# Balancing Automation & Human Touch | HRTech Europe Discover AI's impact on recruitment and get practical advice on how HR leaders can leverage automation effectively while keeping the human element front and center. ## Watch the Replay ### Balancing Automation & Human Touch Duration: 18:40 Well, can everyone hear me? Fantastic. Lovely to see you all. Thanks for coming. Today, I want to talk to you about automation, efficiency, AI. And I guess kind of, why am I talking about this? Well, I'm old enough to have seen many, many generations of AI. I'm old enough to remember expert systems and machine learning, or maths, as most of us call it, but I think it's really important to understand how AI fits the new landscape and how it exists with the other other tools we have. So my name is Adam Bird, co founder and CEO of Cronofy. I'm one of those rare CEOs that can actually code, so I know the tech, although I would probably know enough to be dangerous rather than useful as far as my team are concerned. And just public service announcement, if you want to talk about any aspects of what we discussed or what I'm sharing today, we have pastries over at the stand afterwards, it's just that way. You can't miss it. The colors are like this. So let's get going. Cronofy is actually a technology provider to lots of software companies. So you've ever done any kind of scheduling on the internet, there's a good chance that Cronofy is the technology that's powering that under the hood. Now you'll see something you recognize, GoDaddy, Indeed. You book an interview on Indeed, Cronofy powers that, now with that, we ended up being quite well represented in interview scheduling, because interview scheduling is a complicated scheduling operation, kind of. Who knew, really? And so over 70 applicant tracking systems are using Cronofy to power their scheduling, which gave us a really interesting insight into the challenges and the flexibility required, and how there are many scheduling operations which are simple and repeatable, with the recruitment, it gets very complicated, very, very quickly. And then what we saw were, whilst these technology teams could build a certain amount of capabilities with our technology, recruiters were being underserved, so we built our own integrations into a series of the leading applicant tracking systems to really prove what was possible. So we've worked very closely with product managers of recruitment software and the recruiters directly to deeply understand the challenges around interview scheduling. So that's kind of who we are, why we're in the space. So why are we talking today? Applications per hire tripled in the last three years, so we run an annual survey. Last year was 12,000 candidates, where anyone who'd been in for any interview, we talked to them across seven different countries, you know, France, Netherlands, Germany, UK, America, Canada, Spain, another one. And what we're seeing is the kind of the use of AI in general. There's a there's kind of a shift in the market where recruiters are drowning in applications, and so people are trying to look at ways they can optimize their flow, because candidates are saying, well, you know what? This technology exists. Let me start using it now. The stats I have here are from our survey in 2024 which is like the first half of 2024 we're just kicking off our 2025 survey right now, just finalizing the questions that we're going to be answering. So one thing you can do is that you can kind of register for early access to the 2025 survey. But we're really seeing some interesting trends. So as of about probably nine months ago, candidates were already using chatGPT, Perplexity, etc, to improve and essentially apply at scale to roles, which, of course, creates more jobs for recruiters, and they're increasingly using tools to mass apply as well. You know, if you can kind of tailor your CV at the same time, why not just mass apply? And you know, you've got nothing to lose, but what, what's been seen in that, and this is what we're looking at in so again, 2024 data, is waiting times for that first interview has gone up considerably, and it's a real symptom of recruiters drowning in the work. Now, yes, we want to get the great candidates in really, really quickly, but candidates aren't being got in. And what one of the challenges with moving candidates through a process is if they wait too long, they disengage. So 37% of candidates in our survey last year said they would disengage with roles after a month or more. Now what that means is, if you don't get to those candidates quickly, they're essentially going to your competitors. They're going to those roles. That's it. Is the kind of role of rule of the market, I guess, really, and what we're trying to do is, or what I hope you will learn today is looking look for opportunities to improve that without losing that human touch, because a massive 81% of candidates said a human touch is important, which, yeah, come on, guys, you're using technology to mass apply and you know, but you want us to treat you like a human candidate's kind of candidate? Yeah, I'm a father of children. Anyway, actually, most people have parents of children. That's how it works. But