AI for HR Leaders: Smarter Sourcing, Stronger Retention

Hello, and welcome everyone to today's Achieve Live webcast. My name is Zach Doms, President of Achieve, and as your community leader, I appreciate you all being here with us. I'm so excited to keep unpacking new frameworks, research, best practices, stories from the field on how we can continue building a better world of work. So we're very lucky to have some amazing experts here with us today. Before we get started, just a couple housekeeping items. Yes, today is being recorded. We'll share the resources and the session with you afterwards if you are like me and you want to listen back and take more notes. We'll also be sharing SHRM and HRCI credits at the end of the session, so make sure you take advantage of those and stay to the end to get that. And then we'll be sharing some additional research throughout this, so make sure you're taking notes. You're live with us here today. I would love to see, if you're with us live, where you're calling in from. We already see someone in the Philippines. If you're at in the chat right now, let us know what part of the world you're in. I always love seeing that type of footprint with us. And today, we have a very special and challenging topic that we're going to unpack. If you have any involvement in the world of recruiting and hiring and talent acquisition, you know how hard it's become to find the right talent and access who are the right fit and where actually some of your top candidates are. I've hired a couple roles myself, and I'm getting hundreds to thousands of applicants per role. So I'm always looking for more ways of how do we actually source and find talent. Like, what are the best ways to understand who's going to be perfect for this role when you're rolling through a sea of AI-generated resumes and people automating their submissions into your job hiring profile. So I'm excited to sift through that today with you all, and we're joined by our partners at Indeed, who are the experts in this space. They obviously have been working in this space for so long, and now they've developed even more tools and research and frameworks to actually understand how to do that at scale for your company. So that being said, let's welcome these three amazing individuals. We got Dan, Kyle, and Yael joining us to really be the experts on this. So I'm going to stop sharing and welcome these individuals to our community. Dan, Kyle, and Yael, thanks for joining us in the network today. It's a pleasure to host you. And to get us kicked off, I'll hand it over to you, Dan. Thank you so much, Zach. It's a pleasure to be here with everyone again, another Achieve webinar. Every recruiter knows the feeling that Zach was describing. You're trying to find the perfect candidate. But will someone else find them first, right? The war for talent never stops, never sleeps. And this fear that recruiters have continues. 72% of hiring managers say they're missing out on top talent simply because there's too much volume to sift through, and 93% have lost a strong candidate because their process took way too long. It takes way too long from the recruiter end, and it also takes way too long if you're the actual job seeker. But the talent is out there, and the problem is finding them in time because other people are looking for the same talent as you. It's very competitive out there. And we're going to dig into that challenge today, and more importantly, what you can do about it. Indeed is here to introduce Sourcing Assistant. It's a new AI feature within the Indeed Smart Sourcing Suite. Say that 10 times fast. Smart Sourcing Suite. Smart Sourcing Suite, right? But it has a nice ring to it. And it helps recruiters discover and engage with top talent faster without sacrificing quality or control. And I'm joined with our two panelists today. The first is Kyle MK, is a senior talent strategy advisor at Indeed and one of the people closest to thinking and strategy of employers today. Kyle, welcome. It's so good to be here, Dan. Thanks. And Yael Smith is director of talent acquisition and recruiting at Stern At Home Therapy, a company that has put this tool to work in the real world. Yael, great to have you here as well. Thanks, Dan. Happy to be here. And really the idea here with this panel is we're going to show you the new tool and how it's useful, as well as how it is applied to a company like Stern At Home Therapy. And I think that that's really useful because we're all trying to improve sourcing and, and this is definitely one way to do that. So we're going to go back and forth between the panelists so that we can extract that type of knowledge and application so you can benefit. And throughout this webinar, if you have any questions or thoughts, please use the chat or Q&A box. I have both of them set up. So whatever you feel more comfortable doing, and we'll be sure to answer those questions as we go through this. So we're going to start with Kyle. The numbers in this space are striking, like I shared. So especially that 93% of employers have lost this top talent because things have just taken way too long. So from your vantage point, how did those structural challenges shape the case for building Sourcing Assistant? Yeah, good question. I think they shaped them completely. Sourcing Assistant came out of one primary realization, which was that the talent is out there, but the recruiters just can't get to it fast