High ROI AI for Talent Teams: What to automate vs ignore

Hi, everyone, we're just gonna wait about… 30 seconds, a minute or so, just lets more people attend, and then we will… we'll crack on with today's session for you all.

Amazing, I think we'll crack on. So, hello everybody, welcome. I'm really glad you've all been able to join us today for this session. My name is Nathan, so I, head up our Talent Advisory Service here at Scede.

For those who haven't come across Scede before. We're a company that strongly believes in building world-class teams. And doing that being the cornerstone of any successful company that we work with. So, we've had the ability to see firsthand the hurdles that businesses face when it comes to acquiring and retaining top talent, especially when scaling quickly.

So, that's why Scede was founded back in 2013, to essentially be a recruitment partner that seamlessly integrates with a company's culture and hiring goals. Offering both the strategic guidance and the hands-on execution that actually makes an impact. What I'm going to do now, just very quickly, is hand over to Adam, who's joining us today from Teamtailor, to give you a little introduction to himself, and also to Teamtailor. And then what we'll do is we'll continue on with the webinar.

Good afternoon, everyone. Thanks for coming along. So yeah, I'm Adam Nichols. I'm the Customer Success Manager for the UK for Enterprise Clients for Teamtailor. My background is actually from TA, so I've spent the last 21 years in recruitment. Now I've switched over to Recruitment Tech, working with Teamtailor. Those of you that don't know who Teamtailor are, we're a leading ATS provider, really leading the way when it comes to AI within the ATS platform. So yeah, looking forward to going through this session with you guys today.

Amazing. So, before we kind of continue, we're… we do have the ability here with the webinar service with Zoom, so if anybody has any questions throughout the webinar. Please submit them in the Q&A section, we'll make sure we cover those all towards the end. And yeah, so what I'm here to do now is I'm gonna set the scene for what we're… what we're doing today.

Essentially, what we're here today to discuss is that hiring teams are under real pressure right now. I think we all know that, being talent acquisition professionals ourselves. So what we're seeing at the moment is more applications, flatter headcount, and essentially the same hours in the day as we had before, right? And alongside this, we've got an expectation to move fast. and hire well, and that expectation's not changed. So, one of the things we're hearing quite vocally, regardless of where you're looking at the moment for your recruitment information, is AI is getting thrown around as the answer to all of this. So, some of this conversation does live up to the hype, some of it doesn't.

And what we want to do today is cut through that, and give you something concrete, a clear sense of where AI genuinely removes friction, and where it's probably not worth your time. So, the aim for today's session is for you guys to leave with some ideas you can take back to your team, take back to your stakeholders, and you guys can engage in this, and… join the journey that we at Scede and our customers at Scede are also going through as well.

So, in regards to Scede's philosophy, let me share how we think about AI over at Scede, because it really does shape everything that we do with our customers. And I think it's a useful frame to start with before we get into any kind of examples. So our view at Scede is quite simple. AI doesn't replace great recruiters. It makes them significantly more capable. We'll talk about. frequently talk about, sorry, on LinkedIn being AI-enabled recruiters, and we're people that use these tools thoughtfully to get better outcomes, but we're… we're not people that have handed over judgment to a machine.

The way that we approach AI as a company at Scede is through experimentation, right? We don't come in with a rigid playbook. Or a one-size-fits-all solution. We're a company that tests, we learn, and then we measure what actually moves the needle for each of our customers. And we're happy when we're honest, when something isn't actually working. That mindset of experimentation is really important, because the AI landscape is moving fast, and what works today might look different in 6 months, 8 months, 12 months.

So now, in our experience, top of funnel is usually where the biggest drag is for most hiring teams. Volume of applications is up, significantly in a lot of sectors. But the time available to screen hasn't scaled with it. As we said earlier, we're stuck with the same 24 hours a day. So that's typically where we look first. These are repeatable, low-judgment tasks. That are quietly eating hours every single week. What we're able to then do by eliminating these is recruiters can redirect their energy towards the work that actually needs a human being.

