52 Issues Before ROI: Hung Lee on Building Brainfood and the Future of Recruiting

Show notes

In this episode of the Leaders in Talent podcast, host Adriaan Kolff interviews Hung Lee, founder and curator of Recruiting Brainfood, the weekly newsletter much of the recruiting industry reads every Sunday. Hung spent ten years as an agency recruiter, became the zero-to-50 hire for London tech start-ups in the early 2010s, and founded the talent marketplace Workshape. Recruiting Brainfood started as lead generation for Workshape. The first issue went to one subscriber, himself, and it became the business.

Hung talks about the quit points of the first year and the advice that got him through them (do it for a year before you look at ROI), the sponsor he tried to send away with a made-up price, why he runs the business alone and still writes every word by hand, and the restaurant hours that get it out every Sunday. Then the big argument: staying on closed, cloud-vended AI leads companies to headcount zero, the way the cloud wiped out in-house IT departments, while on-premise AI turns every hire into an investment that compounds the company's own intelligence. He explains why recruiters should back on-premise AI, why the future is recruiting for talent rather than tasks, and closes with three pieces of advice: invest in your network, raise your profile, and practise with AI every day.

Timecodes

01:23 Welcome: Hung Lee, founder of Recruiting Brainfood

02:56 From zero-to-50 recruiter to tech founder: Workshape

06:23 How a lead-gen newsletter became Recruiting Brainfood

08:04 Quit points: one subscriber, 52 issues and issue 30

13:20 The first sponsor and the go-away price

18:39 A one-person media business, by choice

20:51 Writing by hand in the age of AI, and what counts as brain food

25:16 Restaurant hours: how the newsletter ships every Sunday

28:13 What TA leaders ask about: AI-native tech stacks and context

30:43 Tower Bridge, headcount zero and the case for on-premise AI

43:16 The AI dopamine loop, AI labs hiring recruiters and the end of IT departments

49:28 Three pieces of advice and the 2026 events list

___________________________

Connect with us on LinkedIn: https://www.linkedin.com/company/matchr/

Get in touch with us: https://matchr.io/contact

___________________________

Connect with Hung Lee: https://www.linkedin.com/in/hunglee/

Connect with Adriaan Kolff: https://www.linkedin.com/in/adriaankolff/

___________________________

Resources mentioned in this episode:

Recruiting Brainfood, Hung's free weekly newsletter, and the crowdsourced Big List of Recruiting and HR Events to attend in 2026: https://www.recruitingbrainfood.com/

___________________________

RSS feed: https://media.rss.com/leaders-in-talent/feed.xml

Transcription

[00:01:23] Adriaan: Ladies and gentlemen, welcome to another episode of the Leaders in Talent podcast. Today, for me, feels a little bit like a full circle moment, because my guest is none other than Hung Lee, founder of Recruiting Brainfood. Hung, welcome to the podcast.

[00:01:41] Hung: Great to have you all. Great to come on the podcast, Adriaan. A real pleasure having a chat with you, and I'm looking forward to this next hour we have.

[00:01:49] Adriaan: Awesome. Hung, for the 1% of our listeners that do not know who you are, can you give us your two cents: who you are, what you do, and then people will also understand why I'm so excited to have you on the podcast.

[00:02:02] Hung: You know what? I basically just write a newsletter, Adriaan. I'm a recruiter at heart. I still self-identify as a recruiter. I started off as a recruitment agent and did 10 years of that, then I went in-house and stuff. So I've basically been in this business for a long while. The main thing I do these days is write a newsletter, do a few podcasts and get involved with community. I see myself as an ecosystem support person for the recruitment industry, via all of those media channels, basically.

[00:02:31] Adriaan: And you're front and center. Your weekly newsletter, Recruiting Brainfood, is read by thousands, hundreds of thousands of people on a weekly basis.

[00:02:39] Hung: Well, not quite 100,000, but basically, yeah. Quite a lot of people in the industry read it. It's free to read, by the way, so sign up if you want on recruitingbrainfood.com. It comes out every Sunday, and on Monday I write another newsletter as well, which you can follow on LinkedIn. So that's two newsletters every week.

[00:02:56] Adriaan: Wow, amazing. Well, we're definitely gonna dive into that. But what I feel not everyone knows is that before starting Recruiting Brainfood, you were actually a tech founder. Tell me a little bit more about the journey from in-house recruitment to becoming a tech founder, and how, organically, did Recruiting Brainfood come out of that?

