For years, a candidate experience that delivers truly personalized feedback at scale has been an unattainable dream for talent acquisition teams. Candidates have always wanted tailored feedback, advice, and the opportunity to ask questions, but the sheer volume of applications has made this impossible
So, is AI now revolutionizing the candidate experience and finally making mass personalization a reality? What does it take to create an ethical, transparent system that provides candidates with two-way interactions and personalized feedback, all while improving efficiency and reducing workload for TA teams?
My guest this week is Sam Dhesi, CEO of Popp AI. Despite only being 18 months old, Popp AI is already partnering with FTSE 100 companies and global staffing organizations. In our conversation, Sam shares real-world results on how AI enables conversational assessment, provides personalized candidate feedback, drives engagement, and vastly improves hiring efficiency.
In the interview, we discuss:
• How AI is changing the way employers interact with assess candidates
• Building in an ethically and responsible way
• The importance of human oversight and explainability
• Delivering constructive personalized feedback at scale
• Facilitating mass candidate engagement
• Using assessment via conversational AI as a replacement for the resume
• Which employers are doing all of this well, and what tangible results have they achieved
• What does the future look like? Where will hiring be in five years time?
Follow this podcast on Apple Podcasts.
Follow this podcast on Spotify.
Matt Alder [00:00:00]:
Support for this podcast is provided by Popp AI. Popp AI brings next generation conversational AI to RPOs in house talent teams and recruitment marketplaces, making every candidate interaction faster, more human and more efficient. From automated screening and voice note transcription, with support for 40 plus languages, the AI powered platform engages top talent through natural multichannel conversations. Popp AI transforms early stage hiring, capturing insights, collecting documents and seamlessly scheduling interviews so you can focus on building stronger connections with the best candidates. Visit joinpopp.com, which is joinpopp.com to discover how Popp AI is redefining recruitment automation for the modern workforce.
Matt Alder [00:01:14]:
Welcome to episode 663 of Recruiting Future with me, Matt Alder. For years now, a candidate experience that delivers truly personalized feedback at scale has been an unattainable dream for talent acquisition teams. Candidates have always wanted tailored feedback, advice and the opportunity to ask questions, but the sheer volume of applications has made this impossible. So is AI now revolutionizing the candidate experience and finally making mass personalization a reality? What does it take to create an ethical transparent system that provides candidates with two way interactions and personalized feedback, all while improving efficiency and reducing workload for TA teams? My guest this week is Sam Dhesi, CEO of Popp AI. Despite only being 18 months old, Popp is already partnering with FTSE 100 companies and global staffing organizations. In our conversation, Sam shares real world results on how AI is enabling conversational assessment, providing personalized candidate feedback, driving engagement, and vastly improving the efficiency of hiring.
Matt Alder [00:02:33]:
Hi Sam and welcome to the podcast.
Sam Dhesi [00:02:36]:
Hey. Hi Matt. It’s great to be here.
Matt Alder [00:02:38]:
An absolute pleasure to have you on the show. Please, could you introduce yourself, tell everyone what you do.
Sam Dhesi [00:02:44]:
Sure. My name is Sam Dessi, one of the founders and the CEO of Popp AI and we build and sell digital resources to both enterprise talent teams and global staffing companies. Yeah, that’s the headline.
Matt Alder [00:03:04]:
Fantastic. Well, give us a little bit more. Tell us a little bit more about how long you’ve been going and what is it you do.
Sam Dhesi [00:03:10]:
Absolutely. So we’re a pretty young company, so we’re about 18 months old, just under 18 months. And in that time we’ve managed to achieve quite a lot. So we’ve actually raised 3.6 million. We’re a venture backed company and more importantly, we’ve onboarded numerous FTSE 100 companies as customers. We’ve obviously built A great product with an emphasis on how we can apply AI responsibly and ethically in the process. And I’m sure we’ll get onto more of that. And we’re now, we’ve grown from a co founding team of three to now 12 people in Central London. So yeah, it’s been a really interesting year.
Matt Alder [00:03:55]:
Fantastic stuff. And yeah, that’s an incredible way to come in in just 18 months. So let’s, let’s dig in a little bit further. So talk us through what you’re seeing as the kind of the main current use cases for AI talent acquisition and how quickly is the adoption currently going?
