Agentic AI in Hiring Before It Was a Buzzword

Welcome back to Hiring Excellence. I'm Johnny Campbell. As we wrap up production on new episodes, we're back in the archives today. But this isn't just any replay. This is probably our strongest, we call it early episode of all time. If you've been anywhere near the TA space in 2026, you know that Agentic AI is the absolute dominant talking point. Everyone is buzzing about it. But nine months before it became the biggest buzzword in the industry, I sat down with Pavan to explore exactly what this technology was doing and going to do to hiring. Listening back to it now is kind of wild. What sounded like a bold forecast at the time now reads like a literal road map for what recruitment teams are dealing with today. This is the Agentic AI episode we did before Agentic AI was a buzzword. Whether you're trying to wrap your head around where this tech is going next or how to actually harness it for your team today, Pav's insights are pure gold. We'll be back in the feed with fresh new episodes very shortly. But for now, grab a notepad and enjoy this incredibly preient conversation with Pav, you're very welcome to the show. It's great to have you here. I wonder could you perhaps tell us where you're joining us from? Tell us a little about your own career background and maybe the team at 8fold. — Yeah, lovely. Uh Johnny, thanks so much for having me and uh it's a privilege to be talking to you and your uh business here. So I'm uh joining all of you uh from Bangalore, India. Um I head the global talent acquisition function at 8fold. Uh a bit about 8fold if I can share before I jump into my career background. 8Fold is a 10-year-old uh startup in AI. Uh we're headquartered in Santa Clara Bay area with offices in London, Bangalore, NOA and elsewhere. Um know we're backed by for you know venture firms like Soft Bank, Lightseed, City and other uh such you know storied backers. We sell to large enterprises. Um if I can count uh some big customers like know Coca-Cola, there is Starbucks, PayPal, Ericson, Department of Labor, Department of Defense. As you can see, these are really very different kind of customers that we sell to. There are mining companies, the retail companies, big pharma companies that we sell to. So the uh product is really um tracks the talent life cycle from acquisition to management you know development of talent. So it's not like a classical CR and ATS uh that's sort of the background I going to give um also pioneers in the area of employing AI for you know matching skill development skill taxonomy and now agentic AI and I'm sure that's where we will spend most of our time uh today. Um a bit about me u I've been with 84 for last four years. Um I joined uh to scale the India uh operations. Uh but after a year I got the opportunity to lead the global TA team here at 8fold. Uh prior to 8fold I was with company called Akami Technologies. I was leading the TA there for APJ. uh before that I spent a decade at Juniper Networks other Silicon Valley you know story company that's where I you know uh cut my teeth uh so to say and you know this has been my journey so far nothing else if you want me to add I'm happy to double take on that give me some content context on so as the VP of TA in in a software company in our sector it's kind of a unique one right because a lot of folks listening might be customers or they might be looking at fold as a product and you run the recruiting team there. So you know tell me about that recruiting team its size where are they based what kind of roles are you hiring for? — Yeah so we are um as I mentioned we are in you know Bay Area, Bangalore and London. So I have uh recruiters in all these locations. Uh we are at 20 25 member team right now. Um I have recruiters in Bangalore, recruiters in Bay Area, there a few remote recruiters in the US on the east coast side. Uh I have recruiters in London as well. Uh we also have a global sourcing uh practice in Bangalore. Uh so this nimble team supports all the recruiters in terms of you know early sourcing, lead generation, community building, all of that good stuff that eventually translates to a great talent. Uh plus we have a college recruiting leader here in Bangalore again running the global charter for us. Um we recruit from top colleges in India among the you know what we call day zero recruiters of the early ones to get stake. Um so roughly this is how the

team has been you know designed or shaped. So Pin, you know, you're in this unique situation where you're not here today to sell as eight-fold or talk necessarily about eightfold as a product. Um, you're here representing your peers in TA, you know, you run recruiting, but uniquely in a company like 8fold that builds technology for recruiters, um, you must have a kind of benefit of maybe being customer zero or getting the inside track on new products. So how does that work? that relationship between your team, you and the product teams. — Well, it's very fascinating and I always consider it as a privilege of know for me and for my team to be able to partner so very closely with our know engineering team, the customer success team, the sales team. So my team touches all these you know the entire product life cycle right from design of the new features new products to the deployment to the selling to the post you know deployment care that we give to the customers. Now I can double click on this one part of the right like um know we have several product managers within the company know uh they are owning several parts of the product like there is a product manager for you know AI there are sourcing product managers for career side the way we have shaped of the structured that the arrangement is one of the recruiters owns one of those streams like you know almost like they're associate product managers product manager and you know they work very closely with this these teams of the of their team to you know one provide early design feedback second you know provide validation of the design itself the user research um the nuances making it even more sharper and you know more perfect u for as an example I can give you just last month we launched the AI recruiter and um you