Application volumes are climbing fast, and AI has made it far easier for candidates to produce a strong-looking resume. For talent teams trying to give every candidate a fair hearing, the traditional model of recruiting is starting to break down.
Some employers are now handing the first conversation to an AI voice agent, and that raises some obvious concerns. Does automating the first step strip out the human connection that recruiting depends on? The teams doing this well are finding the answer isn’t what a lot of recruiters expect, and that getting it right depends as much on how openly it’s done as on the technology itself.
So what does it look like in practice?
My guests this week are Jean-Baptiste Anne, Global Director of Talent Acquisition at Mirakl, and Anneliese Muscari, their Head of AMER and Global Go To Market talent acquisition. In our conversation, they share why they made the change, how candidates have responded, and what it means for the future of the recruiter role.
In the interview, we discuss:
• Managing unprecedented application volume
• The growing limitations of resumes
• How do you give every applicant a fair chance and a great candidate experience?
• Handing the first conversation to an AI voice agent
• Moving from skepticism to trust
• Publishing AI guidelines for candidates
• How candidates have responded
• Keeping humans in charge of every decision
• What does the future look like?
Follow this podcast on Apple Podcasts.
Follow this podcast on Spotify.
Matt Alder [00:00:00]
Recruiters are wary of letting AI run a candidate conversation, and candidates are wary of being on the other end of one. So how do you win over both sides and still get the value that AI promises? Keep listening to find out.
Sponsor message — Maki
Matt Alder [00:00:17]
Support for this podcast comes from Maki. Maki began by replacing the resume screen with a fair, structured voice interview that assesses real skills before anyone formally applies. Now that same intelligence is extending across the whole funnel, from the first conversation to the final decision. They recently launched Tomo, an AI interview assistant for hiring managers, the next step towards one connected system that screens, interviews, and gives every candidate a consistent, fair experience at scale. See how the end-to-end picture comes together by going to makipeople.com. That’s makipeople.com. And Maki is spelled M-A-K-I.
Introduction
Matt Alder [00:01:22]
Hi there, and welcome to episode 807 of Recruiting Future with me, Matt Alder. Application volumes are climbing fast, and AI has made it far easier for candidates to produce a strong-looking resume. For talent teams trying to give every candidate a fair hearing, the traditional model of recruiting is starting to break down. Some employers are now handing the first conversation over to an AI voice agent, and that raises some obvious concerns. Does automating the first step strip out the human connection that recruiting depends on? The teams doing this well are finding the answer isn’t what a lot of recruiters expect, and that getting it right depends as much on how openly it’s done as on the technology itself. So what does this look like in practice?
My guests this week are Jean-Baptiste Anne, Global Director of Talent Acquisition at Mirakl, and Anneliese Muscari, their Head of AMER and Global Go-to-Market Talent Acquisition. In our conversation, they share why they made the change to AI interviewing, how their candidates have responded, and what it means for the future role of the recruiter.
Hi, JB. Hi, Anneliese. Welcome to the podcast. It is a pleasure to have you on the show. Could we just start off with you introducing yourselves and telling everyone what you do?
The conversation
Jean-Baptiste Anne [00:02:49]
So I’ll start. I’m JB. I’m the Global Director of Talent Acquisition, Permanent, at Mirakl. We are based in Paris and operating across the globe, and Anneliese is a testament to that. We are running a full TA team across the globe and across all types of, let’s say, jobs. So we are hiring tech, G&A, and go-to-market roles from Boston to Australia. And maybe a quick thing about my background, because I’m not a traditional HR guy or recruiting guy. I took the lead of the TA team two years ago, but I joined Mirakl four years ago to lead the B2B customer success team. I have a background of, let’s say, engineering consulting, and I worked in global corporate companies like Société Générale in France or Amazon in Europe.
Matt Alder [00:03:54]
Fantastic. Anneliese, tell us about yourself.