Sam Dhesi [00:04:12]:
Yeah, sure. So I think broadly speaking there are two buckets that the application of AI in hiring kind of falls. So I think there’s the low hanging fruits and this is where most of the kind of applications and products have cropped up over the last year or so. So this is for example generating job descriptions, helping to transcribe interviews, helping to transcribe intake calls and summarizing that information. Essentially it’s the summarization and the generation of text doesn’t represent a high risk to the process. Now that is a great application of AI, but it is to me that is low hanging fruit and I think there’s a very low barrier of entry when it comes to those kinds of products. On the flip side, you have the second category which is slightly higher risk and this is where we’re actually changing the way that we interact or assess candidates. So for example, when a candidate applies for a role, how are they currently assessed? I mean for most talent teams it’ll be an individual on the talent team looking through applications manually, going through them, applying their human intelligence and trying to figure out whether somebody is suitable and then ultimately reaching out to that person and initiating a conversation which might include further pre screening, it might be an interview, it might be to schedule a time with the hiring manager. Now that part of the process classically has been run by human operators. It’s still very repetitive, very manual, very kind of high volume in many companies. And that seems to be a difficult place for AI to penetrate because it’s not just about the underlying AI, it’s about how we fit into the workflow, it’s about how we, how we allow the human to still have control and oversight of the process, but just at a larger scale. So it’s much more difficult to build a product around there. But that actually is a great opportunity and part of the reason why we exist.
Matt Alder [00:06:35]:
Absolutely, that makes perfect sense. And I Suppose, picking up on what you said a little bit earlier about building something ethically and responsibly, what’s your kind of view on that? How does that, you know, fit with the kind of the opportunity for AI that you just described?
Sam Dhesi [00:06:50]:
Well, it’s everything to us and we know it’s very important to our customers too. It’s really hard to get this right. So we can take a regulatory point of view on this and we can also take a moral point of view on this. But you know, there is quite a lot of overlap. So I’ll just give you some examples of where we need to think about the way that we’re applying AI. So for example, should the candidate know that AI is involved in the process? So we take a very hard stance on this and we say yes, absolutely. So whether they are being, whether their application is being assessed partly by an AI system or whether they, they’re being reached out to by an AI system, and we do both of these things, we actually make it mandatory that the very first communication that our conversational AI might have with a candidate mentions that it’s an AI assistant that they’re interacting with. This actually greatly improves the candidate experience, which might seem paradoxical, but we’ve actually got the numbers to prove it. People know what they’re getting into, they know what to expect, they appreciate the transparency. And actually there’s a kind of feeling that you’re not really being held accountable to another human on the other end. So maybe you’re a little bit more open in the way that you respond. So that’s just one example. Another example is ensuring that there is always human oversight in any process that is being handled by an AI system. Now, one of the most dangerous things I think anyone could do is is allow an AI to autonomously be handling very important decisions or processes between a business and candidates. And when you’re faced with a product that is applying AI in recruitment, you’ve got to really look at this aspect, like what controls do you really have? And of course, one of the ways in which we built Popp is to enable you to always see what is going on. You can jump in, you can pause any AI enabled process. If there’s a conversation happening between a candidate and our conversational AI, you can actually pause it, have a one to one conversation and restart it. We even tell you when we think you need to. So just a couple of examples there. One more actually that’s extremely important is a kind of transparency, not in terms of just telling a candidate that AI is involved in the process, but actually in the way that we explain how we’ve come to some kind of conclusion about a candidate. So, for example, if we are using AI to screen a candidate’s resume, you know, what does the. How did we come to the conclusion that they scored, that they met a mandatory requirement? Well, we need to make it as explicit as possible what information that we looked at in that candidate’s application, whether it’s in their resume, whether it’s a response to a question, what information we used to come to that conclusion and make it very clear to the person who’s handling the process. So three clear examples really of how we can improve the kind of ethics and responsibility of applying AI here.
Matt Alder [00:10:10]:
So you mentioned the candidate experience there and how that can kind of be improved and augmented. We’ve been talking about candidate experience for, for many, many, many, many, many years now. Why is it so important right at the moment to kind of really focus on the candidate experience?