know we've been part of daily standups design reviews um metrics evaluation know fast loop closing of all the feedback. So there's a lot of know hand you know um handle glove kind of you know arrangement that we have with the product and engineering team in building the product in shaping the product. So that's one part Johnny. The second part is post uh you know we sell the product. We are part of several design workshop with the customers on how do you define your process how the how to best get the ROI on the product because we've been using this product so very well for so many years now. uh we have very strong point of view what works in the marketplace what does not work and also been part of several customers you know uh interactions where you know we spent a day at recently in Kansas uh my recruiter and I to you know help one of our customers you know adapt our platform better we exchange more stories we exchange what are the gaps what are the opportunities and how best to leverage that so the the extent of partnership is really very deep very embedded I would say. — I love that you mentioned one of your customers being Ericson. Kevin Blair has been on the show several times and he was sharing with me as an EO customer that he had seen a few months ago preview of some of the stuff that's coming in the second half of the year. He's blown away particularly by some of the agentic stuff and that's as you said earlier what I'd love to dig into. Maybe we can start if you don't mind Bob, by just defining what a Gentic AI is and isn't in the context of recruiting from how you and maybe the business sees it. — Yeah, I think the uh the misconception can be that you know is automation agentic AI, what is agentic AI? What is automation? Agentic AI is beyond automation. You know, with agentic AI is very thoughtfully designed workflow where the AI takes over the much of the operational workflow uh with a human in the loop. It at very thoughtfully carefully designed you know checkpoints. So the for example you know you can let an agent talk to a candidate do initial recruiter screening summarize the feedback get all the notes you know engage very thoughtfully with the candidate the way a human would interact come back to the recruiter and give a very good summary uh that that's one example of you know agent AI the another example could be like you know you uh on the hiring manager side the manager wants to open a position the agent can very easily you know take an instruction saying that They open a position in you know Nevada for uh you know customer success manager and an agent the agent is able to collect all the know past information like calibration the skills required job description the salary range you know who are the ideal candidates for this role and open the role publish the role and give it to the hiring manager for review saying that this is what I've done already this is the final outcome now these are all not

really you know a fictional or no skyfi kind of scenarios but it's a scenarios that are coming that are already you know uh at play. Um, and in terms of I guess a lot of folks I remember when I first heard the word agentic AI, I was going, "Oh my god, here's another word I got to get used to. " Um, you know, we were just getting around gener generative AI, you know, thinking I don't know what agentic is and told someone said agent. Agentic, that's, you know, it's agent. It's doing things for you. So rather than going off and producing lots of text, which I guess generative AI produces text and produces images and produces video even, so produces media. Um, agentic is an actionoriented thing. So I guess the difference between a talker and a doer, right? Generative AI is the talker. Agentic AI is the doer, right? Go actually perform these actions. And — you know, I've been excited by some of the new Agentic AI developments this year. You know, the likes of Alexa plus being launched by Amazon that can take actions for you from again a conversation. So oftentimes it's it there's a mix. Generative might be the understanding piece to interpret and then agentic will be take the actions in recruiting. You know, we've seen in the last year several big announcements by different companies of so-called agentic capabilities. You've seen LinkedIn launch their hiring assistant to talent connect last year. still isn't released yet, but you know, in early release promised to be able to write job specs, post ads, source candidates, reply to those candidates, short list for you, answer queries. um talk to me about where your experience overlaps perhaps with product like that and then goes further and maybe you know why that's different to perhaps some of the other stories we've heard where you know for example you know McDonald's or UPS or 7-Eleven are using Agentic AI to hire hourly workers you know in a textbased workflow you're doing this in a professional environment hiring software engineers so for example so I bring it back I'm not I'm asking too many questions at once but if you could walk us through maybe the agentic AI kind of um use case what are the different points in which you're using AI Gentic AI have begun recently started using Agentic AI — yeah so the u the biggest use case so far and more to come uh we're working on new innovations but the right now what we have just released is AI recruiter um you know a company gets ton of rums applications you get hundreds of applications now recruiters are very busy creatures. Um always overworked constantly juggling multiple priorities. What happens typically is you know you would look at top few resumes and then you shortlist a few and then talk to those and then that becomes the pipeline and any good candidate that applies later on in the process or someone who was know the bottom of the pile could get overlooked. I'm not saying we'll get over but could get overlooked. Right? What agentic what agent or AI recruiter is doing is you are unleashing uh you know a tireless 24 bar 7 available recruiter to look at all those know applications you know the candidates can talk to that AI at their convenience could be weekends could be you know later in the night during lunch hour at their convenience