enough. And- I think recruiters have known for many years now that sourcing produces the best candidates and the highest engagement, but it's also the most manual thing in the job, and it's usually the first thing that falls off when there's no time left or when the day fills up. So, it's kind of funny that the most effective method to find great talent is also the one that nobody has time to do. But, in the meantime, the volume still keeps rising and you've got the same recruiter with more applications and less time, and every one of those applications, they represent a real human being that deserves some sort of response and some sort of connection. And with Sourcing Assistant, it's pretty neat. At least in the early customers that have used it, both in the beta program and since it's launched, they've shaved off on average seven hours per week. That's a full day of just manual searching. And they've been closing those roles, each role about six days faster than they would without it. So, it's a pretty neat tool. Yeah. And Gal is one of the customers that Kyle is probably alluding to. Before we get into the solution, grounding us in reality, give us a picture of what recruiting looked like at Stirn at Home before you started using Sourcing Assistant. Yeah. I mean, what Kyle said earlier was basically a perfect description. A lot of roles, not enough time for the sourcing. So, I would really have to manage my time efficiently and select the highest priority roles to source for because everything was manual. I had to find the person through a long search, and then craft a specialized message that wasn't too broad that people think it's spam and don't read it, get no responses. So it was very challenging being the only recruiter here at Stirn to juggle all that, as well as with the other admin tasks and, of course, the key thing, which is having conversations with active looking candidates. So when I was presented with this sourcing AI tool, I felt like my prayers were answered. This was something we've been looking for for months, and I can't wait to get into further details on how it helped us. And back to you, Kyle. Walk us through what it looks like when a recruiter sets up Sourcing Assistant for the first time. What is the experience like, and what does the recruiter need to bring to the table? Yeah. The experience is basically, it's three and a half steps. And the thing that you need to bring is just a clear job description. That's essentially the steering wheel. But the first step is just you turn it on from Smart Sourcing homepage. You give it plain language instruction, so the must-haves, any other criteria that you might have, and you basically would describe it the way that you would describe it to a colleague, describe the role to a colleague. But there's no Boolean strings, which is a huge benefit. I hate Boolean. The half-step would be just training it. So this is kind of an optional thing. It brings back initial matches, and you correct it, and it adjusts. So you can choose to do that or not. Step two is you're just approving the outreach. So what it does is it drafts a template that you edit, or you can of course drop in your own, and you're setting the tone and the volume thresholds for these outreach. And then step three is you start sourcing. You watch the outreach go out, you monitor the responses, and see the analytics on either the manual outreach versus the automatic side by side, and you can adjust at any time. But basically at every stage of this, you're choosing whether it's manual or assisted, and that's kind of how it's built. It's built as an assistant, so that way you've got control the entire time, and that one real dependency is the job description. And Gal, back to you. Recruiting licensed therapists is a credentialed, specialized search. There's real nuance in what makes someone qualified for a role. How do you describe what you needed to the assistant, and how quickly did it understand those requirements? Yeah. So when you start the whole process, you will assign a job, and from the assigned job, the AI will scan the job description and pull a bunch of requirements, nice to haves, things like that. And there's a little tab over to the left which would allow you to make changes and edit if there's a few things that it didn't pull that are required. It did a great job. I really didn't need to edit anything. And then from there, like Kyle said, it will give you a sample of five matches that it just wants to make sure it has the right understanding. From there, you will just check to make sure that the license is accurate to the state for the job. That's a very key point. And for example, we hire in New York City and right across the Hudson in New Jersey. Some people commute from New Jersey into New York, but only have the license for one. So it's very key to check that and correct the AI for it, and it learned it very quickly to scan it, and it would actually populate a message saying, "This candidate states has an active license in New York or New Jersey." That way, it assured me that the tool was working accurately. Yeah. It's smart to double-check, and I'm glad it worked really well for you. Kyle, with 370 million profiles available for sourcing, how does Indeed distinguish between people who are actively looking versus people who haven't been on the platform in a while, and how does the assistant handle both groups? A really great question. We hear this a