Sam Dhesi [00:10:28]:
Well, I think candidates are demanding much more. Just like, you know, any, any kind of consumer group, candidates are also demanding a lot more from any hiring process. And that’s partly because it’s becoming more and more competitive when it comes to kind of some of the kind of very popular roles out there. So if let’s, let’s take the average candidate experience today, it might look something like this. You apply to a role you are in amongst 300 applicants, you have more questions about the role beyond what’s kind of on offer in the job description. And you, you apply and you reach out to the hiring manager, obviously you hear nothing back. And you might be wondering what’s happened to your application. And you hear nothing back. Most people at this stage will just be ghosted, so they’ll never hear back. And it’s a terrible candidate experience. And whenever anybody talks to you about the brand of a company that you applied for, you’ve got these kind of negative feelings, whether it’s as a candidate, but maybe even as a consumer. So there’s a kind of snowball effect there. If they eventually do get back to you, it’s typically with generic feedback. And that’s because of the limitations that we have when it comes to providing feedback. Right. It’s hard to give everybody personalized feedback at scale, and maybe even when you’re going through and progressing to the next stage, there is a lack of information about what you can expect, what’s happening. And maybe you have a question at 8pm on a Saturday and you’ve got nobody to kind of talk to that’s the only time you’ve had some headspace all week to kind of, you know, think about the process. So there’s a serious lack of human capacity generally within talent teams to actually have the level of kind of high touch communication that is required with every candidate throughout the process. And that’s really where things fall down.
Matt Alder [00:12:32]:
Absolutely. And obviously we have this fantastic opportunity with AI to not just fix this, but to kind of make it better. How does that work? How does AI kind of improve the candidate experience from your perspective? And I suppose particularly when it comes to sort of mass personalization, because I think that’s something that as an industry we’ve never been able to get right in any way, shape or form. So where do you think we are? Where do you think we’re going with this?
Sam Dhesi [00:13:01]:
I think where AI really comes into the hiring process, particularly the candidate journey, is in the top half of the funnel. When there’s a lot of volume, there’s not enough human capacity to serve that volume. And what ends up happening is if you’ve got a very large kind of volume of candidates entering the top of the funnel, let’s say there’s a thousand across a few roles, you might, you might not get around to assessing half of them and you may well even end up kind of jading them because of the poor experience. And you may well be missing out on some great candidates. And those that are going through the experience will have a very long process and that will increase your time and cost to hire. So there’s a huge opportunity there at the top of the funnel to actually address some of those repetitive manual, high volume touch points with candidates using AI. So, for example, we can now look at every application that comes in, in real time and assess it just like a human would. You know, you can actually program products like Popp to think just like you might think about a candidate’s application and to actually provide a really detailed candidate report that would otherwise be kind of sitting in your mind kind of fleetingly and actually, you know, put that in front of you and do that consistently across thousands and thousands of candidates and give you all the information you need to make a decision. And therefore the important bit, the decision is then your role as a human. But the processing of all of the data then goes away. And that’s a huge amount of time saving. That means suddenly, just with that alone, you can suddenly massively reduce the time to hire. You can increase your ability to get back to candidates. Because now, having processed all of that information, you can actually provide constructive personalized feedback to the ones you’re not taking forward, to those that you are taking forward, you can actually give them great feedback on why they’re being taken forward. I know this gets to your point around personalization, then you can even take it one step further. And if there’s a product that also combines conversational AI, then we’re talking about reaching out to these candidates, scheduling interviews on Autopilot, providing answers to questions they have questions that might be, why did you choose me? Can you give me some feedback? We have all of that information already because we actually process that data. We did that step at the very beginning, so we can absolutely dive into that. But that’s my response at a high level.
Matt Alder [00:15:45]:
What else do you think might be possible in terms of improving the candidate experience, making it feel more personal, more sort of tailored to the individual?