like we are seeing we're seeing a lot of people talking to the AI during the weekends when they're at home you know with a you know good you um at a mind with a good mind space all of that and then they talk to the recruit the AI share their experience and the AI is really very thoughtful this is not uh you know uh a list of questions to be asked and check those boxes and come back we respond react to the questions to the answers ask smart follow-up questions bring all that rich data summarize that and bring it back to the recruit for an evaluation the recruiter then does a thumbs up thumbs down based on you So what has what has been summarized there? This is useful because people are writing really smart rums. Chad GPT has help people write really tailored made resumes. So recruiters might end up talking to a lot of false positives if you don't use something like this. Uh second the experience for candidates is really superlative. They can talk to a recruiter AI recruiter at will at ease. All right. And you know we have seen people u asking for a second chance to the AI saying sorry I'm not fully prepared today. Can I talk to you tomorrow? And then they can go back to the same invitation, click on it and talk to the air recruiter whenever they want. So these are really very guided, experience-enhancing, recruiter, experience-enhancing, you know, features um that we are building. The question could be what would recruiter do? The recruiter is actually then able to then distill that pipeline, talk to the really good or the really matching three four candidates, invest all their time in prepping that candidate, understanding that candidate even better, you know, ensuring they get

a white glove treatment and they're successful. Say at scale this is going to you know uh do wonders for all the experience and we are ensuring there's a human in the loop to make decisions. So AI is not saying this person this candidate is good or bad. We are not there yet. Um all we are doing is you know uh getting really insightful comments about the candidates's experience about their background um they're very structured consistently applied in a rubric format back to the recruiter. So taking let's say an hourong interview and distilling it into the key information you can summarize and and absorb in three to four minutes as a human and compare as you said a standard rubric of information for every candidate to regardless of how they structure the resume because the interview the agent is conducting a consistent interview. you're capturing the same data or at least asking the same questions consistently of everybody and using those data sets to review and to your point the human still makes the decision is that a good or bad answer what's the waiting who do I progress etc so would you talk about — if I can add one more point to that Johnny about the questions that are being asked right uh the AI is tailoring the questions based on one the candidates's resume their background and the job description so it's not a you know one template question being asked to every candidate. It's actually very thoughtful. He talk he's talking to me. He'll say, "What did you do at Akami? What you at Juniper? Why did you choose this path? — Why not that? " — So, it's very it's extremely smart, very intuitive, you know, um recruiter who's tirelessly working for you. — I love that. Sorry. I think you're it's a really valid point because it's trying to get the same evidence from everybody, but it may take a different path. It may ask different questions, different probing depending on each unique candidates's background. I guess what it's doing is presenting hopefully the same, you know, evidence in different categories. So the same categories are trying to be covered. I imagine if a candidate shares three data points in one answer, the agent's not going to answer ask the two questions it was waiting to ask because they've been answered, it moves on again. It's intelligent that way. and and in your own experience and I'll come back to the full workflow just a second pin but so far as of the recording of this about how many interviews have you conducted across how many roles uh just give me a sense of is this something you've done five times 50 times 500 times what kind of numbers are we looking at just yet in your own team — yeah we launched roughly one and a half month ago uh Johnny u we so far interviewed some 600 plus candidates lyrics. — Wow. — Over last you know over what some 35 40 different roles uh spanning US, India, EMIA across the globe wherever we have open positions and the feedback uh both from recruiters and the candidates has been really phenomenal. — Wow. H how are you capturing that feedback? Is that uh quantitative in terms of they're giving a score in a survey or qualitative they're sharing notes and emails or a bit of both? a bit of both. Uh one is at the end of the call there is also you know a survey which is fairly standard there they can give how do they rate the whole conversation. Second we also talk to these candidates as a post followup conversation you know whoever we want to take them forward to the next round of interviews. We actually go back and pop to them and prep them for the further conversations. any questions that were unanswered, anything that they want to know more that wouldn't have been answered by the AI and during that time too we asked uh know all the candidates about their experience and they are sharing some really very good you know feedback and people have written LinkedIn post about their experience know some people have know told the AI though they know they're talking to a machine not a human but they concluded the call saying this is the best experience they had with the machine they thought it would be you know very robotic in that sense you know but it has been fairly humanlike — I know most recruiters when you talk to them about this concept if they haven't already done this so most of them haven't you know recruiters conceptually and recruiting leaders conceptually will say I can see how AI can schedule interviews I can see maybe how AI can source maybe even how AI can write emails and respond to candidates but there's no way you can replace a human doing an interview when you take that candidate journey can If you walk me through what you've learned about all those different steps where you have personally seen AI to be strong and where would you say AI is not there yet and perhaps you although you don't have to have a human in the loop in a part of a process you prefer to keep a you in the loop what's been your experience there I think the um curiously enough right if you have more than one human in the loop it is difficult for AI to you know do a efficient job for example scheduling of