Examples of this, as we all know, would be building relationships, assessing culture fit, making the case internally for a great candidate. These are the things we need a human for, these are the things that AI can't do. But here's the thing. Before you can fix any of that, you need to know where the hours are actually going, and that's where a lot of teams skip a step. So, before we rolled anything out here at Scede, we did a proper time study internally. This was across all of our teams, all of our functions, to truly understand where the pain points were. We tracked how people were spending their time.

What was creating the most friction, and where we were losing hours to work that wasn't moving things forward for us, our clients, or our stakeholders. That was especially important for us, because we work across a huge variety of customers. Different sectors, different hiring volumes, different team sizes, even different maturity levels when it comes to hiring processes. There isn't really a one-size-fits-all answer to these things. So we needed to understand the patterns before we could start solving them.

It wasn't a big formal project, it was about being genuinely honest with ourselves about where the friction really was, rather than assuming we knew what the problems were. And that study told us where AI could genuinely help, and equally, where it couldn't. What we were then able to do is figure out where a human touch was still essential. And that's shaped every decision that we've made since. So if I had one piece of advice for anyone in this session who's thinking about where to start with AI in their hiring processes. Do that exercise first. Even just a week of consciously tracking with your time, guys. It sounds simple, but it changes what you prioritize, and it stops you from investing in solutions to problems that you don't really have.

So what we have for you guys today is two examples. These are projects that we've done at Scede with our clients, and essentially been able to reduce a lot of the drag. A lot of the time waste in process, or Essentially fighting against the grain that we're seeing in procurement at the moment, which is, again, high numbers of applications, no extra hours in the day. But how can we make sure that we're still delivering an excellent service and hiring success for our clients? And if you guys are internal yourselves, how we can deliver exceptional service to our stakeholders. So, the first example that we have is cutting screening time in half. So, what does that look like in practice?

The first one we have is with our customer, where CV screening had become a serious bottleneck. They were a growing tech business, hiring across multiple functions simultaneously. And the volume of applications coming in was significant. The recruiting team was spending a huge chunk of their week essentially, we're talking the majority of their time, in some cases, just getting through CVs. So, before a single conversation happened with a candidate, before any real recruitment work was beginning. The hour's already gone. It's a problem we see a lot, it's a problem we hear a lot on LinkedIn, and the top of funnel becomes this enormous time sink, and everything downstream suffers as a result. Short lists take longer, hiring managers get frustrated. And as the rule, as long as time has been, as good candidates lose interest because the process feels slow.

So what we introduced for this client was AI-assisted CV reviews, and these were able, essentially, to enable the AI to do the first pass. So this is flagging relevant experience, surfacing patterns across applications, cutting through the noise, essentially to help our recruiters get to the candidates worth looking at more quickly. And the result? Screening time was cut in half. Half the hours at the top of funnel, which meant faster shortlist, a better experience for hiring managers, and quite frankly, a better quality recruiting work further down the process, because the team had the sped headspace for it.

And now, this part is really important, and this is something we've all heard a lot about on LinkedIn, or if any of you guys go into the pits of Recruitment Reddit. AI was not making decisions. And I want to be very clear about that, because it matters enormously to how we operate, and how we operate with our customers, and how you guys operate with your stakeholders. AI still makes a significant number of mistakes. Around 56% of firms worry that AI may inadvertently screen out qualified candidates, and it's a legitimate concern. It can be too rigid in how it interprets experience, it can misread unconventional CV formats, and it can miss the kind of contextual nuance that an experienced recruiter like us picks up on instantly. This could be such as a career changer with transferable skills, someone whose title doesn't reflect their actual seniority. AI can struggle with that.

In fact, a number of companies use AI exclusively for rejections at initial screening. This means candidates never have a human look at their profile. even before it's been eliminated. And that's not how we operate, and then, frankly, that's what we think is a problem, both ethically and practically, with AI usage at the moment. So our approach is the opposite. AI helps us tier candidates to get our recruiters into the strongest profiles faster than they could do manually. It's never declining someone, or making a functional judgment on a profile. Every single decision still sits with a human. AI just means that human gets there quicker, with less noise. And less admin time reading through hundreds of CVs, hundreds of cover letters. think of it less of AI screening people out, and more as AI is helping us find the right pile to look at first. And then, our recruiters take it from there.

