African Talent Meets Open Source AI

Gerard Holland of TalentMatch Africa explains how African remote talent, compulsory AI fluency on every placement, and open-source model economics are reshaping offshore staffing. On The GEO Show with Paris Childress, he traces the Mandela-sparked founding story, Starlink-era connectivity, Claude-expert embeds inside classic enterprises, and why LLM recommendations can out-trust Google rankings for B2B conversion. Paris connects the same thread to proprietary knowledge bases, GEO measurement, and agentic publishing workflows.

[00:00:00] Paris Childress: Hi, everybody. Welcome back to The GEO Show, to you by GEO Forge. Today, I have a special guest, Gerard Holland, who is the founder of Talent Match Africa, today we're gonna be talking all about interesting business, how he's balancing, both the human labor and AI, and and then we'll get more into AI search and GEO. Gerard, thanks for joining. Welcome to the show

[00:00:28] Gerard Holland: Thanks for having me, Paris, and looking forward to the conversation

[00:00:33] Paris Childress: Same here. Why don't you kick us off with a little bit about your background? 'Cause I didn't do much of an intro there for you, but I'd love to hear how you got it all started, what led to Talent Match Africa and where you are today.

[00:00:45] Gerard Holland: Yeah, sure. So I started my career in, in accounting and finance, and I worked for a a large, you know, accounting and advisory firm in Melbourne. I got my chartered accountancy and moved to London as most

[00:01:00] accountants in Australia do, and spent a bit of time working in v- investment banking at Royal Bank of Scotland during the GFC.

[00:01:07] That was a, a pretty fun time. Two years there and returned back to Australia, and I set up an outsourced CFO division for the firm that I, I was originally worked for as a grad. So went back there and you know, set up a pretty substantial outsource CFO business. And one of my clients was in international education.

[00:01:25] I started doing a master's of entrepreneurship at the same time, and eventually I decided I didn't wanna be in a, in a firm anymore, and I wanted to go out and start my own business. So I I started a company called Outcome.Life, which was to help international students with employment after their degrees.

[00:01:41] So I, I saw the universities were being a pr- doing a pretty terrible job of, of actually helping in- international students get a return on investment, which I believe is getting a job at the end of your degree

[00:01:52] Paris Childress: seen that, I've seen that myself too. I can totally agree.

[00:01:56] Gerard Holland: Yeah. I mean, the, if you're going to spend you know, $100,000 to

[00:02:00] move to another country and get an IT degree, you wanna get a job in IT. But the unis don't actually see it that way. And, you know, international students are just treated as a cash cow. And, and what annoyed me the most is that their fees would actually be spent on financing other parts of the university's business, like research.

[00:02:19] So, you know, it's not on... You go and spend a whole lot of money on a product, but most of that money is spent on a different product that's not even related to you. So that sort... I, I felt sorry for international students, and so I started a business to try and to help them, and we fell into doing internship placements, and we just built a tech platform called InternMatch to, to facilitate internships.

[00:02:38] Then, you know, ironically, the universities all started becoming our clients as graduate outcomes bes- started becoming more important. So I took that all over Australia, i- into the UK, and into Canada. And then in in January 2020, I met Zondwa Mandela, who's the grandson of Nelson and Winnie, and he asked me why I was helping privileged people.

[00:02:58] Yeah. And I, and I said, "Well, hang

[00:03:00] on, mate. Like, the international students aren't all privileged." And he said, "No, what I mean is we need a model like yours in Africa. You know, we have a really young population you know, that is entering the workforce, and we need a business model like yours to, to help get people, you know, young people into jobs."

[00:03:15] I said, "Okay. Well, I've never been to Africa." But that sort of started something inside of me, and, and it's called the, the entrepreneurial instinct that you get. And I thought, "Okay. Well okay, what would life look like if we sort of branched into Africa?" It's not what a... Businesses in Australia don't usually expand into Africa.

[00:03:32] But I eventually jumped on a plane and flew into Ethiopia, then Rwanda, Kenya, and South Africa, and I was just blown away by the quality of the talent on the continent. Just really educated people, great English, enthusiastic, determined, want to work, tech-savvy. I was like, "This is the world's best-kept secret."

[00:03:49] So just launched Talent Match Africa on the back of that. And yeah, fast-forward to today, and we've got people working from Addis, Nairobi, Johannesburg, Durban, and Cape Town, and from our offices and,

[00:04:00] and working for clients in the USA, Australia, UK, Canada, Dubai, France, and China.