an interview you If you look at your own calendar, it's so very difficult. It's all choco block. It's all blocked, right? There's no place and a human coordinator can, you know, call you and say, "Hey, can you move your this meeting? lunch or can you move this one in inside an interview with with AI or with automation what happens is you can only look at the vacate spots. It can be one week from now. It may not be convenient for the you know candidates. So that's a very hard problem to solve if you look at it that way. So the scheduling is a harder problem. That's where you know uh we are also trying to introduce a functional interviewer like you know not just the recruiter but also the AI can talk on behalf of the hiring manager — right and you know that's again very similar benefits at will you can talk you can scale that um to hundreds of candidates uh so it's it's another great experience for everyone there while again the human in the loop still decides whether this is no was the right answers. Do we have the right background and everything? So the the difficult part is the scheduling part. Funnily enough, um the interviewing part is fairly straightforward. The then the offer negotiation, the offer design part, those can't be done by AI. uh if the jobs are you know difficult to hire if you're in a know tough talent market but if say the jobs are like vanilla jobs you hiring hundreds of people in a month for same job uh and you have a very set compensation set package philosophy uh you know the AI can of course roll out the offers too so right from creating of jobs to you know screening of rums to you know offer design offer negotiation that's where AI can do the job if the job is fairly simple commoditized job that you're trying to fill what's difficult is um still the sourcing is quite difficult uh especially if you're chasing top talent if you're in an industry where the talent matters more than you know um anything else then know I don't know yet if AI can effectively replace you know um a human sourcer talking to somebody you know wooing them coaxing them you know through multiple strategies you know that's something which is going to be hard uh that's where human ingenuity creativity grit all of that comes in and that's where I see you know u recruiters can focus more of their energies towards you know sourcing building community doing branding know storytelling pitching your product pitching your company and that would still continue uh to be in vogue or to be in — it's interesting. I was looking at a post that Glen Cathy, good friend of ours, posted recently on LinkedIn about a University of Zurich study where they had um unbeknownst to the users uh put AI uh profiles into Reddit into several forms and engineered situations of negotiation and persuasion. And they found that uh the AI performed in the 99th percentile for negotiation and persuasion in these real world examples over three four months with unbeknownsted real humans. And so maybe the time for negotiation is actually moving towards AI. And we were thinking about this because what you've described is probably the reverse of what a lot of recruiters think which is oh you know posting sourcing engaging can be automated. Interviewing no way that's a human thing. And you're saying actually scheduling is difficult. Sourcing if you're looking for high-end talent might be a human uh thing. Interviewing is a standard you know structured thing that actually lends itself much more to this. And it brings to mind a comment that uh again in a you know in a podcast that we were doing with Glen Kathy recently he shared another um you know idea uh to think about. He was saying you know a lot of people give AI the admin work and he said but imagine if you think of AI as an assistant that does all that can do any amount of work for you but this assistant has an IQ of 140 150. — Would you give it the admin work? Isn't that a waste of its resources? And like with that in mind, you know, where do you go next? Like what would you be excited to see? And I'm not necessarily talking about eightfold's product strategy here at Puffin, but you as a TA leader, where would you love to see Agentic AI go next? And what problems would you love to see it solve next? — Yeah, I think the again I will speak from the know from a tech recruiting, you know, lens because that's what I've done my entire career. So I can only speak from that standpoint. Um authentically I would say uh AI can really help more and more in candidate rediscovery. You know you don't really have to source. So you know um so man will if I can say you know you do all the heavy lifting AI

should be able to tell you that if you're looking for this kind of a you know talent you already have in your talent universe somewhere hidden could be a former you know employee could be offer declined candidate a silver medalist you know how there is always talent hidden in the mind of extremist you already have like you know can AI proactively bring that up to you which you know Again a lot of AI centered ATSs can still do it but being even more productive saying that you know these are the five candidates know you should send an outreach and I can help craft that bridge right this is again based on how bad the search has so far been we are seeing so many candidates getting rejected now I can come to your rescue without even being asked to right that that's one second is of course being able to like periodically engage with that talent uh know can force the what kind of positions you're going to hire. You have had this skill last year or this time consistently. So I think you will hire again