Anneliese Muscari [00:03:56]
Sure. Hi, Matt. Thank you so much for having us on. My name is Anneliese Muscari. I oversee all go-to-market hiring globally. So, as JB mentioned, that includes the AMER, LATAM, and APAC regions, and then also EMEA as well. I do come from more of a traditional recruiting background, starting in agency, then spending some time in corporate, and then most recently, four and a half years ago, joining Mirakl.
Matt Alder [00:04:26]
Fantastic. So, obviously, lots of things going on in the market at the moment, so let’s start with this. What are the biggest challenges that you’re seeing in recruiting right now?
Jean-Baptiste Anne [00:04:38]
I think, like everyone, we have seen a huge spike in the volume of applications. Last year, we processed more than 95,000 applications, and it’s growing year on year. As an example, on internships alone, we received 14,000 applicants, and it’s growing by 43% year on year. So this is really huge, and honestly too much. When I took the role, I had an engineering reflex, saying, you know, more inputs mean more outputs. So I was happy with the volume, thinking maybe it’s going the right way.
Jean-Baptiste Anne [00:05:18]
It was a hard thing to understand, after a couple of months, that no, not at all. The quality signal is not directly linked to the number and the volume of applicants, not at all. And I would say it’s even worse, if I might say, with AI, when everyone can polish their application and it becomes almost free to apply. The big challenge we are facing is how to manage that high number of applications while at the same time — and this is our key obsession at Mirakl — providing as good a candidate experience as we can. Everyone is an individual. Someone applies to us, and even if there are a lot of them, there’s a human behind each application. Our strategy and our thinking around this is to make sure that everyone gets feedback, whether it’s positive or negative — that’s not the point — but valuable feedback, so they can learn from our application process. So on our side, the biggest challenge is the volume, and at the same time finding the right balance with a great candidate experience.
Anneliese Muscari [00:06:37]
I’d agree with that. I think we’re dealing with unprecedented application volumes, with sometimes seemingly perfect resumes, because everyone can use AI to polish them. And recruiters are now tasked with going through this very high volume of applications while still providing an excellent candidate experience for every single person who applied. It’s causing a lot of friction in the recruitment process, and it was something that was a huge priority for us to try and solve.
Matt Alder [00:07:11]
So you’ve also been going through a few layers of AI transformation. What was the motivation behind that, and how did you start that process?
Jean-Baptiste Anne [00:07:19]
Just to be super honest, I would say it didn’t start with a vision. We faced that level of applications, and it was a math problem. And again, maybe it’s because of my engineering background, but it was like: we need to find a way to manage through this volume. Last year, as a company, we generated more than 9,000 scorecards. So we needed a way to go through that volume challenge. The idea was not to use AI just for the sake of using AI, or to use a solution that would serve everything. No, let’s go back and say, okay, how can we face this?
Jean-Baptiste Anne [00:08:08]
So we started to explore a couple of things, and it was very clear that we needed to find a way to allow the team to take a better decision when reviewing applications. Our main objective was to find a solution that would sit at the very beginning of the funnel — from application to, let’s say, TA, talent acquisition screening. In that way, we would be able to filter out much more of the application volume, and without guesswork. My mission here is to gather more information so the team can take a conscious decision on whether to move forward. The challenge, when you are facing such a volume, is…
Jean-Baptiste Anne [00:09:00]
…is that you don’t have time, and so you are looking for the perfect candidate. Through that volume, honestly, it’s pretty easy to pick, I don’t know, let’s say 50 or 100 almost perfect candidates and move them forward. And honestly, we were missing some talent, because they are not that perfect, but the team has a difficult choice to make every time they see an application. It’s either I invest one hour meeting with you — with the preparation, with the scorecard — or I pass. And it’s easy to pass, even on the best candidate, because we have so many of them. The challenge is we only have, you know, one page, or let’s say two pages, on the resume, and maybe a LinkedIn profile, and that’s it.