Sam Dhesi [00:15:54]:
Yeah, absolutely. So just to double click on some of the areas I mentioned there, so one of the things that we certainly do is take the candidates that you’re not going to be taking forward and have a look at what the gaps are in what you are looking for, sorry, the delta between what you were looking for and what the candidate presented with. And then using this is a great application of generative AI. But using generative AI to create a really personalized human like email that actually outlines why the candidate’s not being taken forward, some pointers for improvement, and actually can even point them, signpost them to certain resources on the Internet to actually fill those gaps. We can even within the same message, the same email, actually analyze other roles that might be available in your ats and explain to the candidate why that might be a great fit for those other roles based on the characteristics of those roles and the characteristics of what the candidate applied with. So there’s a huge amount of personalization here. We’ve got constructive feedback, we’ve got recommendations for the roles and explanation that actually takes into account the candidate’s profile. And that isn’t just in a single feedback email for a candidate that’s being rejected at this stage. So there’s a great experience there, even for those that aren’t being taken forward. And for those that are being taken forward, of course, we’ve already talked about reaching out and personalizing kind of feedback for those who want to know why they’re being taken forward. But actually we can go much further than this. Like if you look at the entire candidate journey, I mean, what we’ve talked about is really the stuff that happens at the top half. But what about as candidates progress and they go from interview one and two and maybe there’s a third interview, maybe they want to know what the feedback was based on those interviews and what they can do to improve and maybe what they need to look out for when it comes to maybe the third stage interview. And so if that is being recorded somewhere and you’re happy for that to be in some form shared with the candidate along the process, what Popp can allow you to do is actually allow the candidate to ask questions in the same thread about what they can improve, how they performed, and actually get responses back that were based on their actual interviews. Right. So that’s one great step. And then even further from that, how about when you’ve hired somebody, you’ve given them an offer, they’ve accepted it, and now there’s three or six months before their first day, there’s a real risk that they could be poached, they could drop off. How do you keep them engaged and nurtured? Well, at that stage, again, Popp can come in and we can begin to engage the candidate over a period of time, periodically remind them why we think they’re great, drip feed them information that’s tailored for them. There’s all kinds of opportunities across the entire journey.
Matt Alder [00:19:00]:
Absolutely fascinating stuff. And from the candidate’s perspective, obviously, in the same way that TA teams are enhancing what they do with with AI, candidates are obviously moving in that direction as well and using AI to help them at various stages of the application process. What’s your kind of take on that? Are we sort of moving to a world where it’s like the candidate’s AI is talking to the employer’s AI and the humans are kind of out of the loop? What do you think? How do you think that’s going to progress and what are the implications?
Sam Dhesi [00:19:33]:
Yeah, sure. So I think there’s this deluge of applications that have been created with AI and now that’s really driving up application volumes. And often what we’re seeing is that there are some embellishments of somebody’s achievements or skills and experience that have been created by ChatGPT or some kind of large language model that the candidate has used and also fed in the job description. So how do we combat this? Well, there are two kind of ways that we can kind of think about this. Now you’ve mentioned one of them and that is almost using AI to battle with AI. So if people are going to continue to apply with these kind of AI generated resumes, then you can have an AI system in place which is handling that really high Volume. But what that doesn’t really fix is if somebody is really tailored their CV to a job description and the requirements then and you know, they’ve actually ended up kind of embellishing some of the skills and experience they have and that could pass an initial kind of basic AI screening system. So how can we prevent that from happening? Well, one of the things that we’re seeing our clients do is actually be less kind of granular in some of the requirements for a particular role and be a little bit more general about kind of the what, what a candidate can expect from the role, what’s expected of them, generally kind of what kind of experience that they’re looking for. What this means is that even if a CV resume is tailored to a job description, as long as you have kind of a more granular screening process in place, then it’s very difficult for them to overcome that and know how to overcome that. So that’s one, one area. The second area is let’s say there’s a world in which actually we don’t want to necessarily use, we don’t want to allow a candidate to apply with a resume or a CV or a cover letter because of the risk of this. And actually we just want to, you know, we’re going to actually assess them and we’re going to try and assess them in a way that it’s going to be difficult for them to use AI to assist them. So again, this is something that many of our clients demand. They’ll skip the screening process and they’ll actually just use the conversational AI element where we can actually have a conversation with a candidate. We have a scorecard that we know that we’re going to be assessing candidates responses against. Now the candidate hasn’t seen that scorecard, but we know what it is and we can ask them questions. And as they respond, whether it’s in text or using audio or uploading a video, we can actually assess their responses against that scorecard which is sitting behind the scenes. So this is actually mimicking a first stage interview. Right. And you can now do this at scale using AI. You don’t need to do this with, with a person. And the output is essentially a percentage score telling you kind of, you know, how closely matched the candidate was to a scorecard. So obviously that is something that is very difficult for a candidate to overcome, to score extremely well, even if they’re kind of enhanced with AI.