lot. So there's two layers, essentially. Job seekers opt in by selecting something on their profile. It's basically a checkbox that says, "Let employers find you." So that's one signal. And then the assistant reads activity signals to figure out who's genuinely in motion right now. So those signals include the recency of them updating or uploading a resume, or maybe recent activity on the platform itself. And we're looking at a taxonomy of a little more than 160 skillsets. So the combination between those two matters because qualified and available are two completely different things, and obviously the perfect match who hasn't touched their resume in three years is typically a dead end. So those are the two primary ways that we're able to discern whether or not someone is active or passive, and then still the recruiter gets to choose whether or not the outreach goes out to a passive candidate. I think that's a really good point you bring up, active versus passive candidates. From my experience, I've heard that passive candidates are actually looked highly upon as well because they're in it, right? They're not actively looking and maybe unemployed. Do you ever find that people who are passive might become active again if there's a reach-out or some reason to? Yeah, I think far and wide, people like being pursued. So- Yeah ... when they think they're reached out to, it might get some folks on the hook that weren't typically looking. But in this specific market, at least the one that we've been in for the last year or year and a half or so, we've heard about job hugging quite a bit. There's this passive talent. These are individuals who have chosen not to look because they feel that the market is too unstable- Mm ... or unstable to go look. So I think right now the passive candidate is an interesting unicorn because some folks are interested in new opportunities and they're not looking because of the market, and others are like, "I'm actually very happy with where I am and I don't want to move." So yeah, it's kind of why it's a gamble when you reach out to those folks. In the past, I think it was likely that you would get a response. Now, I think it's maybe about a 50/50 shot. That makes sense. Yael, back to you. What did your pipeline look like in the first week or two using the tool? Were the quality and volume what you expected? Yeah. I was very impressed and astounded in the results. Within the first week or two, we actually got six qualified candidates for a range of three positions all in different locations. And I probably set the outreach to maybe 20 to 30 outreaches a day and kind of had it capped. And those are people who most likely I would never have contacted because I was busy doing something else. So it allows me to focus on other areas while it works on other priority areas, allowing us just to increase the top of funnel and talk to those much needed and seeked out candidates. Yeah. Time saving, but not at the cost of quality, which I think is really important because obviously anyone wants to save time. Everyone's so busy all the time, but you still are responsible of getting the best candidates. And you said six. Six seems like a lot for a very specialized position where credentials are needed. Yeah, 100%. The field is very competitive. I know with my talks with my Indeed rep over the years, it's in the 98th, 99th percentile, so- Wow ... if you can get someone who wasn't looking, like the passive candidate, interested, that's amazing. And the velocity and just needing to talk to the individual quickly is necessary. And Indeed AI just helps that be possible. Yeah. And Kyle, the outreach aspect I think is really important. As Yael's kind of pointing out too, you got to get back to all these people, right? And you don't have hours and hours and hours and hours to be able to do that. The assistant drafts personalized outreach on behalf of employers, right? And I think that that's a really, really important value add. How do you make sure the messaging still feels human or reflects the employer's brand rather than reading from a form letter or people just knowing it's AI slop? Yeah. This is something that's very important to me personally because- Yeah ... I am sure everybody has received the messages in LinkedIn where it's like, "Your background really caught my eye." And they don't know what your background is. I'd say there's three things here. It does personalize it based off of the candidate's real details, and calls out specific skills. It's also tuned to the brand's tone, which it picks up through both the job ad and the company page. And then, of course, the recruiter reads every message before it's sent, so you've got that extra layer of safety. But the candidate-specific detail, what's neat there is it names the relevant skillsets that it thinks the candidate will want to see, like, "Well, why do you think I'm specifically good for this role?" So it says, "We're looking for XYZ, and we saw that you had these specific skillsets." And then some of those skillsets are inferred based off of their experience. So it's not just copy and pasting a word that they might have in their resume. It's a lot smarter than that. And then the tone piece where it pulls from the job ad. This is pretty cool because obviously, a senior care nonprofit and a tech startup, they shouldn't sound the same from an outreach standpoint. And obviously candidates can tell. So I love that we kind of thought