Jean-Baptiste Anne [00:09:53]
It’s really hard to take a great decision based only on something that you can also pretend, that you can make up or polish through AI. So the idea was to find a way that is effective and scalable, where we can say, okay, we can have, I don’t know, 90,000 applications running through that system for us. It’s both scalable and, at the same time, great for the candidate experience, because if we add too much friction to our application process, the best talent won’t apply to us because it’s too complicated. It’s not a good time investment from their perspective. This is why we thought — the beginning of the discussion was how we could survive, if I might say, on 10x the application volume. It’s not even 90,000, it could be 200,000, whatever. We needed to rebuild something, not as an efficiency play, saying, okay, we are going to absorb 10 or 20% more. It was more: what are we going to put in place, any process or system, and I think we will discuss that later, that means we can do the same but with a million applicants or whatever.
Matt Alder [00:11:22]
Yeah, absolutely. And you’re not sacrificing — well, you’re improving candidate experience and improving quality at the same time. So it makes perfect sense. Let’s talk about the technology here. What role has technology played in that, and what is it that you put in place?
Jean-Baptiste Anne [00:11:39]
I think technology and AI are one of the solutions that can absorb the volume and scale with a limited cost. When I speak about cost, I mean everything from resources to euros and dollars. And it’s also able now to provide a pretty good experience, even a great experience, mostly through vocal AI agents. This is why, you know, the very good news is that we started with an issue and said, okay, we can take the technology to offer both at the same time. Because we can always say, let’s add a ton of questionnaires, let’s add a motivational letter, but that’s not providing a great experience. So, the technology — and because we did this in late 2025 — we started to discover the whole world of AI vocal agents, because I think the technology was mature enough to provide a great experience. I wouldn’t have said the same, and the conclusion and the decision wouldn’t have been the same, if we were in 2024 or even before.
Matt Alder [00:12:52]
Things have moved on very, very quickly, haven’t they? Anneliese, tell us a bit more about the tech.
Anneliese Muscari [00:12:55]
So we decided on a company called Maki People, which is sort of the background technology of Emma. You’ll hear us maybe speak about Emma throughout the rest of the conversation, but that’s who we’re calling our vocal AI agent. Maki is a great company that has a variety of types of AI assessments. They also have more in-depth skill assessments outside of their voice agents. And their scoring is based on science-backed data to help score and evaluate candidates. So we felt really comfortable and confident about not only the technology they provided, but also how they score and evaluate candidates.
Matt Alder [00:13:44]
I mean, it’s an interesting shift, isn’t it? Because recruiting is such a human-centric activity. As you said earlier, when you’re talking about the candidate experience — what’s it like to hand that first conversation over to an AI voice agent? And did your feelings around that change?
Anneliese Muscari [00:14:00]
Yes. Honestly, I was a little worried about it, I’m not afraid to admit it. I went into it with a great deal of scepticism, perhaps. As a recruiter, I really value that one-to-one human connection. I’ve really leveraged that both in the way that I recruit, but also as a core tenet of how I lead my team. So when thinking about handing off the first interaction to AI, it almost felt against those core values, and I really thought it could put our candidate experience at risk. On the other hand, I could see the logic clearly, particularly around the benefits of consistently providing a high bar on fairness and consistency. So as I started partnering with the Maki team to launch Emma, and was really involved in how we trained her and how she would be integrated into the recruitment process, I started to realise that perhaps I was thinking a bit too small about candidate experience, and maybe more so in a traditional way.
Anneliese Muscari [00:15:17]
Really, what Emma provides is the ability to take the first step with Mirakl at any time of day, anywhere, regardless of the recruiter who’s on the search or who reviews their resume. Candidates can be assured that their experience is the same as everyone else’s, graded on a consistent, science-backed rubric. And so I think, in a really competitive talent market, that consistency builds trust that ultimately can turn candidates, maybe even into advocates for the company, regardless of the outcome. So yes, my feelings did change as we went through the implementation and eventual launch.
Anneliese Muscari [00:16:02]
I think, too, the other thing I started to really realise is that this also freed up the TA team’s time to focus their energy on the interactions that require that real, in-depth human connection with candidates — to fully engage with them, understand their background, prepare them for what an interview could look like at Mirakl, but even also internally, to fully immerse ourselves with our hiring teams, coaching them, and ultimately spending more time with the business. So I also saw the benefits there, around having a bit more time to focus on some of those uniquely human connections. One thing I will highlight, though, is that Emma does not make any of the hiring decisions. At any critical juncture or decision-making point, a human is always making the decision. So that’s one piece that will not change, and it’s really important for us to emphasise.