anyway, we'll skip over that. And that's the problem. We talked about the kind of delays in interview scheduling. We talked about kind of all aspects of it, when you're drowning in just processing the applications and trying to get people through the process, then there isn't time for that human touch. So but there is technology out there, and if you kind of listen to lots of the vendors here, they're all going to solve your problems, they're all going to give you this time. What I'd like to do is not necessarily propose a particular AI tool or technology that you should be using, but I think talk in more general terms about how you think about composing your workflows and composing your capabilities. So how do you cope with three times more work, and fundamentally, you can't hire this away. Gone are the days, I think, where people would basically hire a bunch of coordinators out in outsourcing the Philippines or wherever, to just go through the process and try and get people through. That doesn't scale and that doesn't work. So that's kind of, that's what I think, where we are in the market that we're trying to cope with increased kind of workload from candidates. We're trying to find the best candidates for our company. And there's lots of tools out there. Now I'm here to break it to you that AI is not magic, it's just software. That's all it is. But there's two words that come up quite a lot when we talk about software. There's deterministic and probabilistic, and they're horrible words, but basically deterministic software is software as we know it. It's kind of, it's the same answer every time, whereas probabilistic, it's kind of, your chat GPT, your kind of, LLMs, it looks like the right answer. And I deliberately chose those words. It looks like the right answer. It's not necessarily. It's not right. Doesn't know it's right. It just looks like it's right. So deterministic software is software as we know it, same inputs, same outputs. You can predict it. So we use this for commission calculations for finding out whether interviewers are available, for scoring multiple choice. This is why multiple choice exists. It's because software can't interpret or deterministic software can't interpret an answer. So therefore you need to have a kind of multiple choice or fixed answer questionnaires in order to do that. Now probabilistic software is different, same inputs, you get the different outputs, but it's probably correct, and that's all around that's what training is all about, is trying to get to a point where it's probably correct and it's really, really good for conversational interactions, interpreting free text answers against a framework and summarizing free text, but remember, it's probably correct. So you've got these, this new type of component in your workflow. So pre AI, we had humans and deterministic software. With AI, we've got humans deterministic software and probabilistic software. That's the only thing that's changed. The AI is not replacing anything. It's just another bit of tooling in our in our tool set. Now this software, probabilistic software, makes mistakes by design, but then so do humans. Humans already made mistakes in many ways. That's how we've evolved. The reason humans are have kind of succeeded, as it were, as the just the apex species on the planet, is because we make mistakes, and sometimes those mistakes turn out to be a new way of doing things. And so our business workflows already cope with those mistakes. There's already the checks and balances. There's already processes, processes to follow, and a company's risk appetite is what decides how they control those mistakes, those chances or mistakes. So if you're doing kind of consumer customer service doesn't really matter. The Jeopardy isn't great. If you're processing mortgage applications, Jeopardy is quite high, because a mistake could take down the company. So what I would encourage you to do is, if you're starting to look at kind of implementing AI tools into your workflows, and I would absolutely encourage you to do that, treat them like humans. Treat them like they make mistakes. And think about the checks and balances you need within your workflows in order to protect against those mistakes. It is just software, but it's software that you cannot completely trust. But what's different about this software to humans is you can hire it for an hour a week, 1000 hours a week, 10,000 hours a week, doesn't really matter. It only really does one thing, application, assessment, compensation, interactions, it will be very good at doing one thing. It doesn't have any human HR challenges. It's not going to take a sickie, it's not going to come in drunk is not going to break the photocopy at the Christmas party, but mistakes are inevitable. You have to be able to understand that these mistakes are going to happen, and so it's up to you to think about your workflows. If you had an intern doing this process, what would you do in order to check that work was happening? So let's talk through an example for recruiting. So I've got to go on this perfectly kind of wonderful flow that's going to get all the way to first interview for candidates. So firstly, an AI agent is going to process