[00:04:07] Paris Childress: Wow. your experience sounds a little bit like my mine in when I arrived in Bulgaria twenty-one years ago. also realized while there is, This-- It felt like this is one of the best-kept secrets, at least in Europe, because no, no one really thinks about Bulgaria, and at that time, it was even before they joined the EU. But I think there was a great ratio of quality of life to cost. So it wasn't the best of life in Europe, but it was pretty good. The costs were outstanding. Those have gone up quite a lot. But I always think about that, that ratio and everybody I met just had a hunger, it seemed like, they wanted to be part of the game.

[00:04:48] They wanted to be seen as capable, and I think they all felt a little bit overlooked and I just sensed that there was a real ambition, especially with the young people. that drew me in. That was one of

[00:05:00] the reasons I stayed, 'cause I came here on a, what was gonna be a one year, a one-year project with a USAID USAID grant, the now defunct USAID.

[00:05:12] And then I wound up staying. But yeah, kind of similar stories. You land in a place and you realize, wow there's great people here. they're hungry, they're ambitious, they're smart, they're actually just as good as the, the people we have in the US, and in some cases better.

[00:05:26] Gerard Holland: And I think

[00:05:27] Paris Childress: wanted to

[00:05:27] Gerard Holland: Africa's, Africa's like that now 100% where I, I describe it like Africa's loo- looking forwards and, and so people there they, they know that things are, are on the way up. There's prosperity, there's opportunity, economic development, like it's just so much going on and, and you know which direction it's going.

[00:05:44] And I compare that to when I'm in America and it's like America's looking backwards almost and worried about where they're heading and where it's all going and, and it's like they're going in two different directions. So if I was gonna be part of any journey at the minute, I'd probably prefer to be part of the African journey than the

[00:06:00] American journey.

[00:06:01] And the next 20, 30 years is, is going to be so exciting on the continent

[00:06:06] Paris Childress: Of the things I've loved about being in Bulgaria is the rate of change that I've witnessed over these 20-odd years. It's phenomenal. I think I've seen more change here than I would've seen

[00:06:16] Gerard Holland: Yeah.

[00:06:17] Paris Childress: of in the US, and I really like that too. That gives me a sense that my life has more meaning when I'm in a place changing a lot, and I even feel like I'm part of that.

[00:06:27] In a

[00:06:28] Gerard Holland: Yeah, yeah

[00:06:29] Paris Childress: that. So it's more rewarding that way.

[00:06:32] Gerard Holland: Yeah, couldn't agree more

[00:06:32] Paris Childress: that's really cool. Yeah. And when I think of Africa, I th- and when I think of tech, technology in Africa and the stories that I read, one word comes to mind all the time, which is leapfrogging. Africa leapfrogged into mobile payments.

[00:06:48] Gerard Holland: Yeah

[00:06:48] Paris Childress: they were late to the classic internet, but then they leapfrogged into mobile. is that happening to any extent with AI, or am I off the mark there? Or can you talk to me about the tech

[00:07:00] scene overall in Africa?

[00:07:02] Gerard Holland: No, I think you're spot on and I've just come from Kenya a few days ago and M-PESA is a really good example of that mobile banking that, you know, a lot of people don't have bank accounts in Kenya, but they have M-PESA. And you literally, you trade everything on M-PE- and M-PESA was built before WeChat and it, and it's a similar type product.

[00:07:19] You can do everything on it. And now the same thing's happening with AI. The young people, actually young and old, are using AI extensively, and I think the open source, weighted source models that are now becoming more popular is going to create an enormous amount of opportunity because, look, Anthropic is expensive, but Kimi 3 is 112 times cheaper per million tokens than Anthropic.

[00:07:42] So as these weighted models have got so good now and, you know, they, they outperform on a lot of metrics. So if you've got a model that's 90% as good as the latest OpenAI or Anthropic model that is actually pretty cheap, then just think about what capability that actually brings to people on the African

[00:08:00] continent.

[00:08:00] If you've got a computer, you've got internet you can utilize these tools. And that I think is a huge leapfrog opportunity where, you know, you can just start accelerating into the global workforce by, you know, utilizing these tools and technologies that just allow you to do your work three times better.

[00:08:17] Like, you know, I'm obsessed with it. I, I make sure that I'm trying to w- operate at three times you know, more the productivity, efficiency, capability than I could a year ago. And I say that to all our team in Africa that, you know, you've, you've got to be spending at least an hour a day sharpening the ax.