through the candidate experience first and worked our way back to the technology of how do we build a tool that connects people in a way that matters most to each of them? Yeah, you need speed, scale, but also personalization, right? Because as Eyal was saying, he's able to get back to so many more candidates, and even those he would never have found on his own, right? And I think that's the kind of the value, obviously, the value proposition of Indeed and what you're supporting here. But what do those profiles look like, Eyal? How much do they align with the types of candidates you're looking to hire, and maybe were there any difference between those profiles of the ones you'd never have found on your own and the ones you typically find? Yeah. Of course, there's some overlap just based off the details it'll pull from the job description. Like we provide home care, so it'll look for those, but then it also dives a little bit deeper and realizes that we do outpatient home care, which is different than the traditional Part A. So it would actually go a step further and look for resumes with a certain type of outpatient ortho experience, which I might not have the time to find. But the real kicker that I realized is it was able to contact these travel contract therapists. Now these are like, I don't know, finding a treasure chest. Their contracts are like every few months they come and go. They have 10 licenses everywhere. So you really have to be vigilant and quick. Now, obviously, I'm so busy with everything else, I can't do that. So the AI was there scanning all the resumes, was able to pick out these people before it was too late, allowing me to have a conversation with them. Very cool. And is that a group that you wouldn't have thought of before, or because of this tool, it kind of surfaced. You're like, "Oh, this is a type of candidate maybe I should pursue in the future." Yeah, exactly. Awesome. It wouldn't follow the prototype of what we're looking for, but the skill set was still transferable and qualifying. And it actually allowed us to expand and offer something that we originally didn't consider. So now when you're looking, you can expand the talent pool like you wouldn't have before. Exactly. We now have something to offer to travel PTs, which opened up an unidentified pool of applicants in such a tight market. Yeah. And Kyle, candidates sourced through Sourcing Assistant are nearly three times more likely to get hired based on your research. What is driving that outcome? Is it the match quality? Is it the speed of contact, or something else? I think it's both, and they definitely compound. So the match quality puts the right person in front of you, and the speed gets to them before anybody else does. For the match quality, if you don't mind me going into these, I think the keyword matching cuts both ways. So a great candidate gets skipped because the resume doesn't have the right word that a traditional ATS might be looking for. And then another gets surfaced because it does, but it doesn't promise that it's still right. The auto body shop is an example that I keep coming back to because, or that I think about quite a bit because there's an auto body shop. We have a story of an auto body shop that had a very difficult time finding folks because they kept hiring mechanics based off of technical skills alone. They were looking for those exact keywords. But then when they started using this tool, it started looking for customer service experience as well. And it turns out those were the ones that worked out and stayed because they were the ones who were able to talk to customers, since the technical skills you can teach. And that's kind of where this deeper skills analysis comes out. It's the candidates that a traditional keyword search might bury. And then on the speed piece, obviously, as I mentioned before, being invited and being pursued is a completely different experience than just looking for a listing. And I think people are really engaged whenever they're reached out to. But as Eyal pointed out, the speed in which we can reach out to multiple people all at once is pretty great. And so one candidate doesn't have to wait for another candidate to be reached out to. They're not in a line. It's kind of all happening at once, which is really great. A much faster way, obviously, to find great talent, to get to them without any delay. And a couple stats, if you don't mind me listing off a few. Let's hear it. I just thought that it was really interesting is that the candidates that are matched, that receive an invitation, they're more than 15 times more likely to apply than obviously just finding the role in a search. So just being pursued jumps the likelihood to apply 15 times, which is crazy. And then there was a Harris Poll that was done that said 81% of hiring managers believe that retention is higher among those people who are sourced. They feel a little more loyal and a little more connected. Because I think this is a lot more intentional in finding talent, whereas just posting and praying is- It was a little more luck. And so this is just a really great way to find great talent quick. Yeah, and building off that, Eyal, has the quality of hire held up over time? So maybe you think you have the right selection and have hired the right person, but has it worked out over a long period of time? Has this person stayed in their roles? Are they successful in their job? What are