[00:03:18] Hung: Yeah. I mean, this story is easy for me to tell, Adriaan, 'cause it's burned in my mind. There were a number of forks in the road where I could have made a lot more money elsewhere than down this road here. So basically, the story: I did a lot of in-house recruitment. I was the zero to 50 guy back in the tech boom here in London, so this is the early 2010s. I would be the person that would come in. There would only be the founder. He or she had raised like 5 million quid or whatever, and I'd be the guy that converts that into the founding team. I did that for a number of projects, maybe two years or so, actually three, four different big chunky projects. Through that you obviously end up meeting tons of engineers, right? You build a massive network of engineers, and some of them become your friends even though you don't want that to happen; they end up becoming your buddies. Before you know it, one of your friends says, "Hey, I got this idea. I think we should change how recruiting works." I said no. But then the next week he comes back with a prototype, and it was like, "Oh man, you might as well build the other half then." And then the week after he comes back with the other half, and I thought, "Okay, we've got this far, this may be something." The concept basically was a talent marketplace. We were in the same era as Hired.com, if you remember those. MoneyPot in the Netherlands was part of that milieu. Everyone was thinking, "How do we disintermediate recruitment agencies, and how do we get employers talking directly with engineers?" Our innovation was: can we somehow stop the first conversation, which is the sentiment conversation? Like, what is it you wanna do? Can we pick up the information about what the person wants to do, and can we communicate that to the employer such that we don't have to ask that question? Then you can go straight into a substantive call about why the two people might wanna work together. So that's Workshape. Brilliant idea, brilliant product, brilliant team, terrible CEO, Adriaan. That was me. Now I think, "Okay, I now know why this was always not gonna work." I'm a bad CEO of a product, 'cause it just required a bunch of different decisions and qualities that I don't think I had for that particular product. I was too emotionally invested in the concept, which you can't be when you're a proper CEO. You have to be much more ruthless in how you make your decisions, and I wasn't. Essentially what happened was that over time, the business had stopped iterating. We were slowing down, losing momentum, and my job was just to try and solve it all. "More sales solves all problems" was my mentality at that time. So I thought, "How do I make that pipeline as thick as possible?" My current process was okay, but it wasn't scaling. And I thought, "Okay, maybe I should set up a bunch of media channels, start communicating with my audience in a different way, which is less intrusive, more value-driven, see if I can create more of a community vibe and get them to be aware of the product I was running after having consumed some of that content." So it was basically an inbound marketing strategy, which is where Recruiting Brainfood started.

[00:06:23] Adriaan: Really? So take me to those early days, right? You're looking at your business, Workshape, the platform. It's not really going the way that you hoped it would go. So how did you think about it then? Was it immediately a newsletter? Was it creating content? How did that start?

[00:06:41] Hung: It was immediately a newsletter, and this is pre-newsletters being hyped. So one thing I had going for me was basically that I started it before everyone was doing it. I wanted to make it, firstly, vendor agnostic, so I'd never even talk about Workshape on it. I didn't even label it as a brand. I thought, you know what? It's just gotta come from me, just my observations. Now I know it's probably called founder-led marketing. You're not talking about your brand at all. You're just talking about what your observations are. I always read a lot, and part of my problem was obviously encountering so much crap as you sift through the internet. I ended up just identifying a number of channels which were consistently high quality. Before you know it, you have a bookmarked database of these sources. And I thought, "Okay, the quickest way I could do this is just create a curated newsletter: here are some really cool articles I've personally read this week, I reckon you should read them as well." That was generally my vibe. So no concept of growing it as anything other than a quick, easy way for me to communicate who I am to the market. It didn't cost me a lot of work, was my thinking. It externalized a lot of my thinking as well, which is a form of marketing. And then there was a hope that maybe people would breadcrumb their way back to the product.

[00:08:04] Adriaan: So in those early years, when did it start to really grow? Was that right from the get-go? Was it constantly a straight line, or was it for a very long time nobody was really reading it and then all of a sudden it kicked off? What was that journey?

[00:08:19] Hung: Dude, there were so many times where I nearly stopped the project. I call them quit points. You would know this as an entrepreneur, man.

[00:08:27] Adriaan: My goodness. Oh my goodness. Yes.

[00:08:30] Hung: You know immediately what I'm talking about. So quit point number one was the first newsletter, man. I sent the first one, and I was so incompetent as to even how to set up a newsletter that I sent it only to one person, which was myself.

[00:08:47] Adriaan: Great. 100% open rates. 100% open rates.

[00:08:50] Hung: 100% open rate. Literally one person opened it. It was actually me, and I thought, "What the...?" You know, I spent my Saturday afternoon doing this, and I could have done something else. So the quit point was really early. I thought, "What's the point?" But then I remembered a conversation I had with a friend of mine, and he said, "Look, Hung, if you're gonna do it, do it for a year before you even look at ROI." And I thought, "Do what? Do a year?" And he said, "Yeah, just do a year." And I thought, "Okay, 52 newsletters seems like a lot, but I think I could probably do that. I'm gonna pay attention to that one idea." That pulled me back from the brink of quitting, and I just said, "Oh, I've gotta do the second one." So I kept going. Issue number 10 was another quit point, man. I'd got to something like 33 subscribers by then, and I thought, "Wow." Every subscriber was amazing. But for some reason, I lost like three subscribers that weekend, and it was like, "Oh no, I've gone back to 29." I thought, "What am I doing?" I'm sending to 29 people. But then again, I remembered this conversation. I thought, "Okay, I'm gonna keep going, man. This is my personal mission. I don't care if it goes to zero. I'm gonna keep going." So anyway, I kept going. We talk about quit points, but the turning point was probably issue 30. So we're now half a year in plus, and I've got to the point where it's very consistent, it's very rhythmic. There's no longer any thought of stopping, so it's almost like you've habituated yourself to doing it. And for the first time, I saw on LinkedIn someone recommending the newsletter without me soliciting it or calling them out. I didn't even know the person, but they said, "Hey, this is great. Come and read it." I thought, "Wow, that's nuts." That gave me such encouragement, and there were lots of people signing up. And then from there, I just kept on plowing through. So yeah, the first year was full of quit points. And a message to anybody who's doing any content or any business, as you obviously know as an entrepreneur, Adriaan: the quit points are all there. You do need to have a serious conversation with yourself at some point. Don't quit too early. Don't quit too early. People fail 'cause they quit too early.

