Re_Boot: AI in Recruitment
Helping solo and boutique recruiters reboot their relationship with AI - not through expensive enterprise solutions, but through practical, human-centered approaches that actually work for real recruitment businesses.
Hosted by Kat Kingshott and Nick Rickards, and AI educational morsels from Marley!
Re_Boot: AI in Recruitment
LinkedIn Wants to Pay You $70/Hour - To Replace You
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Episode 5 | LinkedIn Wants to Pay You $70/Hour — To Replace You
In this episode, Kat and Nick (husky voice and all) dig into one of the more eyebrow-raising stories in recruitment tech right now - LinkedIn quietly posting roles to hire top tech recruiters as paid consultants at $70/hour, tasked with training their AI models.
The irony isn't lost on either of them. LinkedIn has been charging recruiters significant fees for years to use its platform, and is now turning to those same recruiters to fix what their AI still can't do: understand how recruiters actually think, work, and influence.
Nick's take? It's an admission. Nobody he's spoken to thinks LinkedIn's AI recruiting tools are genuinely good. The agents pull back irrelevant candidates, don't ask the right questions, and fundamentally misunderstand what recruitment value actually is. Because here's the thing... recruiters don't just find candidates. They influence them. The nuance, the emotional intelligence, the reading between the lines... that's where the fee is earned, and that's exactly what AI can't replicate. Yet.
Kat goes a step further, reading the LinkedIn move as something bigger: a play to eventually build an AI marketplace that cuts recruiters out entirely: employers on one side, candidates on the other, agents doing the matching in the middle. Nick's not convinced it'll work at the senior end. But for high-volume, transactional hiring? They both agree the shift is already happening.
The episode also covers some practical ground with Nick flagging Claude's native connectors as one of the most underrated quick wins available right now (the little + button in Claude that lets you hook up Gmail, Slack and more in a few clicks). And there's a timely warning about tools like Claude Cowork and Perplexity Personal. They are genuinely powerful, but dangerous if you ask them to do too much in one go. Context windows fill up fast, mistakes compound, and suddenly you've got garbage data you're manually fixing.
The analogy of the episode: treat AI like a trainee. Give it 20 things, not 120. Feed it small, specific, relevant information. And remember, just like a new starter who doesn't want to let you down, it'll confidently make things up before it admits it doesn't know.
Context is king. (Yes, it's officially the show's unofficial second tagline now. 👑)
Like our page and connect with us!
LinkedIn: linkedin.com/company/re-boot-ai-and-recruitment/
Nick: https://www.linkedin.com/in/ai-agents-for-recruitment/
Kat: https://www.linkedin.com/in/katkingshott/
You're good. You always want me to start, don't you? Not this time. No. All right. Good morning, everybody, or afternoon, or whatever time of day it is. It's your two favourite hosts.
SPEAKER_00One slightly clearer than the other today, Nick. How how has your voice been?
SPEAKER_01It's um it's a rusky, sexy voice, isn't it? But uh I haven't been able to talk very well at all, actually. I had a bunch of uh client meetings a while ago, like a few days ago, and they literally couldn't hear anything I was saying, so the microphone was literally in my mouth, and I was shouting, and it was still coming up as a crazy weird whisper. So um it's a lot better. So thank you for your concern and your question. What is the topic of the conversation today?
SPEAKER_00I wanted to talk about LinkedIn. They have just posted an opportunity for recruiters to be paid$70 an hour to give their experience, their insights as a recruiting paid consultant in metrics analysis. What is that, you might ask? They are seeking top tech recruiters and corporate recruiting generalists to provide expert insight as part of a project for our AI marketplace.
SPEAKER_01So, what they're basically saying is we acknowledge that our AI agents are really shit. And we need to understand how recruiters work, think, and act. So we are going to interview loads of them, pay you$70 an hour to try to improve our models.
SPEAKER_00Yeah, I think LinkToo was founded in 2008, 2007. So it makes sense that they haven't quite figured it out yet.