you seeing from a longer-term standpoint? Yeah, good question. It's great to get people through the door, but retention is key, especially in healthcare. Out of the initial six applicants that I mentioned, we ended up hiring half of them, which is outstanding. And two out of the three are still with us, hired about three months ago. And when I trialed this AI system, it was about the end of March. The third person had to withdraw due to personal reasons, so it wasn't anything about it not being a fit. So overall, it's been a huge success, 50% conversion plus essentially 100% retention. That's amazing. And how does that compare pre using the tool? Yeah. Is it only one out of every five? Or give us a sense of success using the tool versus not using the tool. Yeah, definitely with the tool and the sourcing overall, the retention rate has to at least be 20% to 30% higher. Wow. Keeping it into the 78th to 80th percentile. With home care, it's a great setting, but also has its challenges. And for those individuals who've never experienced it before, it might be a little bit difficult. And we all know that healthcare has high turnover overall. So if we're going to spend the time and the resources finding the candidate, we want to make sure that they're a fit and they stay here long-term, not only for the company's sake, but also the patients'. We have to keep the patients in mind, so that always comes into play. Yeah, because retention really saves you a lot of money as well, so you don't have to keep on hiring and hiring and hiring and hiring and hiring and investing money in training, and then they leave, right? So it's a huge cost savings to have those strong retention numbers and have the quality talent come in and onboarded. Kyle, you mentioned that employers are saving seven hours a week per job and closing roles six days faster. How should recruiters think about what they do with that recovered time, right? And this is obviously the promise of technology and AI, too. It's like you're saving time because of the scale of getting back to people, the speed in which you're finding the right people and kind of latching onto them and inviting them in to maybe be interviewed. So what are these recruiters doing now because they've saved time? Going on vacation? Yeah. If their workload allows for it, for sure. Take that four-day work week. Something I always say is that a business is nothing more than a community of people who work together towards a common mission, often with the same values. And TA are the community builders of this community. So arguably the most important part of a business is the community builders. So in this case, I think what recruiters can do is they can put that time back into the parts of the job that are surrounding that community building. So the interviews, obviously real conversations with hiring managers rather than quick check-ins, like really deep dive intake meetings. Or look at the reqs that have just been open for too long, that have been collecting dust. Having an extra day per week is pretty crazy. And the hiring manager conversation, I think, is probably the highest value one because most bad matches and searches are because the hiring manager doesn't necessarily know what they want. They're going to highlight the basic requirements of the job and maybe fill in some pet peeves that they have about their current team and say, "I need someone who knows how to use this tool," or has a little more initiative or whatever it might be. But it's probably been many, many years, if not decades, that a hiring manager has been pressed to list out all of the technical skills that are needed as well as what are the relational skills, which that's what we call soft skills here, our relational skills. Do we want just a bunch of people who know how to do the technical thing but suck at talking to one another and providing feedback and collaborating and talking to customers? Probably not. We need people who can work well together, and we need those Ted Lassos, the folks who are really great at the interpersonal relationships as well as the technical stuff. So that would probably be my number one suggestion is to strengthen the relationship with hiring managers and then strengthen their understanding of what real talent looks like these days. That makes a lot of sense. For years it's like cultural fit, right? Do they get along with people? Do they fit in the culture? Are they... And that leads to a successful hire and people being retained, too. Everything is connected. That's 15-plus years doing this. Everything is interrelated, right? 100%. And obviously if you nail recruiting and people feel like it is a fit from the job seeker standpoint They're ready to go. Hopefully onboarding will be great, and they'll stay at the company for a long period of time. Right? Yeah. We have one question I'd like to hit on first. While AI-powered talent sourcing is a helpful tool for HR practitioners, how will the AI system adjust to constantly changing country labor and employment policies such as the USA, where employment is heavily impacted by immigration policies or other policies? Oh, that's a really great question. So every tool that Indeed has that involves AI has touched Indeed's responsible AI guidelines, which basically puts the human job seeker first. That's one of its primary goals. We have an internal model evaluation by both our legal team and product, and all of that is within