[00:10:55] Adriaan: It reminds me of two things. My very first executive coach taught me a very valuable lesson. He said, "You need to do something 30 times," he literally named that number, "before it can become a habit," right? Those 30 times, that's where the discipline comes in. That's the grind of doing it before it becomes as normal as brushing your teeth, before it's part of your system. So that's one. The second thing: MrBeast, right, the most watched, most well-known YouTube influencer in the world. He often gets asked, "Okay, I wanna be an influencer. What do I need to do?" And his advice every single time is, "Make 100 videos. With every video, you look at your previous video and you optimize one thing. And then when you've done 100 videos, you can come back to me, and then we can have a conversation, and I can help you grow." And two things happen. One, 99.9% of the people don't follow up, because they don't wanna do the work. Two, the 0.1% of the people that do the work do not come back, because they now know how to create video.

[00:12:00] Hung: Yeah. Those lessons, those stories, resonate so much because, number one, it's the frequency and the volume. This is why I've kind of turned into a bit of a volume merchant, or at least an advocate of doing lots of the thing, because you need the reps. You need the reps. It reminds me of Boris Becker. He didn't come out of nowhere. He literally said what he did was take a tennis racket, go to a wall and just hit it a thousand times. That's basically to get your muscle memory in such that it becomes automatic. And now I know that when it comes to doing content particularly, you've got to keep that cadence at the right place if you want to make change. Same with going to the gym, by the way, or some people journal or do meditation or whatever it is. You can't just suddenly decide to do it and think a miracle will happen. You have to systematize it such that you're consistently doing it and it becomes part of what you do, part of your life. And from there, the workload actually drops away, because it's integrated as part of your life. It's no longer an administrative task; it's just your normal workaday. And secondly, of course, the number of iterations you do means you get better. You just get better, faster, and everything becomes much more natural simply because of the reps. So, totally.

[00:13:20] Adriaan: Hey Hung, so newsletter number 30, right? All of a sudden you actually get that dopamine rush, like, hey, something seems to be working. But your main business was still Workshape at that time, right?

[00:13:30] Hung: Correct.

[00:13:31] Adriaan: So tell me about that journey. You were originally doing this to draw more leads to Workshape. Tell me about that moment when that all of a sudden shifted, when you realized that Workshape might not actually be the business that you need to focus on, but Recruiting Brainfood is. How was that transition? Was there a specific moment? Was there a catalyst? Take me back to that time.

[00:13:56] Hung: There was. I mean, there's a two-step process. Firstly, it was a lead gen thing, and it was never designed to be an independent business. But I remember after it got to about 5,000 subscribers or so, one guy came in and wanted to sponsor it. I'll never forget his name: Glenn Farrell, an Irish guy in Slovakia, for some reason doing a tech business in Slovakia. He said, "I wanna sponsor the newsletter." I didn't understand what he meant. I said, "Glenn, you're gonna have to dumb it down for me, man. What do you actually mean?" And he said, "Yeah, I wanna put my logo here. I wanna give you a load of copy. I wanna put it in here, and I'm gonna pay you to do that." And I said, "This is not how it works. This is a lead generation thing for me. I'm not gonna do leads for you. We can't do that." But he was also an ex-recruiter, so he was very determined. He kept on coming back at me. And in the end I just made up a number, a go-away number. You know, I picked a number out of the air: he's gonna go away on this. And he obviously didn't. He paid. He said, "Okay, I'll do it." I thought, "Oh no, I'm gonna have to do this sponsor thing now." So I inserted it into the newsletter, and basically, it went out. I thought, "Okay, that's it. He must be disappointed with whatever happened. He'll definitely go away now." But he came back the next week and said, "Hung, that was great. Let's do it again." And that's when I realized, oh, I probably underpriced myself. I'm gonna double it as the new go-away number. And in fact, he paid that as well. And that's when I thought, "Well, what is the number then? How do you make...?" So long story short, first year...

[00:15:30] Adriaan: So interesting. So how long into that journey was it when you got that first paid sponsor? Was it a year in, a year and a half, two years? When did that happen? Do you remember? About a year. Okay, about a year. So you'd been doing this for a year, and did you not think of putting Workshape front and center as the sponsor or driver? Did you ever think about that, putting it more front and center?

[00:15:54] Hung: Adriaan, I should have done it. Workshape may have been the thing now.

[00:15:58] Adriaan: Yeah, who knows? Who knows?

[00:16:01] Hung: It didn't occur to me. I made an ideological commitment for it to be an unbranded, unsponsored thing, and that was always what it was gonna be. And sometimes when I make those ideological positions, I quite like sticking to them. That's actually part of the reason why Workshape did fail in the end, because I wasn't able to pivot from those very strong early beliefs. So anyway, it never occurred to me. But what happened was, once that first sponsorship went in, suddenly I was inundated with companies that wanted to sponsor it, and then I realized they hadn't known that you could sponsor it. The first year I made enough to pay the rent. It wasn't big money. I was making 40K or something, 30, for the year. I thought, "That's nice," you know, nice to pay the rent maybe, but nothing where you could ever say, "That's what I'm doing." But then I started thinking, "Okay, what is the level I can get it to? Can I add another channel? Can I build another thing that can then increase the inventory?" That's when Recruiting Brainfood became alive. And probably around about that time, after the first year, maybe after the first six months of sponsorships had gone in, I started considering Brainfood as its own thing, as a revenue-generating option. That's when I realized that mentally I'd already moved in this direction, and I was kidding myself a little bit to say I was still running a tech platform, because in the end I was really running a media business and the tech platform was dying away. So yeah, we kinda made that self-serve, wound it down and kept going with the media.

[00:17:34] Adriaan: And how long into that journey was the moment when you closed down Workshape and really focused on Recruiting Brainfood? Was it a year and a half or two years in?