The second example we have is cutting interview stages from 6 to 3. So, this second example is quite different. There wasn't primarily a top-of-funnel problem. This was about process design further down. The customer, in this particular example, had a six-stage entry process. And look, I understand how these things happen. Stages get added over time, different stakeholders want their moments in the process, and before long, you've got a pipeline that is more about managing internal politics than genuinely assessing candidates.

And also, from a candidate perspective, 6 stages is exhausting. We've all been through long hiring processes and know how quickly enthusiasm turns into frustration, both from personal experience and also from speaking to our candidates. From a hiring perspective, like, It creates drag. Great candidates drop out. Timeline stretch. And the business ends up losing people it really wanted.

The thing that really changed here was giving the team visibility into their own process that they simply didn't have before. So what we did at Scede is we built a system that ingests all of the scorecards, intake notes, and job descriptions from across the entire hiring process, and analyses how consistently the team is actually screening and interviewing against the criteria that was set up front when the role was kicked off. And what surfaces from that is really revealing. It shows where there's drift in the process, where interviewers are going off brief. And essentially assessing things that weren't part of the original criteria. And it also quickly discovered where two stages are essentially asking the same questions and duplicating effort.

What we then found out is where the process had gradually lost alignment with what the hiring manager actually agreed at the intake state. A key point on this one is it isn't about catching anybody out. Interviewers aren't doing this deliberately. It just happens naturally over time, and we've all seen that. But the impact is significant. When different interviewers are assessing against different criteria, you end up with conflicting feedback. more stages. Try and get clarity, and a process that takes far longer than it needs to. What the bot gave us, and our customer. Was a clear, evidence-based picture of where stages were genuinely adding value, and where they weren't.

leading something I refer to as the MVA. So the MVA is the Minimal Viable Assessment. Essentially, what is the minimum set of stages you actually need to make a confident hiring decision? This, combined with better upfront AI screening, that meant we were working with higher confidence candidates before interviews even began. We were able to strip back the process from 6 stages to 3. The same quality of hire, half the process. And the candidate experience improved dramatically as a result. They weren't waiting weeks and weeks for their sixth, fifth, fourth interview. We were able to act quickly and get the people in that were necessary.

The broader point here is that six stages usually exists. Because a hiring team doesn't fully trust its pipeline. The more confidence you can build through better screening, cleaner criteria, and consistent interviewing, the less you need to compensate with extra stages. And that's what AI enabled in this case. So what we're gonna do now is… this is the lens that Scede has brought through. This is grounded in what's actually working with real hiring teams right now. And what I find fascinating is that Teamtailor's platform data tells a very, very similar story, but on a much, much bigger scale. So I'm going to hand over to Adam, and Adam will go through the next few bits.

Thanks, Nathan. So yeah, just to give you a bit of context to Teamtailor, very much how the system is being built is with two key pillars. It's user experience, both from internal, but then also from a candidate experience as well. So everything that Teamtailor does, it has those two elements, as its core. And now with AI, we've been one of the very early adopters of AI within our platform. Now, these data points you'll see here, we've done a full analysis of all of our clients and pulled that into one place, just so we can get some context to actually what's happening.

Now, if we start to look at, how our co-pilot is working in the system and some of the impacts it's having. Our co-pilot is designed, as Nathan kind of alluded to, how they look at it, very similarly, is there to aid the recruiter. In the tasks that you do day-to-day, to speed those tasks up, to make them easier for you, but also to give you detail, and context that you can then use. So, everything from, sort of, criteria screenings, so skills and traits, it can help you build out job descriptions, etc. There's a load of, co-pilot, ability within Teamtailor, really to take away those day-to-day, admin tasks. Never to make decisions. It's all about how do you get to those candidates quickly.