a job application and decide whether a candidate should be contacted. So based on whether it's a CV, the application, letter, internet research, it's going to work out whether or not this candidate should transition to that next stage, and then another AI agent is going to contact the candidate by text or voice to ask them a series of screening questions. So services like Scotty AI or Hubert do have do this kind of process where they'll actually kind of interact with the candidate and allow them to kind of speak or chat in kind of natural language to assess their fit. If they pass the screen at that point, then they can hand off to a scheduling tool to schedule that first interview, and then scheduling tool hands back into the workflow just to kind of tidy up with the candidate. So massively scalable, you've got four different agents, three, probabilistic, one, deterministic, running this process. But at each of those stages, you're going to want to check the efficacy of that and make sure that these things are working. Because again, think of it like you've got three interns and some software, and you want to make sure that intern is doing the right job. So you'll be looking at individual step conversions. Are too many people getting through to the next stage? Are too few? Are candidates dropping off the process? Are they kind of bouncing off because it's not human enough? Have you kind of missed out on an opportunity to inject a human aspect to the process. Are you kind of reducing your time to schedule any candidate experience you're doing? What's the feedback on the process? Now, with our surveys, we always ask candidates' appetite for automation. And actually candidates have a strong appetite for automation within the scheduling flow. They don't want their time wasted. So it's not the slide, but it's about 38% I think, of candidates want some kind of automation with various aspects of their flow. They recognize that this process should be automated. I just want to take a slight diversion to talk about scheduling and AI. Now our view, my view, is that scheduling is not suitable for a probabilistic engine. And this really comes down to the kind of scope of the frame of reference for these processes. So if you think about answering questions, the frame of reference is quite tight, quite small. There are kind of right or wrong answers. You've got a kind of small kind of body of knowledge. Whereas, if you're deciding whether or not a calendar event should be booked over, the frame of reference is incredibly large. The information is not in the calendar. It could well be just in people's minds. They know that that meeting is more important, because this is a partner relationship that they've been working on for many, many years. And the LLM or other system can never be have a chance of knowing that no, not even neuralink wherever that's going, it's going to give us that chance. And the problem is, getting it wrong is high risk. So this is where having predictable rules and heuristics, controls around offer times, availability, interview pools are all something that people need to set up the rules so they can predict the outcome, so therefore they can trust what's happening and they're not rescheduling candidates or last minute booking. This is a message from our sponsors, because, hey, this is what we do. So something we've been working on very closely is dealing with busy calendars. How do you allow recruiters to, or essentially hiring managers, to mark time in their calendar where they can be booked over? So something we call interview times. So a typical flow of this is hiring manager speaks to the recruiting team. I want this person to start in three months. Okay, that's great. What's your availability for interviewing week six, seven and eight? Mark that in their calendar and but allow them to change that. And importantly, what we're doing there is we're putting something in their block, in their calendar, the hiring manager can then choose to move that but importantly, they're reminded of that commitment, and that's what's really important, because life does change. You know, a meeting with a client, a meeting with a exec, whatever it is, these things can happen, but the hiring manager gets reminded that they've made that commitment, and then they can move the block around, but still make sure they make enough time for interviewing, because they've got to get that candidate in. Now, we've been doing a lot of work with Wise, Mark at Wise. Super keen on automation, super keen on kind of optimizing the flow. You know, one of the things we did with the scheduling is we reduced this, the scheduling, the time to schedule an interview, from six days to 90 minutes by taking a different approach. So they're now conducting the interviews and the time it used to take them to schedule the interviews. So you can imagine the kind of commensurate kind of impact on time to hire. So this is by judicious application of kind of automation or task of tools to solve these problems. They just picked a very, very small problem, worked on that, and now that's part of their flow. And then we kind of work on the other things. And what's great is candidates love it. So at the