[00:17:42] Hung: It was probably about two years in. Workshape lingered on longer than it should have done. Again, it was the first tech platform, the first tech business that I ever ran, and I ran it really badly. I didn't know how to do it, basically, including how to shut it down. I should have given it a graceful exit. It deserved that at the very least. But nothing was graceful about the end of it. You know what I mean? How do you shut a business down? I did not know how to do it. We even had some late offers to buy it. There was a friend of mine who was running a job board and wanted to buy it, and I didn't know how to sell it. You know what? At that point, my mind had just switched off completely from it, and I wasn't able to recover it. So yeah, it wasn't a graceful end, but that's how it was. But at that time, Brainfood was on its way to where it was gonna go, and I was on the other ship, so to say. And off I went.

[00:18:39] Adriaan: Hey Hung, as far as I know, you never skip a beat, right? Every week you send out the newsletter, correct? 52 weeks per year.

[00:18:46] Hung: So far, yeah.

[00:18:47] Adriaan: Yeah. In terms of that tradition, you focusing on Recruiting Brainfood, has it always just been you? Do you have a team? Take us a little bit backstage: what does it take to send out a weekly newsletter that is the industry standard?

[00:19:04] Hung: There's no team. People are shocked by this. I often get emails from people saying, "Hey, can I speak to your sponsor team?" You're speaking to it.

[00:19:11] Adriaan: Let me connect you with him. Let me connect you with him.

[00:19:14] Hung: It's all me. And there won't be a team, 'cause for the same chaotic reasons that I was a bad CEO, I can't operate a business at scale. I can't operate this. The back office is a mess on Brainfood, for sure. But it's handleable, because ultimately it's a very simple business to run at the end of the day: keep doing the content, keep creating great stuff, the audience will grow as a result of that, and if the audience grows, then sponsors will come. That's generally how it's worked. I could have been more sophisticated, I could have made more money, I could have got more sponsors, could have got more expensive sponsors. That would probably involve me getting a salesperson, though. And one of the things I made a decision on, one of the axioms I decided on really early on, was that I wanted to try and run a business entirely on my own. I was a little bit thinking of the Tim Ferriss thing at the time: how far can one guy go? And since that early decision, it was never really a thought process to build anyone around me. I came close pre-COVID. Before the COVID crash happened, I thought, "You know what? I probably do need someone to handle the sponsors. Wouldn't it be great to get someone to build a YouTube channel or something for me?" So I started thinking about building maybe a three-person team. COVID happened and kind of saved me from a lot of payroll there, so it's like, "Great, that didn't happen." It was almost like maybe the fates were saying, "Keep going on the path that you were on." And now AI has arrived, and again, it's driving the idea of building a team further away, 'cause I'm thinking a lot of the basic stuff I could probably automate.

[00:20:51] Adriaan: Which is very interesting, right? So tell me a little bit more about your business: what are some of the things where you are using AI? Then we're obviously also gonna talk about what you see for our audience, and I'll share some of the conversations that I've been having in terms of AI. But you as a founder, right? A one-person company, which I really admire and applaud you for, because I know what it takes, and the pros and cons. Where are you using AI to take some of the workload off your plate?

[00:21:20] Hung: Not enough, is the truth. I don't use AI for any of the content generation. A lot of people immediately think the base use case of AI is writing stuff, right? Producing text. Not the case. I'm still writing everything by hand. I've actually made a point of doing that. The reason isn't purely because I think people prefer it, 'cause I don't think my prose is particularly good. The reason is because I wanna keep my mind working in this way. I've realized that for me, writing is not thinking alone, but writing is sharpening the thinking, and I've discovered things by writing them out that I wouldn't have discovered had I not been writing. So I wanna continue that aspect, and that is still like 80% of what it is. So AI has a limited use case here. Where I can definitely make use of it, and I'm on the way with doing that, is the CRM aspect. Obviously, tens of thousands of people are subscribed to the newsletter. I need to connect with them a lot better than I do. I need to build some sort of process which allows them to interact with the content that's being shared without them coming to me directly to do that. So there are many avenues where AI can be used, but broadly speaking, right now, as we speak today, I wouldn't say it's particularly prominent in how I run the business.

[00:22:44] Adriaan: So it's interesting, right? On the one hand, AI helps automate and can make a lot of processes more efficient. On the other hand, there's so much AI slop, right? Especially in the content world, oh my goodness. There's just so much noise. Because you read a lot, what are you seeing now in terms of trying to tell: is this AI-generated? Is this real? What's your take there, and what are you seeing happening on the content side?

[00:23:12] Hung: I'm not hostile to AI-generated content per se. I think there is a lot of AI slop, but there's also a lot of really good quality AI stuff. For instance, a lot of the YouTube consumption I do now is AI-generated. But it's not just AI-imagined. What it is, basically, is someone who's piped like 20 reports into NotebookLM and created an animated video with a voiceover, and it's actually a very good way to consume some dense information. So AI-generated is not necessarily bad. Where it is bad is when it's genuinely lazy and it's engagement farming, people who are trying to get something very cheaply with the use of AI. Of course you can't put that in. The curation side on Recruiting Brainfood has so far not had that problem, and the reason why is because the editorial policy has turned out to help me. The editorial policy has always been: can I find stuff that basically no one else can find? So you'll notice that it's usually never mainstream stuff. Occasionally a McKinsey report gets in there and what have you, but it's not the industry stuff that everyone reads already, because everyone reads it already. The purpose of Brainfood is to bring stuff you don't encounter. So I'm picking independent blogs. I'm picking out people who write outside of the domain, who have an idea that's adjacent to what we do. And a lot of that stuff I don't think is AI-generated, or even if it is, it's good enough quality for it to still go in. The rule of thumb is: look, it's called Brainfood for a reason. Is it nourishing for the brain? Does it cause you to think? I don't have to agree with it. Oftentimes people come back to me and say, "Oh, I can't believe you put that in there." It doesn't mean I agree with it, man. The reason I put it in there is that it caused me to react a little bit, and if it causes you to react a little bit, that's good. So yeah, it's all about stimulating the brain and giving people something to think about in the week ahead. That's how Brainfood works from a content point of view.