And so you can see, you know, we've looked at 1.1 million applications, within our system, and some of the key criteria that we're looking at now, you know, anyone that is using our co-pilot from a screening perspective, on average, has a reduction in time to progress candidates by 31%. That's a huge, huge, time-saving in itself. When we then start to look further through the process, and looking at how quickly we're hiring, again, coming back to Nathan's point, we're all seeing more vacancies, we're seeing more applications.

actually clients now that are using Copilot before and after, the differences we're seeing in time to hire has, dramatically decreased by 21%. Again, another huge time saving. And this is from before using Copilot and then using Copilot now, so the difference is clear. Copilot is having a huge positive impact on some of the key metrics within talent acquisition teams.

One of our really interesting, Copilot capabilities is our criteria screening. So we have skills and traits that you can then interpret to a specific role using the job description. Companies that are now using that ability are actually seeing candidates being progressed through, a much higher velocity. So, they're able to get to the higher quality candidates a lot quicker. And then able to move them through the process, a lot clearer as well. Again, we've looked at 6.1 million applications in our enterprise, client base to pull this data. So this isn't just an isolated client or two. This is widespread. Part of my job is to work with clients to optimise the use of Copilot. And so you can see, we're really clear on, how, using Copilot, using the system correctly, is really having a huge impact on TA teams. Cool. That's me.

Awesome. Thank you, Adam. We've got about 10 minutes or so left for questions, and what I'd really love to do is make this as useful as possible for everyone on the call. So, if anyone's got any questions, please drop them into the Q&A section. You can continue to see that as we go through this Q&A part here towards the end, but I'll start going through some of them. I'm gonna option Adam to jump in with his perspective as well, as he comes from TA himself.

I can see from the first questions, how do you go about deciding whether AI… whether an AI use case is worth pursuing versus just adding more to a process? I think AI, from my perspective, AI is there to minimise What is in the process, or what you have to do as a person. I would probably say… Again, with, like, a time and motion study, you're able to see what the actual manual ad is from a recruiter. time perspective, I would probably build that around a prioritisation list. We said, as we said during the talk today, we typically find this at top of funnel, so it's usually a good place to look, but every… one thing we learn at Scede is every company is different. We have companies where top of funnel isn't an issue. These may be smaller businesses with less EVP, so they don't get the number of applications, so that may be more about making the most out of the candidates in the process, similar to sort of the second example that I gave, is how can we actually improve our processes with AI and make it so recruiters can spend more time doing that.

If you're a larger company that's getting 4, 5, 600 applications, every single day, as we're seeing a lot of a lot of our customers at the moment, it's probably gonna be more top of funnel stuff. But you should always have measurable things that you're looking for. Don't deploy AI and assume AI is making the difference that you want. You need to know if there's a measurable impact on that. If you're using an external tool. the vast majority of this will come of analytics or reporting, so you can actually see what either the time save is, you can see what the impact of the AI is doing. That would be my suggestion. How about yourself, Adam?

Yeah, I would say just be very clear about what you're… what the problem is you're looking to solve, and… outline that problem really clearly, and not always AI is going to be the right answer. A lot of companies are on the hype train, almost, and want AI because of AI. I think going back to what Nathan was saying right at the beginning, really deep dive into what your challenges are and what your goals are. And then find the right mechanism to… to deal with that challenge. And I think, yes, AI is great, and it has a lot of benefits, but it's not always the answer.

Right, exactly. We've got another question, which is, how would you make sure AI is improving recruiter judgment rather than weakening it? I can give you a short or a long answer, I think, because we've got time. I'll give a long answer. Unfortunately, Adam heard this during an event last week, A few years ago, there was a study by the Washington Study, Where they had, I believe off the top of my head, 528 recruiters were given access to AI to assess candidates, and this isn't similar to what we were talking about today, where AI is just matching skills and prioritising. These were actually assessing candidates through and through, and the system itself had the ability to, like, recommend declines, so this isn't what we're suggesting here at Scede. It's quite the opposite. What we're probably saying is… Not a good, solution for our industry as a whole.