SPEAKER_01It does, yeah. That's hilarious. That's so good.
SPEAKER_00And they have such a go-to list of people that they would already be asking.
SPEAKER_01Yeah. You know what's funny though? Well, I suppose now if you think about it from their perspective, if they do that to five or six hundred people, then what LLMs are really good at is the consistency, spotting the consistencies. So they'll just get the consistency, consistent knowledge across all parties, um, and then they'll apply that to the model. But it's funny actually, because I remember about eight or nine months ago when um this guy was like the recruitment at the recruiter agent's gonna be amazing, and then it told me all this stuff, and I was like, there's absolutely no way it's gonna do that stuff effectively, unless they really understand how recruiters think and act. But I've not spoken to anybody who has genuinely said it's really good. So if you're listening to this and actually you are that person and it is really good, please do reach out to me and to cats because I would like to know what it is good about. But everybody says the same thing, which is it's just basically brings back loads of random people that aren't relevant, um, and it doesn't really ask you any questions or anything like that. So it makes complete sense that they're doing this. So it's ironic though, isn't it? It is funny, isn't it? They charge they charge users an absolute fortune to be on their platform, and then they're using the same people that they're charging to train the models to charge them more.
SPEAKER_00Or well, I I actually kind of read that more as a as a full replacement. So maybe that's me being um being extra a full replacement in what sense? What is in like I think if if you if you could build a really successful AI marketplace, I I I work with a marketplace very closely called uh Paraform, who preferred a lot of recruiters to work with them. They um are looking to work with recruiters within a marketplace. I read this is what LinkedIn um is doing, as essentially looking to circumvent the role of the recruiter and actually take that out entirely. So ultimately what's left on LinkedIn would be a bunch of employers and candidates and an AI marketplace run by these agents that have been fed all the stolen to the LLMs to essentially move folks around.
SPEAKER_01Nah.
SPEAKER_00That might be their plan, but that's not oh 100% would not happen, but I do I do I do think that that's what they're trying to do.
SPEAKER_01Well, and then so so so LinkedIn become the fee bearer. So if they're if their agents find yeah, I mean I can I I get it, I can see how and why. So they're basically I mean they do like jobs and stuff at the moment, anyway, don't they? But they they sell to the end client and the recruiter, don't they? So they're definitely gonna monetize the agents um and probably charge more than 90 quid a month, which is what they're doing at the moment, just to try to flog it out to anyone. Um so yeah, that makes it make sense. Cheers. Maybe we are all screwed. It's a good job, it's a good job. I jumped over the fence ages ago, so you guys are screwed, not me.
SPEAKER_00Look, I I think that that's what some of these businesses are are trying to do. Um Jack and Jill is one who flat out says that that they don't they don't want recruiters in their in their hiring process. That it's uh I think it's Jack might find the candidates and Jill talks to clients. I don't know. But so something either either one of those happens. Um and yeah, recruiters don't don't need to be there. So it's interesting the conversation.
SPEAKER_01But again, though, it's people who don't really understand recruitment who are doing this, which is even more crazy considering that LinkedIn should know everything, they should absolutely understand how recruiters think and act. This just goes down to the whole idea that recruitment is just a transactional thing, right?
SPEAKER_00Mm-hmm.
SPEAKER_01So there's definitely a place for a Jack and Jill. 100%. You know, if I'm a job seeker, would I go on Jack and Jill to find me jobs if I didn't know, you know, how to do what I do? Yeah, I probably would. But most companies are now using AI to filter out all the crappy short lists and all that stuff anyway, right? And that's one of the things that recruiters are for is to have conversations and understand nuances that AI could and never will be able to understand, you know, how we talk, the gaps, the emotional intelligence, that sort of stuff, right? So I think that this is just another really good example of how people are building these tools not really actually understanding recruitment, not understanding the actual value that we bring. We don't recruit recruiters don't find candidates. It's easy to find candidates, right? It's recruiters influence candidates to take the job or whatever else. That's where the value comes in, isn't it? It's not mixing A with B. So I really hope that LinkedIn do create this marketplace and do try to do this stuff because then when it fails, recruiters will be seen as even more value, which then means that they can increase their rates and fees.