[00:25:16] Adriaan: Hey Hung, I'm picking up kind of two things. On the one hand, you self-proclaimed, "I was not a good CEO, I was just not great at that product." Also, in your current business, "I should be doing more," like it's not 100% in control. Yet you've been able to grow the most successful recruiting newsletter in our industry, and you haven't missed a beat. What does your process look like for getting that newsletter out on time every single week, with the highest quality of content in there? What does that look like, and what are some of the habits that you've taught yourself to make sure it always goes out, no matter what?

[00:25:51] Hung: So basically I work restaurant hours, Adriaan. People don't understand, but Saturdays and Sundays are work days for me, and my weekends have shifted to Monday and Tuesday. That's one of the things, 'cause it comes out on Sunday, and the next newsletter comes out Monday, so it's really about that timing. And the reason why it's those days is because I need the bandwidth, the unbroken time, to actually do the writing and reading. Number one, the curation side is easier than people think. I'm not searching for information. What I do is sit in the middle of where information flows, meaning I'm in a bunch of communities, I'm in a load of streams. Stuff gets shared all the time, and I just pick 'em off during the week and file them away for reading later. When I do read it, it's typically on the weekend: Friday evening, if I'm not going out, which these days I'm too old for, Adriaan, so I do that. And then Saturday I do the reading, so I read a bunch of stuff, and then I say, "Okay, this is what's going in tomorrow." Then I'll write it on basically Saturday afternoon, which is the construction part of it, before it gets scheduled. So it's scheduled on Saturday evening for it to go out on Sunday. So it is weekend work, but like I say, I clip off time during the week where I have a different sort of rhythm, so it's not like I'm working seven days. I generally probably work less now than I certainly did before. But what it is, it's always on. So the only stress I have in my life, Adriaan, is the fact it's gotta always go out. I've kept the standard now, so I won't pretend that isn't pressure. It's a form of, okay, I've gotta keep delivering here. But because I enjoy reading, and I'm still passionate about the subject, and the commentary is my opinion and I'm quite opinionated, all of these things don't feel like mega work. I remember back in the day when your boss says, "Hey, I need that report by Monday," it's a terrible strain, 'cause you gotta just work through this thing. For me, it's not a strain. I read it 'cause I wanna; I'm interested. I'm really blessed to have this component, and that really helps. Part of the reason why people say, "Look, pick something that you're really gonna get into and be passionate about," is because the consistency required to constantly produce means that you've gotta have a deep passion for it, I think. So yeah, I'm fortunate to have found it.

[00:28:13] Adriaan: Amazing. You read probably the most content of anyone in our industry, or at least you're in the top 1%, right? You have a weekly podcast. You speak to talent leaders all the time, right? You're really tapped into everything that's happening. What is front and center for TA leaders right now? What are they worried about? What are they talking about? What do they come to you for advice about, to pick your brain, pun intended?

[00:28:41] Hung: I mean, TA leaders always talk about tech. What's happened over the last three years or so is just an explosion of technology products out there. Also, the barriers between technology categories have started to blur completely. So you see ATSs building into HRISs, you see point solutions building into ACSs, you see all kinds of stuff happening. So it's a confusing marketplace for TA leaders, who have the feeling that they need to get these decisions right, because the technology stack that you end up building now, that's gonna be AI-native technology. And the nature of AI is that the more context you give it, the better it's gonna be for you. And with a lot of those decisions we're making today, because you're gonna put in a lot of your context, you're probably never gonna switch off it. So it's not the same as having dumb software that you can swap in or swap out simply by exporting a CSV and importing the CSV into the new system. Now it's like, how do you import the context? Very difficult to think about. So a lot of the requests I'm getting are on that technology decision, which, interestingly enough, I can't ever give a recommendation on, because I have a conflict of interest. I had this email earlier today. 'Cause of course, Brainfood is sponsored by recruitment technology businesses. Again, an early decision I made with the business model was to say, "Look, I will be sponsored by companies. This is how it works, but the sponsorship has to be explicit, it has to be labeled, and I'm not gonna do recommendations, even if I think it's good, because I'm obviously compromised here." So yeah, someone today asked me about recommendations. I said, "Listen, I can't do that. Here are the reasons. But obviously there's tons of great stuff out there. Go ahead and speak to other users if you want." So yeah, that's the most common question, for sure.

[00:30:43] Adriaan: Hey Hung, in terms of front and center: technology, AI obviously is a hot topic. What is your take? We're now about three years in. I remember the early days when it came out. I actually remember you were one of the first to post early on about ChatGPT, five million users in the first week, right? For some reason that post really stuck with me. We're now three, four years in. The dust has settled in some places; in other places it's still very murky. What are you seeing? What are you picking up? Are recruiters going to lose their jobs, yes or no? Are we gonna have AI agents run interview processes? Are we all going to lose our jobs? Are you worried about what AI is going to do to your business? What's your take?

[00:31:25] Hung: You know what? It's very easy to fall into a pessimistic view, because the performance of AI, for anybody who's used artificial intelligence to any degree, is remarkable. It's still stunningly powerful, and you can see, yeah, it's gonna eat into a lot of the ways in which we used to do things. But technology is rarely the complete solution. I went on a very interesting tour earlier this week to Tower Bridge in London. Tower Bridge, for the folks who don't know, is a moving bridge. There's a big engine that moves the bridge and opens it.