What this study showed is when you use these sort of intrinsic, actual, full analytics tools. is when there was no bias, recruiters were operating with a fine. When the AI had a mid-level of bias, recruiters… about 90% of recruiters were able to pick up on the bias that the AI was doing, and they could correct it in real time, and they would write notes saying, no, this is incorrect, this is wrong. But the study itself then showed if they put the system with 100% bias towards certain things, the recruiters actually ended up agreeing with the AI. And these aren't junior recruiters, these are very, very senior recruiters. But because of the way the AI was proposing the information, as you found, recruiters were actually

putting in the notes, justifying what the AI had done. So… it's mostly around what you are doing with AI, right? If you're using a tool like that, which is assessing and declining, it's not… it's what exactly candidates are saying is happening in our industry, and I'm not saying it's not. I feel for candidates to be saying it as vocally as they are, there probably is some truth behind it, right? But that isn't what we would advocate in terms of AI automation within your process.

hopefully that answers the question, but look, if you want to make sure that AI is improving recruiter judgment rather than weakening it, my personal advice would… I would only do it on hard skills, right? Don't do it on anything that can be interpreted in a way of the AI assessing something. You said before, it's very bad with certain formats, it's very bad with noticing if people have transferable skills. If you know your role has to have C-sharp or a subset of C-sharp, you need to put that into the system and say, I am looking for this, and this is the other alternatives that are good, because this is within the remit of this role, it's a hard expectation. Then. it would still need to be reviewed by a human before any action is made, but you should, from that, find better success on the AI's assessment, rather than letting the AI go wild and do what it wants, right? That's definitely not what we're advocating for. Adam, anything you want to add?

You know, I think the only thing with that really is, Actually, no, I haven't. I think you've covered it all, if I'm honest. Yeah, I think that's everything.

Not very often I cover everything, cool.

I've got a question here from Peter. While CVs are ranked by AI, Candidates at the bottom of the list are practically ruled out by AI. How do you square this with European legislation which restricts this practice? It's a really important question. It's one we definitely take seriously. I've kind of touched upon a little bit of question before. It's both an ethical and a legal-based question. So, there's definitely real tension in this, and the legislation is definitely catching up fast. The talk I did last week in Manchester was about the three massive legislation changes in the last 60 days that recruiters needed to know about. The EU AI Act now is…

strictly classifying AI-powered recruiting tools. So, previously, in the past, our industry's kind of been unregulated. We've been quite notoriously known for being an unregulated industry. That is definitely not the case with stuff going through from both the EU and also from the UK and the ICO now. Recruitment's definitely under the lens. I'm not gonna say that's a bad thing either, I think it's a good thing. Our industry has a very mixed name, and… Globally, right, in recruitment as a whole.

But specifically, like, the EU AI Act classifies AI power sort of agreement tools, and that includes, like, CV screening and candidate ranking as high-risk systems. So, these things have full compliance obligations, from August of this year. off the top of my head, I believe it's quite a large fine, or 7% of global annual turnover, so this isn't, like, a theoretical concern, and I think Pete has made a very good point here, it's a live regulatory issue for teams hiring in Europe, especially. And our approach definitely is designed for this in mind, so that the critical… the critical sort of distinction under legislation is that candidates have the right not to be subject solely to automated decisions. So this would be exactly with

I believe it's also under, like, significant effects, so an AI from August of this year, I think it's been rolled back to next year now, they wouldn't be able to be… have an entirely automated rejection. That's the distinction, right? If you are using a system, for example, like Teamtailor, as Adam's mentioned already. the AI makes no actual judgment, it makes no impact onto it. A recruiter will still need to actually review the profile. We said before, AI is not perfect, no AI is perfect. Even the most advanced system we use for the new version of Claude is not perfect. So, the distinction behind that, and I think what tallies quite nicely to what's coming through, especially with GDPR and coming through from the EU Act.

AI Act, sorry, is that no decision itself is being made. There is no significant effect from it. It still needs to be reviewed by a human. Me, personally, I'm a humongous advocate for wildcard candidates. Back when I was on the tools, I would probably say I place more wildcard candidates than not. And that's where the human approach comes through. There still needs to be a human approach. But what we do is we use AI to tier and prioritise candidates to help our recruiters, but the AIs only informing the recruiter's view, not replacing it. That's… how it kind of goes along with the legislation. Adam, anything to add?