SPEAKER_00100%. 100%. I think where it could work is perhaps those lower skill, uh, higher volume, uh some project some project-based recruitment. I could I I see I see the value in some of those in perhaps using like a LinkedIn style campaign.
SPEAKER_01Yeah. Like a construction, that sort of stuff.
SPEAKER_00Yeah, some some blue collar stuff. Yeah, and front, you know, front frontline support. The the there's definitely a ch a shift, a shift happening there. If you go and apply at McDonald's now, you don't actually need to talk to a quote, a human up until I think it is just signing the contractor, it's your first day, everything is kind of virtual and online. And there's a bunch of different reasons why why that actually does work quite well. But for seeing for senior searches, searches that have a price tag of 30,000, 50,000 plus dollars attached to it, I think that it's yeah, perhaps a little bit um I'm not even naive. It's absurd to think that that an AI marketplace can just fully uh fully do away with the recruiter side, but I was reading the actual role descriptions of of this and they're looking for an a software engineering um AI trainer as well. And yeah, I thought it was very interesting that they were going they were going for that. So if anyone has any insights, um, like you say, very interested to hear if it's if it's working so far.
SPEAKER_01Have you um have you come across any good new tools recently? Anything that's been like boom, these guys have nailed it. Maybe some of the users uh the the listeners here might benefit from if you haven't it doesn't matter.
SPEAKER_00No, no, no, no. There was there was one that was spoken about yesterday. Um called well it's a new note, it's just a new note-taker, so I I don't know how game changing that is, but it's called Plawed.
SPEAKER_01Called Claude.
SPEAKER_00No, called Claude, no. Like I applaud you. Claude with a pluh.
SPEAKER_01Oh, okay.
SPEAKER_00Um, so yeah, I've heard about a new note-taker recently.
SPEAKER_01I don't feel like we need another note taker. There's a lot of note takers. It is the best AI tool in the world though. Clawed? No, AI note takers.
SPEAKER_00Oh, yeah. I've got to be able to do it.
SPEAKER_01It's the one the one thing I couldn't live without.
SPEAKER_00There's one actually, and it basically, I need to find the actual name name of it, um, to really talk about it. But essentially it does build your it it uses it probably uses you know, Claude or your note, whatever your note taker is, um, to build your like your second brain, and then actually map the brain as well. So it essentially you could you can feed it any of the information that any of your LLMs have collected about you and everything that you do, and then essentially map how you think, which um for an eight well, particularly for neuro for neurodivergent ADHD types, that would be I think for anyone though, to be fair, it's pretty cool. Yeah, I'll find the name of that one.
SPEAKER_01What I would say though, what I would say though, is everyone's going mad about Claude co-work and they should, it's good. Um, it's basically MCPs for Claude, which is stuff that we were talking about like a year ago. It's they just rebranded it. Um, but it's very useful. And if you haven't used um not even co-work, I okay, I personally I don't use cowork, but I do use connectors sometimes on Claude. So if you don't do that, you should do that. Um, all you basically do is when you log into Claude, next to the query, the query chat, you'll see a little plus button. Click the plus button, and then it says connectors, and then you click connectors, and then it has a bunch of native connectors already. So, you know, things like your Gmail, all that sort of stuff. Like there's there's loads of them to be honest with you. And what you're doing is you're basically equipping Claude with tools, which it can then use. So when when I was doing it ages ago, I would have to like I had to manually like connect them all up, and they've simplified the whole process. So you literally just click a few buttons, and then Claude can read your emails or send emails or draft emails or whatever. So if you're not doing that, you should do it, and that can be like the$20 version. Um, it will completely rinse the context windows and whatever, but um, it's is it's super useful. So useful tools. Uh I would say the simplicity um of that is is is will mean that a lot of people with absolutely zero experience or knowledge or understanding will now be able to connect some of their basic tools and they will see a noticeable difference in their productivity for sure. So um that is that's that there's actually one other thing that I wanted to say, actually. Um, and actually, you know