time Wise, have just going live with a candidate experience tool. But before they did that, candidates were emailing in saying, this is brilliant. Now that is no one praises people, least of all candidates, so that that's that's quite an impact. And really, what they're finding is the key thing is it's making space for more human in their flow, because they don't have to think about the kind of the mundane, the repeatable tasks, the recruiters are now almost treating it like being a master check in the work of apprentice. I talk about this idea of imagine you've got an intern doing this work for you. You're going to check that they're still doing the right thing. You're going to check that they're going to optimize you're going to look over their shoulder and make sure that process is effective, but also allows you to provide more personal touch points. You can reach out to candidates, you can check in with candidates, you can automate a reminder, but allow the candidates to call in. But also those people can upskill as well, so they become less of a sort of admin person for the hiring manager, more of a talent partner. And I think that's a real change that they're seeing. And another nice testimonial, basically, one of our coordinators, using our tool is now spending her time developing her career, because she's not having to spend the time chasing up candidates. So that's it. I would really, really encourage you to leverage automation. But the goal here is not just to make things more effective. It's to deliver more humanity to the recruitment process, because, as I said back on the previous slide, that's how you can differentiate. That's how you can kind of lock candidates into the process. Just over 80% of candidates said a human touch is important to them, and if you're one of the few recruitment teams that can do that, you get so much more license with your your candidates. So thank you for your time. Pastries.let's talk! [Download Presentation Slides](https://cdn.sanity.io/files/ygfbhkqr/production/ffeddc3c5b101fcf98488762396e2fbfdff28532.pdf) [2025 Candidate Expectations Report](https://www.cronofy.com/2025-candidate-report-early-access) ## Key Takeaways #### **AI's Impact on Recruitment Efficiency** - Applications per hire have tripled over the past three years, and candidates increasingly use AI tools (like ChatGPT and Perplexity) to apply to multiple jobs, creating more work for recruiters. - AI is essential in streamlining recruitment, from application screening to interview scheduling. **** #### **Human Touch + AI/Automation = Better Recruitment** - 81% of candidates believe that a human touch is crucial during recruitment. Automating repetitive tasks is great, but human interaction is key to maintaining engagement and building candidate relationships. - AI/automation should be seen as an assistant, not a replacement. By treating it like a team member, recruiters can avoid mistakes and optimize their workflows. **** #### **Beyond Buzzwords, How to Identify the Right Tool ** - Deterministic tools handle predictable tasks, while probabilistic tools are best for complex, conversational tasks. Understanding the distinction helps recruiters choose the right AI tools for different stages of the recruitment process. Not all processes need - or can be highly efficient with - AI; some only need automation. - AI can streamline applications, screen candidates, and even schedule interviews, but recruiters need to monitor and refine the system to avoid mistakes and ensure effectiveness. Tools that support existing workflows - rather than replace them - should be encouraged. **** #### **Automating Interview Scheduling** - Scheduling with AI is tricky, but when done right, it can save recruiters valuable time. Predictable rules are essential for minimizing the risk of scheduling errors. - Cronofy’s best-in-class solution has helped companies like Wise automate scheduling without sacrificing quality or reliability. #### **Impact on Candidate Experience** - Automation, when implemented thoughtfully, can improve candidate experience by reducing wait times and offering more convenience. - Wise’s positive candidate feedback demonstrates the value of combining automation with personal engagement. ### How Wise cut interview scheduling from 6 days to 90 minutes Learn how Wise, a global technology company with a mission to move the world’s money, accelerated their time-to-schedule with Cronofy’s scheduling automation and transformed their candidate experience. Case Study --- ### Are interview scheduling delays costing you top talent? Your time-to-schedule isn’t just a metric for coordinators, it’s now a critical differentiator in winning top talent. But why are teams still stuck with interview availability bottlenecks? Read on until the end to see how to solve this conundrum and empower both recruiters and interviewers with one process, and join an upcoming live session to learn more. Blog Post 2025-02-05T12:28:58.234Z --- ### Maximize Interviewer Availability for Competitive Hiring Scheduling delays cost you top talent. Discover how Cronofy’s new features help reduce availability bottlenecks and enable you to plan interviews smarter. Blog Post 2025-02-28T10:11:15.976Z