[00:08:31] Everything you do, work out how you can do it with AI the next day, and y- you're improved by 1% every single week and after a year and, and, you know, y- half your job you've automated. So I, I think you're spot on that, that there is gonna be a leapfrogging opportunity with AI

[00:08:47] Paris Childress: Yeah. Yeah. I like that sharpening the ax. I was watching a video... i've been binging on Grok Bot use case videos, was one guy, he told his team that today-- and they had Grok Bot.

[00:09:00] He said today, every single task you have to give to Grok Bot, even if it's hard, if you don't think it's gonna work, if it breaks, no matter what, every single task, give it to Grok Bot first, see what happens.

[00:09:10] It's just a exploratory thing." And the result was that they were all amazed at what it could do that they never would have even tried, basically. And I think that's a theme with AI, is it takes a bit of courage to think. C- a lot of times the first instinct is, "All right, that's gonna be too hard.

[00:09:25] Let's wait on that. Let's start with some easy steps." But I

[00:09:29] Gerard Holland: Yeah,

[00:09:29] I've, I've got,

[00:09:30] Paris Childress: your neck out sometimes and just see. Let's try to break it. Let's see if we can... There's no

[00:09:34] way it can do that. Let's see.

[00:09:36] Gerard Holland: Yeah. You don't... You get pretty surprised by what's possible, right? You don't, like, not until you get into it and you go down more rabbit holes, your bel- your understanding of what's possible keeps getting, like, longer and longer and longer. And thanks to you, I'm gonna be spending this weekend in Grokbot and working out what else we can do with him.

[00:09:51] And as you know, I've got a date with my wife tonight, but is that something I should outsource to Grokbot, the, the whole date thing?

[00:09:59] Paris Childress: You gotta draw the line

[00:10:00] somewhere. I've already ruined your weekend, so you better make tonight count because you're If you're buried in Rock Bottom all weekend, that's not gonna score you many points.

[00:10:08] Gerard Holland: If you've, you've also made my weekend at the same time. I actually can't wait

[00:10:13] Paris Childress: I think, I think Rockbot, I, I have one routine that gives me weekend ideas. That's always good. But I think it's still good to un- unplug from time to time. It's hard. It's hard to not go down that rabbit hole though sometimes es- especially when you get a new tool. I wanna j- I wanna jump back to the internet, AI trends and understand the foundation. You mentioned open source, Kimi. Is there a solid internet? Are they getting it a lot from Starlink, or is it from other places? Is there a good f- is there a good infrastructure for internet and connectivity?

[00:10:45] Gerard Holland: Yeah, like in, in the cities where we operate, you know, we've, we've only got people that are in the cities. I said Addis, Nairobi Johannesburg, Durban, and Cape Town. Yeah, th- there's fiber lines into all those cities. If you go to rural parts of in the

[00:11:00] country, then yeah, you're probably gonna have some issues with, with connectivity.

[00:11:03] But that's the same in Australia as well. Like I've- there's pockets in Melbourne where you can't get proper internet. So there's yeah, the connectivity is really good strong internet. There's, you know, the, the there's been... I mean, I think even Meta's laid a l- a lot of, you know, heavy cabling across the ocean and so co- connectivity these days is really strong.

[00:11:21] Starlink's, I think, an interesting and, and that you raise it too, like I love Starlink. I just, I, I invested in SpaceX because I love Starlink and

[00:11:30] I think that Elon could give internet to everyone on the planet and, you know, once they ex- you know, get that technology going a bit where you can go direct to mobile and as opposed to a satellite, which I, I understand they're not far from being able to do, that that's a pretty exciting thing for humanity, that people in rural parts of the world, which could be outback Australia or it could be you know, in a country area of, of Kenya or wherever, to then...

[00:11:54] And 'cause my first experience with Starlink was in New Zealand on a boat and w- you couldn't see land from

[00:12:00] anywhere and I needed to do some work, and they had Starlink and, and it was the fastest internet I've ever used. It, it just blew my mind how fast it was and I was like, "Okay, th- this is game changing for the world if you can give that access and connectivity to people no matter where they are."

[00:12:15] Imagine schools in rural Ethiopia, right? That currently don't have access to say the internet, they're too rural. Now all of a sudden they can get Starlink into these schools. I mean, the, the ramifications and the, the positive consequences of that is, is unreal.

[00:12:30] Paris Childress: Yeah it's massive. And then now with AI on top, you can bring world-class AI. It starts with the internet connectivity,

[00:12:36] Gerard Holland: Yeah

[00:12:36] Paris Childress: then, yeah, these rural schools can have AI and and again, you might have another opportunity for a leapfrog effect. You may have rural schools in Africa with Starlink and access to Kimi

[00:12:47] Gerard Holland: Yeah

[00:12:48] Paris Childress: K3 experimenting and doing things that aren't even happening in the, the, the top Western schools like, New York City has just banned AI,

[00:12:55] Gerard Holland: Yeah.