Yeah, and I think, you know, especially when we're looking at the world of CVs now, that's the world of scrutiny, where, you know, AI is creating all of these very polished CVs. It's very hard to work through what's a good CV and a bad CV, or a good candidate and a bad candidate based on a CV. So. Having tools that enable you to screen a candidate. a little bit more thoroughly, gives you that additional bit more data, you can then go and do your job. I've said for a while now that I think AI is actually going to enable

or actually drive recruiters now back to what we used to do a lot more, which is have really meaningful conversations with candidates. It's going to give us a deeper insight to candidates that we can then go and question. So actually, I think the shift we're gonna start to see, with the use of AI, is more insight to candidates, but then as recruiters, we've got to become a lot better at interviewing, at digging out, as Nathan said, the wild cards. We've got to, you know, liaise with our hiring managers to make sure that we're looking and very clear on what the skills are required. So I think, you know. AI's got a great place, and it's gonna give us some really interesting insight. It's definitely not gonna be allowed to make decisions for us, which I think is correct, but I think it's gonna give us a lot of insight, a lot of help as recruiters, but we will have to adjust.

Yeah, no, completely. We've got another question, In your opinion, what signals tell you a hiring process has too many stages and what needs to be simplified? How long is a piece of string? Might be the short answer to that. it kind of goes back to what I mentioned earlier, so there's… a process that I go under, it's not a Scede thing, it's actually a risk management terminology. We have a good thing about using different acronyms for different industries here at Scede, which is really interesting, but it's called Minimal Viable Assessment. it's a very hard thing to find out manually. I would say in the past, if you go back a few years, it involved sitting on interviews, as many as you possibly could, throughout the whole process.

had the risk of making candidates a bit uncomfortable, why is the recruiter on it? With that, you were able to find out about duplication, you were able to find out about are things being fairly assessed? It was a very manual process, and if we're in the mindset, which we are at the moment, which is reducing admin time from recruiters. this is where you can kind of use AI. If you're a business that uses… I'll bring it back to Teamtailor, Teamtailor Video and uses their co-pilot function. You get the transcripts from the call, you can run those transcripts for a bot you've built yourself. It isn't something I would actually recommend you buy a product for if you guys have Google and Gemini, or if you have ChatGPT, or if you're…

You've got a very nice Claude. account, you can build something very easy yourself to go look through this systematically and find out any forms of duplication. Then, what you're able to do then is go back to the hiring team as a whole and say, look, we've got a number of duplications here, we need to pull it down. But if you're looking at a 6-stage process, an hour of pot, that's 6 hours of shadowing, that is… two hours of looking through your own personal handwritten notes, and then bringing that back to the business. What you're able to do with AI is nominally quicker than that. You're probably looking at 20 minutes, 30 minutes admin to make the same human-led impact. But processes need to be assessed beforehand. Adam.

Yeah, I think the only thing with that as well is, is, I think we've become… a little bit unsure, as a recruiting population, especially from the hiring manager population. So, training, is a big element at the moment, I think, that needs to be looked at. So, is there a reason why you've got all the stages? What's caused that increase? Is it a lack of confidence in… what you're seeing coming through in the ability to interview. So I do think, you know, looking at your stage, you've really got to do a deep dive into what's going on internally throughout that whole process.

Amazing. how do you make sure speed gains do not come at the cost of missing strong but less obvious candidates? I think this kind of goes back to what we were discussing earlier, Humans still need to be involved in that, models need to be adjusted. I don't think anything is ever perfect in recruitment, and I don't think anything is ever perfect in AI. I think if we go at anything with a perfectionist point of view of, we don't need to do anything more, we don't advance as an industry, and we don't advance as a team within our clients. it kind of goes back as well, wildcard candidates have always been a secret power of me back when I was on the tools, shall we say, where being able to have a conversation with a person and find out more from them. I think… you can get the speed gains with AI to get to at least the most immediate

Like, high-value candidates quickly. But I think… from using AI the way that we've already mentioned today, which is having that human approach to it, not letting AI do any declines. Obviously, that's illegal coming on soon anyway. You wouldn't and shouldn't miss out from that anyway. Yeah, that's my answer. Adam?