[00:31:57] Adriaan: In London. In London, yes.

[00:31:59] Hung: Yeah. Did the tourist thing on my day off, mate, right? I did the tourist thing, and there was a really interesting component of the exhibit where they talked about the jobs that were being performed on that bridge. There was a coal stoker, there was a lamp lighter, there was an oil man oiling the machine. All of these were just jobs. And then you think, how did those jobs emerge? They emerged because new technology emerged, but the technology was not complete. It needed maintenance, it needed repair. The technology itself created new demand and new issues that needed other people, and human beings basically came around and filled in the gaps as to what that technology couldn't do. And I think that's still gonna be the case with technology innovation today, even though AI is a different beast, because it may well be that the gaps will be filled by artificial intelligence itself going forward. So your artificial intelligence will create a demand, and then it may fill the demand, and I think the impact will be significant. So yeah, you can take the positive view. I choose to do that. I do think this time is different; I do feel that as a vibe. But there is a pathway out. I've been thinking a lot about AI, where it sits and where the intelligence coheres. So people might wanna bear with me a little bit on this, 'cause it gets into some fairly marginal opinions, Adriaan. If people call me crazy at the end of this little monologue, I'll understand.

[00:33:28] Adriaan: Great. Great, so tell me.

[00:33:30] Hung: So the bottom line is, I think if we stay on the current default, which is closed-source, cloud-deployed AI, and we rely on that, I think we head to headcount zero, and indeed we will lose a significant amount of work without knowing it. The reason for that is because we are training an intelligence that is vended to us and that we don't own in our business. OpenAI owns that, or Anthropic owns that. And I think without us knowing it, our businesses get hollowed out. Now, those vendors, by the way, look at your business as a set of workflows that can be automated. They've got no interest in defending your headcount and no interest in defending your payroll. In fact, they want the headcount to go down so that your payroll can be transferred to them in the form of subscriptions. Right? If we stay as we currently are, then I think yes, our companies will be on this journey of constantly shrinking, constantly automating. Now, individual companies may increase, because you might actually be cannibalizing the market share of your competitors, 'cause you're more efficient or better priced or whatever it is. But in aggregate, as an industry, the numbers of people doing it will probably go down. That's one future. We're on that current track, I would say. The other future is, if we switched off this and went to open-source, on-premise AI, the polarity completely reverses on what it is to hire someone. If, for instance, all of the human workers you're currently employing end up training artificial intelligence that's on-premise, that doesn't go into the cloud, that you actually own, then it changes the entire calculus as to what hiring a human being is. You're no longer thinking of hiring as a cost-per-hire metric, or of whether the person you're hiring is just a simple task executor: this person does 100 tasks, I'm gonna pay that person money to do those 100 tasks. Instead, you're getting someone who is a capital investment in building and training your AI. In other words, if you have on-premise AI, every human being you end up hiring will compound the value of that intelligence. The organizational intelligence compounds. But it is dependent on what is currently a technical decision: do you host AI on-premise, or do you have it vended to you on a cloud? I think recruiters should unequivocally support on-premise AI, selfishly, 'cause I think it means growing businesses and growing headcount. The other side means shrinking businesses in terms of headcount.

[00:36:17] Adriaan: And why do you say that in the second scenario headcount will grow? Why would the scenario still not be true that I, as a business owner, have an AI running on our local servers, understanding all the tasks that need to be done, and I will just be able to improve my margins because I have AI doing some of the tasks of some of the people?

[00:36:41] Hung: Yeah, and you could choose to do that. However, having an employee gives you the opportunity to constantly train that artificial intelligence. If you stopped training it and said that's the end of what the AI's gonna do, then that's a different story. But if you wanted to have an organization that continually evolves and continually grows its intelligence, you're gonna need human beings to input into it. And human beings are very unique in a number of ways. Firstly, we're embodied intelligences. We're bringing in intelligence and information that the AI sitting in a box cannot get access to. It doesn't have eyes and ears, doesn't have sensitivities, doesn't have any of those things. All of those human workers, when you think about it, they all go home. And when they go home, they're absorbing new information, and the next day they come back and bring that information back to your business, feeding that intelligence. With the AI, you switch the box off, and there's nothing happening. That isn't the case with a human-AI symbiotic business. The second thing is that we are all enmeshed in a system of relations. In other words, your colleagues, your work colleagues, your peers, you're all generating intelligence simply by having off-the-cuff, undocumented information. For instance, you and I were at a conference not so long ago. Wasn't that fantastic? Some of the best conversations that we had, most of them were not documented and never fed into the training data. Human beings give you access to that information, that intelligence which the AI currently doesn't have. And thirdly, we're all enculturated. In other words, we're embedded in a cultural system of values. So when we talk about differentiation, when we wanna talk about doing things differently, it is that value system that makes it happen. One of the features of what, in my opinion, will happen for companies that commit to cloud-deployed AI is that their products and their companies will inevitably become homogenized. Because it goes into an AI, the AI's gonna make a bunch of recommendations, and those recommendations will probably be very consistent across the industry. Imagine a tax consultancy. I have a tax consultancy. I decide, okay, Claude is gonna be my strategist for how to run my tax consultancy. I'm gonna feed everything into Claude, right? Everything into Claude. I'm gonna train Claude how to do that. You're training Claude how to do that, but you're also training Claude how to do it for everyone else. Every other tax consultancy is gonna benefit from the training you're giving it, because suddenly Claude is gonna be able to say to all of them, "Hey, this is the smart way to do it." So all of our businesses will stop differentiating, and in fact, we're all gonna go down a commoditization route, because all of it will look pretty much the same. Now, when you have on-premise intelligence, that's a different story. It's gonna grow independently from any other intelligence, and it's gonna look very different from your tax consultancy friend next door, 'cause they're running a different sort of business and a different sort of intelligence. So I think the route forward for us... We're using a lot of biological metaphors, but if you think of a forest purely as lumber, then it makes sense to have a monoculture, put it all in a certain set of rows, and then you can harvest the lumber. That's basically what it is. However, if you think a forest has a lot more to offer than purely the wood pulp you're gonna get out of it, maybe it has medical value, maybe it contributes oxygen to the atmosphere, maybe it has all of those things, then you're not gonna do monoculture. You're gonna create conditions for the ecosystem to grow organically, and you're gonna create this dense, rich, chaotic place that is super resilient to external threat. It's super rich. Some of the longest-lived organisms exist in forests like this, so they're very consistent and resilient over time, and we're gonna get richness out of it which we can't predict or plan for. And I think that's the difference between closed-source, cloud-deployed AI and open-source, on-premise AI: what kind of forest do you want, basically.