adherence of evolving privacy and regulatory standards. So we're constantly looking at what is required from states and governments, and we're including that into how we are evaluating our tools and building them. So it is a ever-evolving process, as you know. I really like this question you already answered, Kyle, that's in the chat. Does the AI also provide labor market data, such as the size of the candidate pool for a specific position in a particular geographic area? Can it also provide insights into how competitive the market is or how difficult it may be to fill that role? Meaning with Stern at Home, can you already get a sense of how big the pool of talent is that you're competing for? Yeah, for sure. So there's this tool, it's my favorite talent intelligence tool, not because I work here, but because- ... because Indeed has more data about the labor market than every government on the planet, and every company on the planet. We have quite a bit of data on what's going on, and there's this tool called Hiring Insights, which gives you full access to all of that data month by month. And it's free, which is even better. There's so many tools out there that have less data, and they charge you a significant amount of money for access. So Hiring Insights directly tells you exactly what the labor market is up to, what the candidate pool looks like in any specific area that you choose. It tells you exactly how difficult it is. We have something called a competition score, so if the light is red, it's a competitive market, and if it's green, it's a favorable market. And then we also have Hiring Lab, which that's our economic research team. And all of this information feeds into this new tool that we built last year called Talent Scout, which works with all of our sourcing tools and stuff. And Talent Scout's just a chatbot version of all of these resources, so you can ask very specific questions. You can drop job ads in there and say, "What does my talent pool look like for this specific job ad?" It's just short of magic, I would say. It's pretty neat. Compared to where we were just a handful of years ago, it's much easier to identify how much budget you should be putting in, what type of leeway you should have when it comes to required skillsets, and so on. So highly recommend, and that's a great question because I think a lot of people are asking it. Definitely. So I think that that's a really, really important point. And going back to what you were saying, kind of building Eyal's perspective here, if we're saving your recruiters so much time, how are you able to better allocate that time to recruiting for maybe potentially other types of roles within the organization? Yeah. Definitely. The most important aspect of my role is to have these conversations, to get the candidates on the phone, and portray what Stern is all about. While we mostly hire PTs, OTs, assistants, there's also some business and admin roles that we have. And from my perspective, having a quality conversation and candidate experience is key. That can be something that sets you apart. But if you feel the constant itch to need to source and find people, you might rush that experience, and the candidate will notice that. So with the AI assistant coming in as a solution, I can rest easy, take my time, very much understand and listen to the candidate, and answer any questions they have. Sometimes the interview could be as quick as 15 minutes. Sometimes it's as long as an hour. And again, with the AI assistant, it's no worry for me. Now I have enough time to do both. That makes sense. And obviously there's so many applications with AI, but this recruiting application makes so much sense based on what you've experienced at your company and your ability to hire faster and hard-to-reach talent. So I think that makes a lot of sense. How do you see that recruiters are spending so much time building their workflows and developing their own instincts. What do you say to someone who's skeptical about handing part of their roles to AI? What should they maybe not hand to AI, and what should they hand to AI? What do you think makes the most sense in terms of their comfort level based on your experience and feedback that you've got from your team? Yeah, definitely. Good question. Humans are creatures of comfort. We like the same thing, repetitive tasks. We're comfortable with that So I was a little apprehensive to give AI the wheel, so to speak, in terms of messaging and if it'll be the right message to the right people. But the training aspect of it put all those worries to bed. The examples of real actual candidates were spot on, perfect, credentialed, everything. And then the tinkering and the tools that it gives you for the messaging, it makes sure that it reflects your company. So not only can you edit it completely, but you can actually change the tone or make sure that the message states certain aspects that you want it to about the company or the role, which assures that it seems authentic, that it's something that I or maybe one of my teammates would compose. So in my mind, at the end of the day, AI is all about efficiency and taking care of those admin tasks that don't really push things too forward. It's, again, the conversation. So you'll be pleasantly surprised at how accurate and successful the tool will be to your team. And the tool gets better over time, too. It kind of learns what your interests are and what's working