what, I'm gonna link in with the whole LinkedIn thing because it's the same principles. Um I've I've I've said for a long time that to really get a lot out of agents um or just AI, whoever's building it or the user, the the um there needs to be a solid understanding of how recruiters work, it just has to happen. That's why, again, you know, LinkedIn's not done so well, and they're now literally saying we haven't done well, we need to get loads of recruiters and train our models again, right? But when you um that there's a lot of stuff at the moment, Claude Cowork, Perplexity, personal computer, right? Um that there's there's tons of stuff where you go in there and you think it's gonna be some absolute mad magician thing, and you type in a bunch of basic stuff, and then just you know, magic appears because that's the way that they're they're marketing this. Um, and there's a client that I'm working with, and and to be fair on him, that's a wait a wee while for me because I've need to build all this stuff. But he he was using perplexity um personal, and he literally just gave it a quote, uh, you know, a prompt just saying, like, map out, map out this market in this state. Like, here is um some examples, here's a spreadsheet which gives you some examples. Find them, connect with Zoom info, connect with bullhorn, and and map out this whole market, right? Um, and initially it works really well, but then what happens 99.999% of the time is that cost goes through the roof, and also most crucially, context is lost, which then means it makes loads of mistakes, which then means it costs more because now it's trying to fix those mistakes, right? So just be really, really mindful if you're using a clawed cowork or perplexity or something like that, that you have to do things in short steps, get it to do 20 things, right? Max, don't get it to do 120 things, because it will absolutely slam you. When we went through the data in Bullhorn, yes, it had found a bunch of people, yes, it found a bunch of people on their companies, but it didn't provide any of the LinkedIn URLs, any domain names, it was complete garbage data, which now he has to go through and add manually, right? So just be just be really mindful. Everybody's jumping on these you know, open claw and all this sort of stuff, and then they are wicked tools, but use them in moderation when you're executing it. Don't just you know say do this for five hours. Do you know what I mean? Um, because they have no idea what you do, they don't have a clue what what how you think, even if you give it a little bit of a prompt for some information, because that gets swallowed up by the context windows very quickly anyway, and it just goes off and uses general knowledge to perform the action.
SPEAKER_00Yeah, it would be the same as if you had an employee starting on your first day. Maybe they've worked in a you know a recruitment business elsewhere for six months, but you say, Hey, go do all this stuff. Uh, I've given you my my my my what you need to be looking out for, and because it doesn't want to let you down. Because remember, these AIs, they they're like they have a bit of a praise kink, some of them. They're like that's a that's a good analogy.
SPEAKER_01Is uh let me just tweak it slightly. It's like you're getting a trainee, giving them a spreadsheet with 25 companies on there, and then saying, right, I want you to find the URLs for the 25, and then I want you to find me 400 more, it will do those 25 really well, it will probably do an extra 10 really well, and then it's like okay, I don't know where to go now, so let me just use some reference points. Do you see what I'm saying? So treat it like a trainee all the time, get to do small things, and the more information you can give it about you and how you work and all that stuff, the better it's gonna be. Context is king. So there you go.
SPEAKER_00In regards to tools, and I think we have we have the tagline now, by the way. Context is king.
SPEAKER_01Context is king. Yeah, we actually should have that. Although I do quite like the whole trying to figure it out together.
SPEAKER_00Well, also that. Well, maybe look, we'll we're building this up, we're doing we're we're we're more regular now.
SPEAKER_01So should we do like which we're figuring this out together and then hyphen step one, context is king. That's one of the things that we figured out. Right. Step two, whatever we figure out next.
SPEAKER_00There we go. That sounds that sounds good.
SPEAKER_01Yeah, cool. All right, well, yeah, it's been emotional. Until next time. Hopefully you'll be able to hear me a little bit better next time as well.
SPEAKER_00All right, see you guys.
SPEAKER_01All right, ciao ciao. I was gonna wave then