Yeah, I mean, I think, you know, quality of hire, it's that, that metric for me is the one which we really need to keep an eye on. You know, candidate experience at the moment is probably… at an all-time low, I would say. and there's a number of different reasons for it, but tracking that candidate experience throughout the process, I think, will give you a telltale to whether too much speed is a negative, or a positive. So, really keeping an eye on that statistic and that metric, for me, would be crucial.

Yeah. No, I couldn't agree more. In your opinion, is there a healthy way to be transparent with candidates who enter the process that their fate is and will be determined by AI? I would hope that their fate wouldn't and wouldn't be ever be determined by AI. I would be my initial answer to that. There are rules that… we've seen it in Canada recently, over the last 6 months, where they've now made it illegal for AIs to be used in a process, If it's not being disclosed to candidates. Teamtailor itself, when you have a recruitment screen through the system, notifies the candidate that there is AI involved in it. It's actually a mandatory post, it can't be removed, because there's obviously so much AI in the system. There are companies that have done really, really good examples of this and put it onto their career website. There's two that typically come to mind that would be good for everyone to have a little look at. Monzo did a paper, which discussed their use of AI, and also what they expect from candidates, because it is a two-way street. We're seeing a massive influx of AI-generated candidates coming through at the moment. It's becoming, especially within

financial services, and… anything to do with crypto or AI, it's becoming a massive wave issue at the moment. What… I would say is work with comms internally, or marketing internally, and build out similar things similar to what Monzo has done. You can find it quite easy online. A client of Scede's, the next client of Scede's, MaidTech, I worked with them myself on building out their one, which was just released recently, where they… they… publicly and clearly go through their commitment for the use of AI within the recruitment process, but they also set the guidelines of what's acceptable use by candidates as well, and I think that's something that's really vital. It can't be, we use AI this way, but you can do it the way that you want to. There is a big mismatch if you're just coming out of the approach of, this is what we're doing. But that is actually a document that's sent over every time that a candidate applies to their website. It's sent to them in their application acknowledgement email, and it's also on the front page of the… each company's, like, recruitment career pages as well. As long as it is mentioned to candidates, I think that's really…

really important, but I would hope as an industry, we definitely move away from anything to do with their fate is or will be determined by AI. I don't think that is where we are seeing AI go, especially with the legislation coming through as well. It's thankfully removing the ability for AI to do that. Anything you want to add, Adam?

No, I think you're right, and I think, you know. we're going through a huge adjustment here, and I think when you're talking about trust, and it's candidates' trust in applying in the role, it's companies' trust in candidates, transparency is key, you know, how are you using it? How are you allowed to use it as a candidate? Just answering those two questions and being really transparent for me is how you build trust. You know, this is very new, for everyone. You know, 3 years ago, it didn't really… wasn't really a huge topic. Now it's everywhere, you know, so we've got to start building more and more trust around it, so the way you do that is transparency.

Yeah, I'll add one more thing. Transparency is… astronomically important in what we do, and especially with an age of so much automation, and this is automation we actually promote, it is definitely removing admin time away from Our recruiters is important, especially as teams are flat. Roles are getting… role numbers are getting higher and higher. But transparency definitely is key. I think it's companies that will suffer are the ones that aren't being transparent where it is used. And they'll be the ones that we start to hear about on LinkedIn a little bit more so than the ones using it for… different purposes when the laws come through themselves. We've got one more question from Julie. How can we ensure the confidentiality and protection of candidates' data when using AI in recruiting processes?

it's also quite a big question. It depends also on what systems that you use. A lot of the stuff that we actually use here at Scede, actually the vast, vast majority of stuff that we use here at Scede is, systems that we have built internally. So, if you're a company that has, like, a Google account, for example, and you have Gemini, all your data is stored within your own internal systems, it doesn't get shared publicly. I will say that ChatGPT does not do that, so ChatGPT, anything that you put into there or build with ChatGPT gets shared with ChatGPT as part of their contract as a business. If you have Gemini, it'll be within your own ecosystem. Same again for Claude as well. Claude will keep your information private. If you're using a vendor, or using a software, an external software partner.