Matt Alder [00:22:49]:
And have you got any sort of success stories from the companies that you work with that you can share with us about using AI tools in their process.
Sam Dhesi [00:22:57]:
Yeah, absolutely. I mean, there’s one very popular, very fast growing digital health brand that everyone will be aware of. I can’t, I can’t actually say who it is because I haven’t got pre approval from them, but it says a very popular brand, often featured in Marks and Spencer’s. And they’ve actually saved, I mean, they launched I think about nine months ago and they’ve actually saved over two months of time by using pop. So their talent team have saved of, of the nine months of human effort it would have taken them ordinarily without pop. Actually, two months of that was freed up across the talent team because they had Popp in place, which is really quite remarkable. And what that means is that they were actually spending far more time at the end of the funnel with the candidates that were engaged and qualified. And the experience both for the recruiters, the TA professionals using it and the candidates has been clearly much better. There’s another huge global kind of staffing brand that we’re working with called Randstad. So you know, it’s the largest staffing company in the world and you know, we’re being used across many of their kind of enterprise clients. And you know, with Randstad, we’ve actually, and this is proven over tens of thousands of candidates they processed using Popp, but we’ve actually reduced the time to screen by 50% so we’ve halved it, you know, and that’s happened over a series of months. We’re seeing an average candidate experience rating of 4.8 out of 5 with every pop powered interaction, you know, and this is a massive scale. So there is some pretty remarkable stuff. And there’s just one other thing, Matt, that I want to mention because I think it’s a great story about kind of how powerful Popp can be in certain situations. And we, we’ve actually. So again, this is a another Ransad related story, but there was a recruiter in India who logged into Popp and actually just before he’d been given a task to hire to actually schedule interviews with for Mexican candidates for roles that were based in Mexico. The problem was he didn’t speak a word of Spanish, right? This, this Indian recruiter only spoke English. And but what he was able to do was log into pop, create a digital version of himself. He was able to look at, actually reach out to 300, over 300 Mexican candidates that were in the database. Popp immediately recognized that they were responding in Spanish, switched to Spanish, had a whole AI powered conversation in Spanish and what the recruiter could see was all in English. And what ended up happening was that within three days he was actually able to reach out to 300 candidates and schedule 90 interviews. And most of that was completely hands off. So it’s pretty remarkable what you can do with this kind of technology.
Matt Alder [00:26:14]:
No, fantastic stuff. And I think it’s really important to share these stories because so much of the narrative and the content around AI for the last sort of 18 months is very theoretical, you know, very opinionated. And, and yeah, it’s great to see the tangible results that people are getting and the type of results that people are getting. As a final question, what do you think the future looks like? Where do you think hiring is going to be in say, three years time?
Sam Dhesi [00:26:42]:
I think the role of the TA professional will evolve considerably and it will become less operational and more strategic. So I think we’re going to see, and we’re already seeing this certainly with our product, but what we’ll see is TA teams that are less focused on performing really manual, repetitive tasks as the ones that we’ve discussed and actually become experts in using tools that allow them to get a lot more done and certainly allow them to automate and eliminate some of that manual admin. And this might look like slightly smaller TA teams or it might be similar sized TA teams. It’s just that they’re doing a lot more and they’re able to process a lot more volume. It will look like, you know, the daily life of a TA professional, I think, will be very, very different. I think they’ll be using a lot more technology. And I think, you know, there’s, there’s, there’s also the question of, you know, whether a lot of this tooling is going to become consolidated or whether, you know, they’ll be using kind of different products and different tools that interact and integrate with the system of record. Our view is that things will become consolidated. Obviously that is what we are building towards. And actually the system of record as it’s known today is going to change dramatically. It might not look like a CRM as it does today. I think things are going to be very, very different in the future.
Matt Alder [00:28:17]:
I couldn’t agree with you more, Sam. Thank you very much for talking to me.
Sam Dhesi [00:28:21]:
Thank you, Matt.
Matt Alder [00:28:22]:
My thanks to Sam. You can follow this podcast on Apple Podcasts, on Spotify or wherever you get your podcasts. You can search all the past episodes at recruitingfuture.com on that site. You can also subscribe to our weekly newsletter, Recruiting Future Feast and get the inside track on everything that’s coming up on the show. Thanks very much for listening. I’ll be back next time, and I hope you’ll join me.