for you. Back to you, Kyle. How does the system learn from recruiters' feedback over time? If someone passes on a candidate, how does that signal improve future results? Yeah. So anytime someone advances or skips a candidate or a resume, that is a signal on its own. The behavior is also a signal. So the system reads, like when someone skips a candidate, the system reads the pattern and what they didn't find, and then the next batch of candidates reflects that change. So anything that you do to define what is a good candidate or a bad match or whatnot, that is all feeding into the next batch of folks. And over time, the ranking and the matching sharpen quite a bit. So you tell it what worked, what didn't work, and sooner than later, it just stops bringing you the wrong resumes. It knows exactly what you're looking for for this specific role, which is really great, especially for evergreen roles that you might be filling out or trying to fill. Because you don't have to go through the same process of training if it's the same role and you're looking for the same thing, which is very nice. But this does calibrate per role. So whatever you set for, let's say, a junior-level position versus a mid-level position, those are going to be two separate matching algorithms that are working on your behalf. And in Yael's case, that makes a lot of sense because if he keeps trying to recruit these credentialed individuals, it's going to get easier and easier over time. Because Yael, you tell me if I'm wrong, the credential will probably be the same, right? Right. Exactly. The type of person you're looking for will probably be similar or the same. So have you seen that this, over time, has saved even more time, has been personalized and improved as you're using it? Yeah, for sure. Again, the accuracy of the candidate was there after you taught the AI. Like Kyle just said, you'll rate from your perspective of who is qualified and who isn't, so that is perfect. What the AI will learn is where are we getting the responses from and go towards that and edit it a little bit. And then you also, as the human, have control over the volume and the type of messaging, allowing you to tinker. So if you feel like you're getting feedback that is negative or not interested, you can then go into the AI, change the messaging, and see how that works out. And it'll show you the results. It'll show you the response rate and then divide that up between positive and negative. So analytically speaking, you get tons of feedback. So in a sense, your search and ability to connect with talent is improved over time, but also your skills as a recruiter improves as well because you know more of what to ask and what you're looking for. Yeah, exactly. I've been recruiting in this field for almost four years now, and the messaging I used in the beginning has changed due to just economic factors and career field factors. So this allowed it to expedite that learning process and not really have to shoot in the dark and hope something hits. If you were to go back in time, Yael, is there anything you would have changed in terms of how you got started using Sourcing Assistant? Yeah. On what we were just saying is to edit the messaging a little bit more. I had full faith in the AI system immediately. Once again, it showed me its learning model and that the candidates were qualified, and we got that abundance of applicants. However, you should change the messaging because one message for, say you, Dan, is not going to be the same message- Yeah ... for Kyle. And you don't want it to get stale. So I always recommend maybe weekly, just change it. Test a little A/B model here and there I think will only help things. So yeah. You still, like everything else with AI, you need that human editor, as they say, right? Exactly. And at the end of the day, again, I think if AI hires the wrong candidate, it's not AI, it's you, right? Right. You need to put it to work. So at the end of the day, the human needs to be involved in the process. So I think that that's a really important message for people listening in today. Kyle, where does sourcing assistance stop and what parts of hiring do you believe should always stay entirely in human hands? I love this question. I'd answer the first one, it stops after the first outreach. So that includes the discovery of the candidate and then the initial outreach, the initial message. Everything past that is the employer. And then the accountability is always on the recruiter. So what's automated there is just the most time-intensive stuff, the thing that doesn't really require a human at all. What should always stay human, though, is the judgment, for sure. Reading a person and hearing what they have to say, deciding whether or not they'll thrive on the team or with the manager and the environment and the work. That is a human decision, and hiring overall is just one of the most consequential decisions that we make about another person's life. So that deserves a human on the other side of it. Yeah. And as a human, your ability to discern and think about cultural fit and personality, you can't even outsource that, right? And especially if this is someone on your team, right? Or someone you're going to be working closely with, there's repercussions for hiring the wrong person. And like I was saying before, yes, recruiting is really important, but good recruiting leads to good retention, right? Which, if you look down the