It is something I think will start to be, if it's not already, I think it should be already, but it is something that's going to become very, very, very important to, like, vendor conversations moving forwards. We've seen a shift from pre-GDPR to post-GDPR of. Where is our data being stored? So, prior to GDPR, anywhere, no one really seemed to mind. Post-GDPR, there's a big push towards it being based in the EU, for obvious reasons, with the way that we're protecting data. That's going to be very similar to… for AI, is what are you doing with the data that we're using in the system? Is it stuck within a closed system? Is it stuck within our systems? Is it going out to you and your cells?

But I think the best way to ensure it is to have those conversations with external partners. it has to have that conversation from them. A good question to typically ask during a conversation with an AI vendor is, if we were to audit you, would you pass? It's quite an upfront question, and one they typically wouldn't be expected, but from the answer from that, if they're coming back saying, yep, this is exactly how we do it, and this is our process for data protection, this is our process for that, it's typically a very strong sign, that they actually know the legislation's coming through, and know what they can and cannot do with candidate information.

Especially within the frames of GDPR. If they get uncomfortable with those type of questions, it'll probably be something you need to investigate a little bit more. But yeah, anything external, you need to have those conversations. You have a duty of care as the holders of the data and the transporters of the data to make sure that it's not being used for any alternative means. If it's something that you're building, it'd be very much based on what sort of LLM or system that you are using, and checking with those to make sure it's within your own ecosystem. Anything on top of that, Adam?

Yeah, I would just say, Education of your teams, that they're not going outside of those guidelines. you know, it has happened, and it is happening where recruiters are using their own personal ChatGPT accounts, or whatever it might be, and putting in personal details, that's a breach straight away. So, really understanding and educating your teams around where you can and can't use AI and how you use it, for me, is the biggest, biggest thing you can do around safety, especially within the processes.

I think that's actually also a really good answer. We discussed earlier already about looking… if you want to look at where to save time, start logging on the time, but if you have a wider team, and you have a team that's also tech-curious, which hopefully you do, especially with the way that the industry and recruitment's going as a whole. is probably having those conversations with your team and finding out what tools are actually being used by people. I think a lot of people would actually be very, very, very surprised, that maybe more stuff is being used as As unknown if you work in a company with access to certain blocks on your browser.

maybe something's fall through the cracks, and they may think, oh, I can use ChatGPT because it's not blocked on my computer. it could be a simple conversation like that, and you would have to actually have a conversation and find out, no, you can't, because we don't know where the data goes. So it would be a wider conversation with the people that you manage. maybe listing out the tools that you're using. You may even learn from it. Maybe someone's using a tool that's perfect for you, but you just haven't discussed that. That could also be an outcome from it. But probably having a full tech map

what people are using throughout their day-to-day basis, and what they're using with candidates, and what they're using with hiring managers, and then assessing it from there, and probably having wider conversations, again, with vendors, or more internal looks at the systems you're using in-house. There we go. So, I don't think I see any more questions, but…

What we'll do is we'll finish up the session now today with, just a general thank you to everyone that's managed to join us today. It's been a really great conversation. Some of the… all the questions that have come through have been absolutely fantastic. They've all made me think, and those are the best kind of questions. We want to thank you all again for joining us. If anything came up that you guys want… anyone attending today wants to dig into further, obviously you have the ability to reach out to myself or Adam directly. We're both always extremely happy to talk about TA and what we do. We've both been in this for very, very long. We wouldn't be if we didn't enjoy it.

I want to say thank you again for the final time. Have a great rest of your Wednesday, everyone, and if you need anything, please feel free to reach out anytime.

Thank you, guys. Appreciate it.