Jean-Baptiste Anne [00:17:00]
And if I might add, because we spoke about the technology part — another part of the hiring experience is being as transparent as we can. Because it’s stressful to have an interview, right? It’s tough, and you need to be super focused. I would really love, as a candidate, to know what the process will be, where I will be assessed, on which parts. And so what we did with Anneliese and the team here was to be one of the first to publish — very early, if we compare to the other tech companies. In late September, early October, we published our AI guidelines for candidate interviews, to explain to everyone what we are going to do. For example, having transcripts, so that we, on our side, can better focus on the candidate during the interview. And also to describe to every candidate when and how they can use AI in the hiring process, and when they should not. For example, a one-to-one interview should be human-to-human. You shouldn’t be trying to read from a transcript just to give the perfect answer.
Jean-Baptiste Anne [00:18:23]
So it was a foundational piece of our strategy and vision — to say, okay, this is how we are going to use AI, and this is how we expect you to use AI. Because I think it’s important to mention to the candidates that it is expected, right, to use AI, because in their day-to-day it will be expected too. So when you are applying, we mention it, I think, at least five times, in every piece of the text: okay, you will be speaking with Emma, Emma is not a real person, and this is pretty much okay, so you need to be comfortable with that. And if you have any questions, please refer to your recruiter, or Google and read more in the AI guidelines.
Matt Alder [00:19:10]
I just think that’s so important. That level of transparency is so important because, as you say, there is a lot of mistrust of AI from candidates. And I think a lot of it is because companies aren’t being that transparent or that clear about AI in the recruiting process. How have the candidates responded — the people who have gone through this process? What’s the feedback been like?
Anneliese Muscari [00:19:32]
Yeah, candidates have actually responded really well. As of early June, almost 83% of participants have rated the experience as four or five stars. 79% of candidates found the process fair. And 80% of candidates actually had an improved perception of Mirakl after the experience, which is a metric I particularly like. So overall, the candidate experience and candidate feedback have been quite good. In fact, some of the qualitative feedback I’m thinking of — many candidates say that they forget they’re speaking with an AI when they connect with Emma. So we love seeing those comments as well. But yeah, I think a lot of it has to do with the transparency we show up front. To JB’s point, we have very clearly stated AI guidelines. And then throughout the recruitment process, we specifically call out where AI might be used. So in our intern campaign, where we’re using Emma, that’s both in the job description and in an automated message that everyone receives when they apply. So we are very upfront about our use of it, and I think that helps build some of the trust at the very beginning.
Matt Alder [00:20:51]
And what about the recruiting results? Has this solved some of the challenges that you were trying to solve?
Anneliese Muscari [00:20:57]
Yeah, I think so. We’ve seen some really exciting results. At a high level, the key win for us is that we are able to screen two and a half times more candidates than in our previous campaign. So two and a half times more candidates had the opportunity to be evaluated by Mirakl, which I think is also a great candidate experience signal, even outside of that efficiency gain. We’re also seeing that quality is maintained. In general, our conversion from TA screen to hiring manager screen has increased by 47%, and application to offer has increased by 7%. And when we look at a matched-pairs analysis — comparing roles that were open both in our previous campaign and this campaign — most notably, our application-to-offer conversion increased by 48%. So we’re actually seeing that stronger candidates are entering the funnel as well, and being converted to offers. So overall, we’re really excited about some of these results.
Matt Alder [00:22:09]
It’s just such an interesting time in terms of how this is going to develop. And this next question is something that’s much debated: where do you think the line between humans and machines should be drawn in the hiring process? Where will it be drawn in the hiring process?
Anneliese Muscari [00:22:26]
I could start with this. I think that machines will, and should, always support. They should be leveraged to improve our ability to make more objective, evidence-based hiring decisions. But I think humans should always be in the driver’s seat of making critical decisions and providing that great candidate experience. So for me, machines and AI play a supporting role, to enhance our ability to make great decisions, but they should ultimately never lead.