road, leads to good leadership development. Everything's connected, right? I think by having a good sourcing assistant, you have assistant that's actually sourcing the next generation of your leaders of your company, potentially. So there's this long-term implications and pipeline development that is created just upfront. Last question for you, Yao, and I want to end on this. What does a win look like for you and your team because of this tool? What does success feel like day to day? Yeah, great question. At the end of the day, again, any of my fellow recruiters or TA professionals who are considering this, it's all about ROI at the end of the day. And for us specifically, it's an increase in top-of-funnel applicants in such a tight and strict competitive job market. But not only that, we want to be having conversations with people who are a fit, not only with the requirements, but also culturally, like you were mentioning earlier. That will then lead to retention. So not only top-of-funnel and more hires, but also years down the road, they're still with us and happy. And Kyle, I'm going to throw this added question at you. What is your hope for how this tool develops over time? And do you have some sort of vision for where this could all lead to in, can't say five or 10 years, the world could be completely different. Like really different. So let's just say in the next one to two years. Yeah. Actually, if you don't mind, I will say five to 10 years because we've- Okay ... had the same vision. And then I'll work my way back. But Indeed's had the same vision since we started in 2004, and that is to get to a place where job seekers and employers just have a single button that they can press that either says, "Get a job" or, "Get a hire." So that is the ultimate goal, and that, of course, takes a lot of work to understand job seekers deeply, what their preferences are, what types of environments that they say that they would thrive in, all the things that make them up as a professional. And then, of course, on the employer side, what type of talent benefits from their culture and thrives. Obviously, what are they looking for? And so that has always been the goal for Indeed, is to get to that singular button. Working our way back, I think the next evolutionary stage, and I don't really have a lot of insight into what's in our product roadmap. But I think the next evolutionary stage for sourcing is to be a lot more proactive for recruiters and say, "Hey, there's some of this talent that you might not be looking at, but we're seeing a trend in this type of skill set for this type of role in other companies. So you're not looking at it right now, but other companies are. Here are some things that you could start expanding your mind into what talent truly looks like." And this goes into what a skills-first future is. And it's my deep belief that talent is universal, it's just opportunity that's not. So I think having a more proactive agent out there identifying trends and then reporting back, and asking, "Is this something that your organization is interested in or not?" would be a really interesting step forward. Yeah. It also requires recruiters to be more open. I remember years ago, there was a lot of articles written about companies saying, "We're open to hiring non-traditional," quote-unquote, "non-traditional candidates, those without a four-year college degree." Mm-hmm. But the question is, are they? And that's just one example, but just in general with this new, as you're describing, a more skills-based economy where people are more open and like, "Oh, this skill can be applicable to this job." And really if you think about Yale's position, there's only so many people who would be qualified because they have to be certified in certain locations, which kind of shrinks that pool. But are there ways to give him a larger pool and make it easier to be credentialed? There's a lot that could evolve in the future. Exactly. So if anyone has any other questions, feel free. I always love these discussions. I've just been in the field long enough where new tools, new technology, especially anything with AI, there's always a lot of interest because everyone's looking to become more efficient and better and smarter and just be better at their job, right? I think that that's really why we're here because if we're better at our job, our companies do well, we do well, and so forth. So thanks so much for pulling back the curtain on AI-powered sourcing. This conversation will continue to be very relevant, of course. As technology evolves, as AI becomes smarter and better and allows you to scale and personalize like has never happened before, right? It's making recruiters better if you think about it because old days, you just send a bunch of messages, and now it's more I'm getting back to people sooner and filling that pipeline. And especially with these hard-to-fill positions like in Kyle's company or Yale's company, that is really critical. So thanks so much for being part of this discussion. The future of recruiting isn't just about replacing human judgment, it's actually about automating everything else and promoting human judgment and relationship building, so giving recruiters the time and the tools they need. So thank you so much to Kyle and Yale, and we look forward to continuing this discussion. Yeah. Thanks so much. And thanks, Shannon, for your comments. Yeah. Thanks for having me. Yeah. Bye, y'all.