and LinkedIn and this and that.

So, uh, within hours of posting a job, you get people interested in interviewing with you. , right. 

William Tincup: Dumb question. Alert. Miguel, does it, does it, uh, cuz it personalizes to their profile, which is, uh, obviously, but does it also personalize to the job description itself? Like, hey, this is exactly what they’re looking for.

You have 12 years of that experience, like, The words one. 

Gal Almog: [00:13:00] Both. Both. Both. Of the job description, they say we’re looking for someone with, you know, three years in Python development. Yeah. Uh, based on what I see on your profile, imagine all of that turning into video in the future, right? So someone will talk to you and tell you why.

You need to speak with the company. That would look completely natural. I was 

William Tincup: thinking if, uh, about conversational bots and conversational AI in that way of either using it. If somebody wants to chat right then, so, so we send ’em a message, but they want to chat. Obviously we wouldn’t have a recruiter do that.

You could do that through conversational ai. You’re probably already thinking that thought and moving in that direction. Uh, 

Gal Almog: definitely. But I think, you know, conversational bots are a step behind because they need to be programmed with the answers. While if you have ai, you know, it could be strictly pre uh, dialogue.

William Tincup: I love that. So just a quick question around Tel and Paycor. Can folks, can they work with Teligia if they’re not a Paycor [00:14:00] customer, or 

do 

Gal Almog: you Yes. So the Paycor, uh, kept our independent, uh, operation selling to companies who are not on a Paycor platform. Okay. And that’s absolutely fine. We’re integrated with almost every possibility.

And, and in fact, you cannot work with us unless you connect with the ATS because the experience is completely 

William Tincup: different. Well, you want that closed loop and you want that learning to happen too. Right. So it wouldn’t make sense. Yeah. We don’t just to dump candidates into a a thing and then not have insight into it.

Yeah. 

Gal Almog: Not only that, but for example, if you are already speaking with someone, we don’t want to connect with them. So we need to gdu them from our list and things like that. 

William Tincup: Right, right, right, right. So why, I mean, first of all, I love the solution. You know, I love the solution. We did a podcast a hundred years ago, so, you know, I love to linea.

Um, why, why do people say no? Like, what’s the op? Um, I, I say the opposition, but what’s the, when, when people say no to you, why, what reasons do they get ? [00:15:00] I’m curious. 

Gal Almog: I ask myself the same question, if you can save 80% of your time Yeah. And you can get candidates that, uh, you know, traditional tools will miss, why wouldn’t you want to have it?

I’ll tell you, and I, I learn new things every day. I’ll tell you what I learn. First of all, people like LinkedIn. 

William Tincup: Far more than they admit. Far more than they admit. Cause if you, if you look on social, everyone hates LinkedIn. Everyone, every recruiter you talk to, I hate LinkedIn. They hate how much it costs.

They all, they hate, hate, hate, hate. 

Gal Almog: It’s a monopoly, but in fact LinkedIn is a pretty good product. Yes. Okay. Yep. And, uh, it’s primarily social network. So they love the social network aspect, which doesn’t come if you buy, seek out or are, uh, easy. Okay? You have to have LinkedIn to connect with people. Now imagine that you build your career on perfecting bullying, search, and keywords and connecting and crafting messages.[00:16:00] 

People are scared. They, you know, think about what will happen to everyone when things like or G P T would come and, and threaten complete industries, unless they embrace it, they will be thrown to the sideline. That’s one reason that we’re seeing. The other thing is that oftentimes until they really try using us for a couple months, they don’t realize the benefits in full.

Right? Right. So we typically do a trial, you know, we let you, uh, post a couple of jobs so you can see the benefit. Right. Uh, and obviously they give us the hardest to appeal job thinking, oh yeah, yeah, yeah. Purple 

William Tincup: squirrel, purple forest. Yeah. Yeah, yeah, 

Gal Almog: yeah. But in reality, we’re not as smart as people. I have to admit.

Okay. Right, right. So, uh, if, if you select each and every candidate manually, the quality of your candidates will be. Perfect. Okay. Right. You, you’re not gonna take a chance with someone who’s missing something [00:17:00] on their profit because you don’t wanna waste your time. Right? That’s right. And you’re limited by the, uh, number of candidates that match exactly what you want.

We do something different. We say broaden the scope. You never know what people are missing and not missing, and we are gonna reach out and contact, you know, two, 300 candidate. Some of them may be qualified and some of them may not be qualified because AI selected them and they selected them based on the size of the talent pool, whatever it is.

Okay? So they need to embrace the idea that there’s a tool. , which not the tool, A platform that would get them, let’s say 60% qualified and 40% non-qualified. And over time we would learn and improve those, those percentages. But that’s a completely new notion, right? For them to understand. It’s not a job board where everyone, you know, 99% would be disqualified, but you know, you cannot stop people from applying.

It’s not. Searching [00:18:00] on LinkedIn, it’s something in between and it takes time for people to embrace that. Well, I think it’s, you 

William Tincup: know, like the Amazon experience, if you log into Amazon, um, like in incognito mode and you don’t log into your account, uh, it doesn’t know you, so it doesn’t know what to serve up to you.

Right? But if you log into your Prime account, an account that you’ve used for years and years, and. . It knows you better than you know yourself. You know what, what I’m saying? . 

Gal Almog: Okay, bye. 

William Tincup: It serves up the things. That’s why people, it fascinates me cuz I’m, I do a lot of stuff on like, Instagram, Facebook, whatever, and like people are like, they’re, they’re upset that, that, that the ads are served up.