[00:12:56] Paris Childress: a c-

[00:12:57] Gerard Holland: So like

[00:12:58] Paris Childress: entire public school system, which to

[00:13:00] me is insane. I think other school districts are, are gonna follow suit.

[00:13:04] Gerard Holland: Yeah.

[00:13:04] Paris Childress: But,

[00:13:05] Gerard Holland: like saying to students, "Oh, look, you're not allowed to use the internet anymore." They go, "Why not?" Well, you'd be thinking, you know, the internet's probably cheating. Oof, what are you gonna do in the workplace? You're gonna use the internet. It, it's no different

[00:13:15] Paris Childress: and they're gonna do it anyway, so yeah. We can, we can go down that tangent, but I think that the just AI, this, this next revolution of AI is just another great opportunity for parts of the world like Africa, where smart people who are empowered, they have the infrastructure.

[00:13:31] The-- It starts with great connectivity, and then they have the tools that are open source, that are next, eith- either free or close to free like the open source models. And then it levels the playing field in such an amazing way.

[00:13:43] Gerard Holland: I agree. Yeah.

[00:13:44] Paris Childress: Yeah.

[00:13:45] Gerard Holland: could be the greatest thing ever for making the world a bit more equal

[00:13:49] Paris Childress: Yeah. I'd like to know a little more about the, the business model. So what, what sort of roles, what are the predominant roles that you're staffing? So it looks like it is a wide... You have a wide country

[00:14:00] coverage geographic coverage. I was looking at the industries here, and I see that there are a lot of industries from accounting to customer service, customer experience, finance. How does it generally play out? What are the most popular roles and what the, the, of the staffing teams that you set up?

[00:14:17] Gerard Holland: Yeah, it's pretty broad because every company I speak to, or we speak to, they, they have different requirements. And, and every time I speak to a, a business I say, "Can you... Do you have someone that can do this? A, B, or C?" And my answer is always yes. And then I go back to our talent team and say, "Hey, can we get people who can do this?"

[00:14:34] And 19 times out of 20, we, we find people that are really capable. So we've ended up doing media buying roles all the way to marketing, and, and marketing's very broad from, you know, SEO and, and and GEO-type roles to meta focus people, to people that are running CRMs and campaigns through that.

[00:14:53] And then finance and accounting you mentioned. Tech roles are very broad. It could be you know, Dynamics and Salesforce-type

[00:15:00] people. But then also, you know, people that are, are front-end developers and back-end developers. With AI, those type of roles are shifting in terms of what you need people to do.

[00:15:08] You know, the, the vibe coding is, is got that good now. And then there's graphic designers as an example. So it drafts people in the construction and, and architecture industries. It really is diverse, but I've been pleasantly surprised that we can always find really good quality people no matter what role is required.

[00:15:25] And, and now the common theme is people that can use AI. So pretty much every company, no matter what role they're, they're looking for, they could be looking for an accounts payable person, and then they'll, they'll say, "But, but I want someone that's using AI." Because what I think what businesses want is that they want people that's gonna help them find efficiencies in the role.

[00:15:42] So, you know, accounts payable might seem like a rudimentary-type role, but you know, the way the CFO would be looking at it saying, "Well, surely a big chunk of that role we can automate." So I just don't want a traditional accounts payable person that's gonna go through the motions. I want someone that's gonna challenge the way we're doing

[00:16:00] things, that's gonna be look at certain AI tools and, and work out ways we can do things more efficiently.

[00:16:05] So AI is becoming like a compulsory add-on to every role that we look at.

[00:16:11] Paris Childress: That seems like a real value add because there are some businesses maybe that might wanna take in some staff from TMA that can actually implement some kind of AI transformation, where you have a young person who has the experience, the hands-on experience, who can actually some of that change.

[00:16:29] Gerard Holland: Yep.

[00:16:29] Paris Childress: seeing some of that happening?

[00:16:31] Gerard Holland: I had a discussion with the, yesterday the CEO of a company in the a- aviation industry. And they were like, "Look, we're not really using AI that much." Like we sh- He goes, "I've got all these ideas about I know we could use it for, but I don't have the time, I don't have the skills." And so we're gonna put someone into their business that is, is gonna be their Claude expert, and they're just gonna, you know, bit by bit work their way through the, the organization and meet with the, the leaders of the divisions, and they're a pretty big business.