this one based on how you are know your positions that be published. I can recommend you to do this event. send these you know initial feeder emails to these n number of candidates in your community. I think that proactive you know planning proactive plotting of the next moves um while the recruiter is busy handling immediate pers that could be a great you know ally for a group to do that. I wonder because could you see a world where you know these recruiting agentic AI recruiters get connected to the HIS and they're able to review quarterly review data and employee feedback surveys and again begin to predict based on employee sentiment what departments might have needs coming up in the next six to nine months might get a bit scary for folks. The opportunity is there. I think the technology can do it. It's just a case of where you know will humans want to have it do that? uh essentially before we kind of conclude it' be remiss of me not to mention guard rails responsible AI practices so you know maybe you can give me a brief overview just in terms of how the HFO product brings responsible AI into place but maybe how do you also then bring that into your processes with your recruiters and your team so maybe if you can share with me those two angles on responsible use of AI as you see it — it's a very important topic Johnny and for you know bringing it up in the product design itself we u have been very deliberate in ensuring there's a human in the loop to make a decision so it's not the machine is not making any decisions there it's all inferences it's all recommendations but never a decision that's one the second is the AI that we have is completely explainable we can actually show you why we are recommending a certain thing like what are the factors going behind it and you know you can always audit that The third thing is we have a lot of external audits, internal audits, a lot of transparency. Uh we have an AI ethics council. The council is helped by uh people from OFCCCCP. There's no EOC a lot of regulators advisor on how we are using AI. Uh we are advising governments like New York state on how to use AI for employment. So there is lot of um you know leadership that eightfold is showing in terms of being uh at the forefront of responsible AR. So that's one part of the whole product the way we are designing the product itself. Second part in terms of processes also we're ensuring that people get uh you know opportunities to opt out of AI is somebody says I don't want AI to evaluate anything you know be it my resume be you know interviewing any touch point it's I want a AI free uh journey we we ensure that they get an opportunity uh to be able to do that the um in terms of feedback surveys all the you know the metrics that we have to measure share the experience of candidate. It's really very you know transparent. We keep a very close tab on what people are saying and how we can you know provide any opportunity to redress any of the deviances that people may have. — A very thoughtful approach both from the company and your team and Paul I really appreciate you sharing that. Final reflections. If we have leaders listening to this call, some skeptical, some excited, but they want to try and experiment at least and begin their journey with piloting Agentic AI, what advice would you give to folks regardless of what system they're using or technologies they might have in front of them? Um, do you have any advice you could share with other leaders around your journey, what you'd recommend others do, how to approach beginning to roll out or experiment with Gent AI? Yeah, I think one I think it should be all centered around building trust. Uh trust in terms of the technology that is being used. Can you trust the recommendations? Validate that

the systems are really you know well built well designed. Um second is build trust in the end user. So don't start with a big splash. You know identify a small group of users who are actually die hard optimist. Sometimes this technology can fail and you know the the initial disappointment shouldn't cloud uh you know the efficiency or the adoption for the entire enterprise. So choose your first team very carefully pilot with a small group of the hired optimists both on the business side as well as the recruiting side and you know measure all the outcomes sometimes you may not get all the data points that you would want. So think of how you can build your own you know data collection points, dashboards, talk to a lot of you know early adopters, users like candidates or managers get their feedback and thereafter you can scale rapidly. I would say centered everything on trust is the way forward. — I love that. If leaders have any questions about more detailed questions about how to get started, what to do, are you okay with folks reaching out to you on LinkedIn? That work for you? Yeah, happy to. Yeah, happy to engage — and we'll share your LinkedIn profile details in the show notes and hopefully you get a lot of interest and I really appreciate you sharing that. I think you're at the forefront of innovation here in terms of adoption as a TA leader of Aentic AI. I only see it going one way for all other heads of TA in time. Um, great to hear how you've, you know, taken that giant leap forward. You're in the privileged position of having a technology team behind you with a great platform that can do that, which is a great position to be in. and maybe we'll hopefully get you back on the show in the next year or two to see what have you done next, what has been the next journey and the next step for you and your team. Thanks for joining us today. — Thank you so much. — And thank you for joining us and hearing Povven's story and how eight-fold are using their own product internally and the concept of Agentic AI and maybe some of the surprises that you might have heard today on how it's working and in what areas it's working and how successful it's been so far. Um, we'll be back next week with another fantastic hiring excellence podcast. Another great leader sharing their stories. If you don't follow us already, please do so on Spotify, Apple, or wherever you get your podcasts. But until then, we'll see you next week.