Jean-Baptiste Anne [00:23:03]
Thanks to AI and all the systems, I think we’ll be able to spend more time as humans — human to humans, right? Because my vision on this is that AI should take everything that is not the best thing in my day. And the best thing in my day is speaking to candidates. The best thing in my day is not scheduling meetings, finding availability from whoever, and so on. So by doing that, I think our job will be even better, because we’ll do what requires human judgment, what is difficult, and what makes our job interesting. So I see our job evolving — whatever the job title — more into the strategist, or the talent strategy, to think about the future, sitting down with the hiring manager and department leaders to think about what the best talent looks like for that open role, or the multiple roles that we have open in the department.
Jean-Baptiste Anne [00:24:07]
I think we’ll also go faster than ever, because sometimes — and that’s pretty much okay — it takes time just to learn from our good, or not so good, experiences. Maybe it’s not the right person, maybe it’s not the right skill set, maybe the compensation plan is not the right one. And it requires a ton of time just to take a step back, analyse, and make sure we understand why we’re not able to close the role.
Jean-Baptiste Anne [00:24:38]
And, as we speak, the target at Mirakl is that we need to fill a role in 55 days, which is pretty fast. Sometimes we are waiting — and most of the time we are waiting to hit the 45 days — just to start having that mechanism to take a step back: are we able to close in the next 10 or 15 days, the role, yes or no? If no, why? And if we’re just, if I might say, discovering at 45 or 50 days that with the past candidates we were all failing at a specific point or whatever, then I think the challenge will be: okay, we can have AI doing this, forcing the mechanism, just to reassess every time — as soon as we input a new scorecard, as soon as we see a new candidate, from all the stages. We can do that at scale. So I think AI will force that discipline, which was hard to catch because we are facing a volume, and we have fewer resources, I think, like every talent acquisition team in the world. So it’s really about applying the same bar and the same level of expectation, an almost mathematical or scientific approach, at scale, on every process.
Matt Alder [00:26:02]
Yeah, I think that makes a lot of sense. And it’s certainly an interesting journey in terms of how what we feel comfortable with is changing as well, when you can see the results come through. As a final question to you both: if we look two or three years into the future, how do you think applying for a job and hiring people is going to feel different from what it feels like now?
Anneliese Muscari [00:26:28]
I think it will feel very different. There could be a world where — and I know JB would like to see this — there might be no resumes. People aren’t using resumes anymore, where perhaps arranging an interview is fully self-scheduled. Maybe the interview process is done asynchronously. I think there will certainly be AI-led assessments versus live assessments. So it could look quite different. As we were mentioning, job seekers will engage with TA teams in different ways — I’d like to think on a more in-depth level than perhaps they have traditionally. But yeah, I think it could feel quite different from how it’s always been done.
Jean-Baptiste Anne [00:27:16]
Yes, I think in two or three years from now, I hope we’ll be moving away from having only a resume or a LinkedIn profile to take the first decision. I hope we find a way to completely remove that. And honestly…
Jean-Baptiste Anne [00:27:40]
…I’m pretty confident we are moving that way. I don’t know how many decades we have been reviewing applications, but I think it will change significantly, so I’m very happy about that. I’m optimistic about that future. Like Anneliese, I think all the admin tasks will be almost removed, I hope. And I also see that — maybe it’s old-fashioned — but we’ll spend more time, like I said earlier, together, sitting in the same room. We need to reconnect with the office, with the premises, with the feeling of being in a company — the atmosphere, the culture you can sometimes see written on the wall, or when you’re in the lift having some small talk. So I really see people being connected together, sitting in the office, even during the interview process. And I think everyone, including the candidates, will be more, if I might say, picky about all those very human parts of the experience.
Close
Matt Alder [00:29:00]
Fantastic. JB, Anneliese, thank you very much for talking to me.
My thanks to JB and Anneliese. You can follow this podcast on Apple Podcasts, on Spotify, or wherever you listen to 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.