[00:40:49] Adriaan: I love that metaphor. I love that analogy. The thing I would question as a business owner is: in the closed environment, with the pace that AI is going and the self-learning abilities that it has, does it really need consistent human input? Can it be, instead of 100 people, maybe 10 people? I don't know. Question mark. Question mark.

[00:41:16] Hung: Question mark. That is true if you're thinking along the lines of a task execution type of method. In other words, if we had people that were very tightly defined in terms of the jobs that you wanted them to do, absolutely. I think that's true in any case, closed or on-premise or cloud. Anybody narrowly task-based, it's not gonna happen for them. Where I think recruitment gets interesting is that we're gonna start recruiting genuinely for talent again. And what is talent? It's not a person that can execute a task. It's a person who has some attributes or traits or behaviors where you're just like, "You know what? I gotta get this person in my team. I don't even know what job this person's gonna do. All I know is I know this person's gonna kick the ass out of it." And when I say this, immediately everyone listening, including yourself, Adriaan, visualizes who I'm talking about: two or three people in your mind. That is the person. Recruiting will go to source more of those folks, I think, going forward.

[00:42:18] Adriaan: I like this a lot, what you just shared. That definitely resonates with me. Just as an example, our head of marketing, Andrew, you've met Andrew. By trade he was a content writer, right? Now he's building very sophisticated workflows with AI. He built our entire website from scratch. He never developed a website himself, but because he has that mindset, he's able to look at and use AI outside of his field and see what's beneficial for the business and our overall work, and he's been able to build that from scratch himself.

[00:42:49] Hung: Yeah. And AI will unlock this, but it will also highlight people that can do this. And maybe one of the biggest disruptions that we're gonna come and see is that the people who are task execution only probably will not be optimally positioned going forward. The get-things-done types that can handle a lot more improv, that can just create great outputs, I think they're gonna really thrive in this future.

[00:43:16] Adriaan: Yeah. What I see with some of my fellow entrepreneurs and founders, and I've been guilty of this, is that they get into this dopamine loop of just building things, because it's so fun to immediately see that result, and like, wow, I'm all of a sudden capable of doing this. And two things happen. One, we even had to catch ourselves: "Wait, why are we actually building this? Is this really necessary? And is this automation or product or app, or whatever it was that we were looking at, worth the time investment to actually build it and make it operational, versus what it's actually saving?" And two, who should be doing it? Because I've seen a lot of founders that should be focusing, whose hours are arguably the most expensive to the company, all of a sudden doing pretty low-level tasks just for the sake of feeling good, the dopamine rush, and being stuck in this AI loop.

[00:44:10] Hung: Yeah. And engineers, by the way, have known this for a long time. They end up just feature building, because they enjoy it, and it feels tangible, and this is the thing. But is it adding value? And what's the opportunity cost? That's something we all gotta think about. I think we're probably not gonna be able to solve the problem, because what will happen is we're gonna end up creating a lot more ephemeral stuff unintentionally. In other words, we'll create a bunch of stuff that no one ever uses. It'll just die on the vine, and then, through natural attrition, some of those things will just emerge as being useful going forward. And maybe we'll develop the muscle memory, the intuition, to know what is worth investing in building, or what is worth just going into a marketplace real quick and getting ephemeral use out of. In the same way, for instance, not so long ago, if you wanted to convert an image file into some other image file, there'd be some free website. You can just go on there and do it real quick. I think we'll probably end up having a lot of those types of micro, nano apps that won't exist permanently; they'll be almost ephemeral. They'll pop in and pop out. You need it, you spin it back up again, and off you go.

[00:45:20] Adriaan: What's actually interesting is that I've been watching Anthropic and OpenAI, how they're scaling their recruitment teams, and they're actually expanding; they're building out their recruitment capabilities with recruiters, right? I find it fascinating that the companies at the forefront of AI are actually following a more traditional SaaS playbook in how they're building out their recruitment teams.

[00:45:43] Hung: My view on that, though, is that ultimately they end up recruiting people that should be working for you. Take the entire concept of their forward-deployed engineering team: they're gonna hire thousands of people, right, to do hundreds of thousands of people's jobs. It's a similar story to... I don't know whether you were in the recruiting business pre-cloud, Adriaan.

[00:46:07] Adriaan: I was not, no.