It’s like they’re looking at Google, they’re looking at this and they’re listening in an Alexa, they’re doing all this. I’m like, I actually like that because the. Because the ads that they serve, which they have to serve up to make their, you know, their models work, their ads are actually right on point.

Like for me, like they serve up artists, they serve up things that I like. I’m like, I don’t like that they [00:19:00] know all this stuff. I don’t care. 

Gal Almog: But imagine, imagine you would, uh, use a tool and you will, it would turn out that what you do now for your living right, is done, you know, somewhat better by ai. Okay.

Alright. , that that is something that, uh, will take a mindset Yeah. Uh, shift for them to embrace it and use it, uh, properly. 

William Tincup: I, I, I, I, I, I think so. I think the more they use it, the more they’ll understand how it actually helps them do their job. Like there’s always gonna be a human component to this, uh, to, to hiring, cuz it’s a very human thing to do.

However, , there’s a bunch of stuff that we’ve been doing through the years that we just don’t need to do. There’s a better way to do that. Um, uh, the last thing I want to kind of cover with you is kind of a, I know you get this question. I know you, you’ve gotten this question, uh, for years around adverse impact and, um, probably along the lines of ethical AI and auditing ai, et cetera.

So [00:20:00] what’s your, what’s your take on just kind of how do the ma the, with machine learning, the technology will get smart. and, and more appropriate to that particular company in that job, et cetera. But how do you go back and look at your AI either independently or do you have your team look at it, make sure that there’s nothing happening in it, uh, with its recommendations, et cetera, that you just don’t like?

Like how do you, how do you kind of keep your finger on the pulse of that? 

Gal Almog: The question that we ask ourself continuously, first of all, you know, we have a component. That, uh, will boost diversity right, uh, in your pipeline. So we are working on the opposite, uh, way to give diverse talent a fair chance to be considered.

Because right now they don’t, especially if you have to look at pictures and, uh, determine who is who. So we. Uh, don’t show your pictures when you do your thumbs up and thumbs down. We don’t show a picture. We don’t show your names. So on a diversity side, I think [00:21:00] we’re probably the best solution out there in the market to, uh, and we are plan to use it only for diversity hiring.

Right. Okay. Then, uh, none of the information that, uh, we’re using is personal information, so we don’t use any of that. So there’s no way for the AI. And we don’t use, uh, you know, in, in Europe for example, uh, we don’t show the clients any personal information, right? Unless the candidate has consented right to, uh, speak with the company.

So, uh, we, we use, uh, the, uh, academia and, uh, research to continuously look at the algorithm to see whether the algorithm is doing something that is inappropriate. But the biggest. Challenge for AI is that, uh, it would copy bias or behavior of humans, which keeps us all up at night. Unethical. Okay. [00:22:00] Yeah. Uh, we, we we’re just giving people an opportunity to.

To, uh, take another job. And I think it’s a good thing to have and, uh, whatever technology can be used for that, I think we are, you know, doing a better job than human in that respect, where we don’t look in any personal, gender, ethnicity or anything like that to choose the candidates and, uh, it’s completely.

Independent of any of that. 

William Tincup: Yeah. Skills and experience and potentiality, uh, and things like that because you’re looking at tangential skills and some of the synonyms that aren’t in the job description. I, I lied. I do actually want to ask you one more question and it’s, we’re driving folks to apply. Uh, do you, do you see Tel in the future going further past apply, like into scheduling or anything like that?

Like once we’ve got them applied, can we, can we do something in the future after they apply? 

Gal Almog: Yeah. Peco already has that scheduling, so we don’t need Okay, 

William Tincup: good, 

Gal Almog: good, good. [00:23:00] But there, there would be a lot of things that we can do once we’re embedded in the a t s and. We will announce a co, but the idea is really to introduce AI to a market that has been, you know, not so developed, uh, with, uh, technology like AI’s and, uh, a company that will not do that, will not be able to, to succeed.

So, right. Uh, the acquisition of, uh, Pega of tele is the first step in getting AI into anything. Picker will be doing in the 

William Tincup: future. I lied again. I have one more question. I, I won’t, this will be the last one I promised. Probably more time. . So. Have your customers already asked you about internal mobility in terms of This is Gal, this is wonderful.

It helps us with everything on the front end of getting new people into the system. We have kind of a quandary around [00:24:00] people already. You know, Siemens has what, 120,000 employees, you know, they’ve got open recs as well. I could see a business case for turning this inward as well and using it that way.

But, but have your customers asked you, uh, about that or do you see a 

Gal Almog: few? They have. They have, and we’re already thinking about that capability. I’ll tell you, I’ll tell you what we have. Okay? Okay. We have algorithm that can predict the nest career path of a person. It’s also ai, right? So if you have a person, if there’s an opening.

and that person is doing exactly the same thing, that it’s a no-brainer. Okay? But if that person is one job behind that opening, the hiring management may not know. The recruitment may not know, but we know, right? So we can say before you go out and consider someone exactly from the same type of job, you know, more money and this and that, give an opportunity to [00:25:00] someone from the inside to do it.

It’s a very. Uh, natural extension of what we do and we’re, we’re building it into the product together with paperwork. 

William Tincup: Well, I, I can’t wait. First of all, come back on cuz I wanna hear all about that once you’re, once you’re close to launch. Thank you for carving out Time gal. Every time we talk I get smarter.

So thank you so much for coming on the show. Alright. Right. 

Gal Almog: Always a pleasure over there. 

William Tincup: All right, thank you. My. 

Gal Almog: Thank you. Take care. 

Intro: You’ve been listening to the Recruiting Live podcast by recruiting daily. Check out the latest industry podcast, webinars, articles, and [email protected]