[00:16:57] And, and, and that's gonna start with

[00:17:00] just certain processes that the CEO already knows. I know this process could be automated. Like, I don't know how to do it, but I know it. I just know it can happen. So having someone that has the time to then go and, and use the, you know, the AI tools, they're gonna use Anthropic's tools and, and just go and work their way through to apply AI.

[00:17:15] And they're not trying to do 100% gain in one go. It's very, you know, how do we get 2 to 3% type efficiencies out of some of these, some small processes, some big. But just having someone dedicated across the business to go and do that. Now, that role didn't exist 12 months ago. Now it's becoming quite popular

[00:17:32] Paris Childress: That's fascinating. Yeah. I totally agree that the best way to tackle something like this one, one small bite at a time, and you start to layer on.

[00:17:41] Gerard Holland: Ja

[00:17:41] Paris Childress: for the low-hanging fruit. What are some of the most manual... The, the easiest targets to go after that are ripe for automation? Start with those. You don't have to go in and build a master plan and then go and try to execute some- something huge, but just go for the easy wins first and then start stacking them. And then at some point the,

[00:18:00] the bigger plan will take shape. I do think some of the foundational things, as we talked about yesterday getting Claude licenses into the hands of everybody that's a good first step.

[00:18:10] Every- everyone should... we have Claude licenses for everyone in our agency it's a great investment because it gets everyone speaking the same language, using the same tools. A, a year or so ago, we were all, all over the place testing Gemini, ChatGPT, a little bit of Claude. Now most of that testing, I still do a lot of that testing 'cause I can't resist it, and that-that's what led me to GroqBot and others. But but our team is still on Claude and now we have a kind of common vocabulary. There's artifacts that we share with each other. have projects set up and people work in those projects, and then contributing and giving context to those projects. Slowly but surely it has taken shape. And I think that a massive opportunity actually to go in and you can have, I don't know if you have a particular name for this agent, but it's a, it's an AI

[00:19:00] transformation a- really that can go into a classic business has been doing things.

[00:19:04] E-especially I see you work with accountants and finance, like these kind of role-- these kind of industries that can be really stodgy and been doing things maybe for decades a certain way, and you can go in and get massive efficiencies. Even helping people, automate their inbox, their email replies and building, building a knowledge base.

[00:19:26] This is one of the big themes of of our GEO product, GEO Forge, is that before you start trying to create a lot of content, especially with AI, you start with building a knowledge base. You start to try to aggregate all the proprietary knowledge that you have in the organization, which AI doesn't have access to. The LLMs have crawled your website, but that's maybe only a fraction of the real knowledge that you have in the whole organization. And then once you have that knowledge base, ma-marketing is only one use case. Of course, you can create content. That content can

[00:20:00] published on your website.

[00:20:01] It can get also published around the web and other places. But that knowledge base can serve almost every department. It can serve HR and legal and accounting and operations. Th-there's this vision of the f-future AI-driven organization and Jack Dorsey, the, the f-former founder of Twitter, and he now has this company Block he has this vision where AI is at the center, everything, absolutely everything in the company gets recorded and gets noted.

[00:20:32] So this-- every meeting is recorded meetings with customers, sales, everything's recorded, transcribed. It goes into the central knowledge base. All emails, all Slack conversations. e-everything that happens, everything that gets typed, written, spoken, gets brought in, and then the AI has 100% of the context of what's happening. And then AI can drive the strategy, so then instead of a

[00:21:00] pyramid type of an org structure where information flows up and down, and then you have inefficient... like a tel-telephone, the telephone tag game, where, things get passed from management down the structure, and then the work has to get communicated back up. actually have more of a central AI in the middle, and then everybody is connected to that. I'm-- I think I'm going off on a tangent here a little bit, but

[00:21:23] Gerard Holland: Yeah.

[00:21:24] Paris Childress: even with a small business, in small ways, if you can start to, if you can start to get businesses to re-record everything, and even that's hard.

[00:21:32] Everybody should have AI note-takers joining every meeting. there is some privacy concerns, but there's ways around that. get, get all the emails, get the Google Drive, or get whatever your file system is, get all the documents in get all the chats, get all the emails and then you can have something really powerful because then you can have something that can dictate strategy.

[00:21:53] Gerard Holland: Yep. Great

[00:21:55] Paris Childress: that-- I think that is an-- that's a really awesome role that you've developed. And

[00:22:00] have you

[00:22:00] had any

[00:22:00] Gerard Holland: just on Jack Dorsey there, just to cut in, like, the, I'm sure some of your listeners might be, might be thinking this, but do you know what technology Jack's decided to use as his sort of central point for all his AI?