[00:46:09] Hung: Okay. So pre-cloud, basically every company had an IT department, because the only way to access computers was to buy a bunch of servers. You'd have a room, a server room, you'd have an IT manager, you'd have a database administrator, a network engineer, you'd have a server guy. Every company, every law firm, would have a team of five, six people doing IT. We called it IT back in the day. Amazon basically sucked up all of those jobs into the cloud, because they deployed a more efficient service, which is, "Hey, no point in you maintaining these crappy servers and trying to manage your own products. We'll do it for you. We'll vend it for you. Here's a cloud service. Here's a SaaS." And that basically unlocked the SaaS world. Suddenly, we had much greater access to much better software as a result of the cloud being there. 'Cause previously you'd have to install the software. It would cost you like half a million quid, and then you'd have to pay X amount of maintenance. It was a nightmare. So anyway, the cloud unlocked the door, brilliant. However, it did wipe out IT departments. And what I think will happen is that AI, cloud-vended AI, will do the same for sales, marketing and everything else within your business, because all of those jobs go into the cloud. The AI will deliver the value and capacity for you, and you will go to headcount zero. As a CEO, it makes sense for you to do that, because you want to get maximum profitability for your company. So I don't blame the CEOs for doing this. I just don't think it's good for us in recruiting, because recruiters obviously need more roles to be hired for. So I think we should come, unequivocally and selfishly, to a different future where AI goes back to being hosted on-premise. That means a reconstitution of the IT department, by the way. That means we need an AI infrastructure person. We need a cybersecurity person. Now, there are lots of reasons why this is nuts and inefficient, but inefficient also means resiliency. It might also mean security. And in my opinion, it also means coherency of your business. In other words, your business doesn't get hollowed out. And it also means sovereignty, institutional sovereignty of your company. In other words, you can make decisions. Right now, we've already seen very early instances, which are the doom-mongering stories of overly relying on cloud providers. We've seen rate limits on the frontier models. What if you've built a workflow that's really dependent on you having those limits open? Well, sorry, those workflows aren't gonna flow. We've seen undisclosed model rerouting. In other words, you've paid for Fable 5 or whatever, but, you know what? "Nah, I don't think I'm gonna give you that, but I'll give you something else, undisclosed." We've seen models give you fake information based on guardrails and security, right? We've even seen, very latest, a complete cessation and withdrawal of service on national security grounds. There are a couple of companies suing the US government right now, saying, "Hey, we built our services based on this, and you've just chopped it out. We can't run our business anymore." So you lose control of your company. And I'm saying, you know what? That's not worth the risk for a lot of people. I think it's worth the risk for some, but not for everyone. For the ones who are not everyone, you need to basically have an alternative, which is on-premise AI.

[00:49:28] Adriaan: Hung, to kind of wrap it all up, what would your advice be to our listeners? What should they do, and what should they not do, in order to stay relevant?

[00:49:37] Hung: Firstly, ignore all advice, because anybody who loves giving advice basically loves the sound of their own voice, and you should always advise yourself. That's generally speaking my own policy. You are the captain of your ship. No one knows your context better than you. In your individual, independent situation, you're the best decision-maker, so trust yourself on this. So that's the first thing. But I'll give you three more, which I think are universal values. I basically give myself this advice, so I'll happily share it. Firstly, I think you need to invest in your network value. In other words, this is the concept of understanding all of the people that you know, and that you've got to put some time and effort into that. Your ability to activate that network before you need it is gonna be a superpower. Remember, a lot of that information, a lot of those values stored in that network, they're not fed into the artificial intelligence. Your ability to activate that is gonna help you out in an undefined way going forward. Number two, and related to this, is that you need to boost your profile, right? This is true for almost anybody in business. It's definitely true for entrepreneurs, and it's also definitely true for recruiters. We are entering into an era of profound distrust over everything that's happening in the digital space, and we're just clinging on to things that we think may be true or authentic. One of the best ways to become trusted is simply to become more visible, warts and all. And this kind of means externalizing a lot more of your life, perhaps, but I think it's useful and it's valuable, and most of the time it's not harmful. So increasing your profile is step two. And step three: we talked a lot about AI, but you gotta dive in. You can read my newsletter, that's great. But you know what? You have to just communicate with this new way of working and do it every day. In the same way your coach said 30 days, and my friend said 52 newsletters, you've got to do the reps with your AI of choice. Make sure you're doing something with it every day in a regimented way. Block out your calendar, and break through. You'll have a breakthrough moment six months down the line or whatever it is, and then your world will change, and you'll probably be much better, much more efficient, much more confident.

[00:51:57] Adriaan: So: expand your network; share what you're up to and be external about the things that you're doing; and three, take time and invest in yourself to play with AI, and do that on a consistent basis. I love it. I love it. Hung, you've always been extremely generous with your time, with how much you give back without ever expecting anything in return. What is something that we, our listeners, could do to support you and help you on your journey, and also as a token of appreciation?

[00:52:24] Hung: You know what? You could bookmark and share the big list of recruiting and HR events to attend in 2026. This is crowdsourced, so it's not mine. I set it up, but basically every bit of content in there has been inserted by the community, the HR and recruitment community. These are in-person events that are happening in 2026. I think you should make use of this spreadsheet. Go and attend some events. If you don't see an event there in your city, set up your own event, even if it's a mini meetup or whatever it is. I am a strong believer that in-person networking, cultivating community intelligence, is gonna be the thing that sees us through this moment. So yeah, do me a favor and do yourself a favor and check out that spreadsheet.

[00:53:10] Adriaan: Love it. Hung, thank you so much for your time. I really enjoyed our conversation. Awesome to have you.

[00:53:16] Hung: Brilliant. Great chat with you, Adriaan. Thanks for having me on.