[00:22:12] Paris Childress: It's probably not one of the big LLMs. Maybe...

[00:22:15] Gerard Holland: is it an open source model that he, he's potentially using? 'Cause I... And the reason why I ask is that I think a lot of businesses now are looking at having, like, one of the open source models as their sorta core, and then you use Anthropic or OpenAI. You might use Astra 6 for some things, use Fable 5 for some things, and that might be, like, 10% of it, but you're actually using some of these weighted source models as being the heartbeat of the actual business that is very low cost

[00:22:43] Paris Childress: Yeah. I absolutely agree there that you're gonna have frontier models are gonna be reserved for the really difficult, most strategic tasks, but these are not the everyday stuff. It's what I've started to veer towards with my Grok bot instance, where I have it running on these loops, these

[00:23:00] routines, and a lot of it right now is based on the podcast production workflows for this podcast. but there's more coming. But then the hard stuff I still give to Claude. If I still do proposals

[00:23:12] and I really want those to be great, that goes to Claude now. Opus Opus 5.5 just dropped, which is supposed to be pretty, pretty nice, and I'm gonna be testing that one soon. So I think there's gonna be this mix of models where, yeah, open source, which is gon- is gonna be your power horse, and then you're gonna have some other frontier models that, that do the, the really case stuff, the difficult stuff.

[00:23:40] Gerard Holland: Yep, agreed

[00:23:42] Paris Childress: I wanna pivot over to search now and SEO. We talked a little bit about SEO and content strategy and how you're thinking about transition into GEO or AI search. Tell me more about that. How are you thinking about

[00:24:00] organic inbound channel?

[00:24:03] Gerard Holland: Yeah. So when it comes to SEO and GEO, like we, I remember talking to my team nine months ago about what are we gonna do about to get mentioned on the, on ChatGPT? Like, when is ChatGPT gonna say our name? And and, and probably being a bit ignorant saying, "Sure, what do we need to do just for them to mention us?"

[00:24:19] And then you, you sort of go and speak to people that actually know what they're talking about, you know, people like yourself, Paris. And and I, I sort of take the view that let's not separate SEO and GEO. Let's just treat them as the, as the same thing, and there's some nuances to it, but let's just make sure we're doing all the right things you do for SEO and then, and then think about how you tinker with it for our GEO.

[00:24:39] So yeah, we've got an, an agency that we utilize out of they're based out of Sydney that sort of help with us with our content and setting things up on the website and what content we're, we're actually producing. And then, you know, making sure that we-we're getting optimized. So, you know, we don't...

[00:24:55] I'm sure if someone googled best tech offshore

[00:25:00] resource, that Talent Match Africa is not going to come up. So, you know, we're, we're nowhere near that yet, or who knows how close we could be. But I think, I just think about like, if you wanna convert you wanna convert leads into really high conversion, if Claude or ChatGPT or, or Grok is gonna recommend your business, then the trust factor is just so far beyond anything that Google recommends.

[00:25:24] So being able to crack that nut and getting the LLMs to actually reference your business is almost a game changer for any business, because they're only gonna reference so many. So yeah, it's something that we're putting a bit of effort into now. And then, you know, ideally that just becomes a, a core...

[00:25:39] Well, you know, a part of all our other marketing efforts.

[00:25:43] Paris Childress: Yeah. I think that's a good approach. SEO is not, It's tempting to say SEO is dying or it's dead. It-- SEO is still there, but I think it's not the dominant channel, and it won't be again the, the dominant channel that it once was. And even people that are still going to

[00:26:00] Google, they're getting a lot of AI overviews replacing the classic Google experience. Other people are going straight to ChatGPT. I think in your sector probably there's a disproportionate use of Claude, I think Claude is number one right now for business, and maybe even to another extent, Microsoft and Copilot perhaps.

[00:26:20] Gerard Holland: Yeah, there are a lot, lot of big businesses that are stuck on the Microsoft tech that have to use Copilot begrudgingly, but they use, stuck using Copilot.

[00:26:26] Paris Childress: sorry for all them too, but

[00:26:28] Gerard Holland: yeah.

[00:26:28] yeah.

[00:26:28] Paris Childress: yeah.

[00:26:29] Gerard Holland: you, you probably saw what happened with Muse over the last week, right? So, you know, Amazon and Google are running scared now that Muse has launched, and that's gonna change how SEO works now too because you, you can't advertise to a bot.

[00:26:42] That's not look- they're not looking, they're not eyeballs looking at an ad. And so I think the whole sector's gonna get turned on its head

[00:26:51] Paris Childress: Yeah. R- right now the biggest differences with GEO and classic SEO is that the goal for SEO was to rank pages

[00:27:00] so that you could get measurable clicks, and then you could then you could follow those clicks all the way through to a conversion. And in this case I'm seeing that the-- your primary call to action on the website is to book a demo. But that was all very measurable. With GEO now, the goal is not to rank. You're not tracking rankings anywhere. You're not even really focused on the page as an entity. You're focused more on the passages of claims and knowledge that, that are inside of these pages. But it doesn't matter how long a page is or, you're-- the, the, the page is irrelevant. And what you're trying to do is just to be in the position to either be mentioned in the answer or to be cited as a source which AI from when it synthesizes its answer, and then to be noticed in a way that people will remember you later. But, later that activity may be that they go and they just Google your name and or maybe there are other ways that they'll get to your website that are not easily trackable.

[00:27:55] So I think that way that you measure ROI is very different. You can't go into Google

[00:28:00] Analytics and see how many conversions do I have from

[00:28:02] Gerard Holland: Yeah

[00:28:03] Paris Childress: There's this many, so there's a value to that, and I know what my ROI is. I think the only real way to do it now is when people-- in your case, when people book a demo, and I'll actually click on this button right now and see what questions you're asking. It's not here, but I would strongly recommend you add the question: How did you hear about us?

[00:28:21] Gerard Holland: Mm-hmm.

[00:28:21] Paris Childress: And make it a dropdown. One of those dropdowns would be AI chat, AI chatbot, like a ChatGPT or Claude. And then a- as those leads, I presume you're flowing those leads into a CRM, and then you're later following them downstream into opportunities, and then there's some kind of a pipeline value associated. And then you can correlate that to people that answer that question with an-- with the AI response, and then ideally, you can then get some sense of attribution. What, what portion of my sales pipeline initiated and initially referred by AI because the people have said it. Most people will remember that they started-- they, they

[00:28:59] found you

[00:29:00] because Claude recommended,

[00:29:01] Gerard Holland: Yeah

[00:29:02] Paris Childress: or may- maybe So I think that's the best way on measurement. But the other big change in strategy is that you need a-- I think you need a lot more content, and you do need to use AI to assist with that content, but it can't be AI slop. You can't just go out and say, write me 20 articles about don't know, th- this topic, that topic."

[00:29:23] It-- the, the articles have to be grounded in some sort of proprietary knowledge. This comes back to the knowledge base. If you're building that knowledge base one of the use cases is that you can ground that for content generation, and then you're feeding AI bots and crawlers with what we call information gain with knowledge about your business that it didn't previously know. It didn't have that from your website, or it couldn't find that on the broad web. And then when you do that, you get rewarded with higher mentions and citations when people are searching. I would suppose if anybody is searching with prompts that include outsourcing in Africa, I think that's where you

[00:30:00] really wanna dominate. if they're including Africa, if they already have some sense of Africa as a destination,

[00:30:07] Gerard Holland: Yeah. And I, I read a report last week that these guys did a like some, some testing, A/B testing almost, and they produced a whole lot of content that was AI generated. And they produced a whole lot of content where a human edited it, and then they checked the, tested the results of it as to how they were scoring in, you know, in some of these things that you're talking about.

[00:30:28] And the ones that were human edited outperformed by three to one. So it was a really good case study that anyone can go and produce a whole lot of content now using AI. You could pump out 1,000 articles if you want. But if it's ju- purely AI driven and the humans are not going in to edit at all, it just doesn't have the level of penetration that actually, you know, taking the effort to at least do some editing by a human

[00:30:53] Paris Childress: Yeah. I think that was Neil Patel. Was it his, does that ring

[00:30:57] Gerard Holland: I can't remember who, who did it

[00:30:59] Paris Childress: guy?

[00:31:00] Yeah, N-Neil 'cause I saw a video from Neil Patel that was exactly like that. He took a big batch of pure AI, compared it to another large batch of content that was AI-generated but human-edited he was actually measuring it on the basis of traffic, which I think is still a SEO-first kind of a mindset, but it performed it phenomenally better, yeah, the human-edited stuff. And I think that is the key is when you start with content production and your goal is to anticipate the questions and all the use cases that are out there across all of your personas. So you've got-- and you can, y-you have more than one or two personas. You probably have dozens of personas that are micro personas where you have one person in a role in an industry in a location. Like maybe a a s- a staffing or head of HR in an accounting firm in London is different than even that same role maybe in in Frankfurt, Germany or something. And then you have these people with these different needs, and then they're gonna go and prompt, and ideally you have content that can

[00:32:00] anticipate all those questions and then answer all those questions. publish a lot of it on your website, but then you also need to get your brand mentioned in other websites. So unlike with SEO, you're trying to get links because links and domain domain authority is kind of the currency. But with GEO, it's more about just getting mentioned places where AI tr-- well, AI trusts those sources, and then if they see your brand alongside other brands, they're gonna have more trust. And that, that actually carries through whether you're asking your Muse agent, your Grok bot, or you're just going to ChatGPT and almost using it as a search engine. These are all agentic experiences now, and these AI agents are gonna need to trust brands that they recommend. And so if they see you, if they see you appearing in reputable places around the web, and they also see that your website is publishing all this content and answering all these great answers to these long-tail questions, then they'll trust your brand enough to go to your website and actually pull out

[00:33:00] one of those passages. And eve- if it finds the exact question in an FAQ page that you've published, even better, 'cause they can then just lift that, that answer and drop it right into the user. And then from there you, you're on the radar screen, and then you're on the shortlist. users don't even really need to come to your website anymore

[00:33:18] Gerard Holland: yeah

[00:33:19] Paris Childress: dig deep and to read blog posts or to navigate.

[00:33:21] If AI says TMA if you're looking to outsource from Africa a-and you want the, you want three names, TMA has gotta be on your shortlist. It's

[00:33:30] Gerard Holland: Yeah.

[00:33:30] Paris Childress: So go to

[00:33:31] Gerard Holland: and this

[00:33:31] Paris Childress: a demo, and then you, then your sales team can hopefully close the deal

[00:33:35] Gerard Holland: So is backlinking still a, a, a used strategy for for GEO? And then first question, then the second one is around even being on podcasts, right? Like, so if you were, you were me and I, I'm on your podcast, and maybe you've got a YouTube video or Spotify thing or whatever happens on the back of this, would you then utilize that content?

[00:33:53] 'Cause I'm, I'm now associated to your brand, right? So how would you utilize these podcast links if you were me in order

[00:34:00] to, you know, do a little, a few more little things that helps with SEO and GEO?

[00:34:05] Paris Childress: Ja Podcast. First, the first question about links is I don't think links are relevant anymore. But what we're doing with a GEO-first approach is we're reaching out and trying to get brands placed, either placed in existing articles that list top tools for AI search and things like that. of those lists are a little bit self-serving, but they still work. in other cases, we're pitching exclusive content to websites that they can publish based on some original research that we've done or something like that. And we're saying, "This is something that's gonna be really high information gain. We'll give it to you exclusively if you'll publish it and reference us as the source." to your podcast question, I think podcasting is still an extremely great way to get, especially for like yourself, founders, CEOs, and leaders in businesses, to get your original thoughts out there. You don't really have time to sit down and write

[00:35:00] or three or four blog posts every week.

[00:35:02] I've never been good at writing blog posts either. I can turn on the camera and just talk and

[00:35:06] Gerard Holland: Yeah

[00:35:07] Paris Childress: So this is the best medium for me. And what we can do with this podcast, and I'm gonna have a repurpose workflow where it's gonna go, it's gonna go to Apple and Spotify, it's gonna go on it's gonna become a LinkedIn article for my newsletter.

[00:35:20] It's gonna get published on my website. I'll have a bunch of short clips that get clipped from this video that are gonna go out through social media. And I hope to get, I hope to get a really big surface area from this one episode. I think the same is true for you. You've been sharing a lot of your original thoughts and opinions about the industry and about AI here. That's information gain. If I go onto your website now, I'm not gonna f- probably I won't find these, these specific ideas that you've shared with me today. And so we have to capture this, and this ought to also be c- coming out on your website and and on your social media. I think that podcasting is

[00:36:00] still just a great way to initiate initiate a large distribution strategy starting with a long-form video. th- there's actually two approaches. You can start with long-form and repurpose it into short-form. So we can start with, like this, a 30, 45-minute video that we're recording. And I am aware of the time too, Gerard.

[00:36:21] Gerard Holland: Oh, good

[00:36:21] Paris Childress: and then we can chop it up. We can get five to 10 great clips out of this. We can get some good images and quotes out of this. There's tremendous repurposed value when you go from long to short. You can also go from short to long, and you can use short-form as a way to audition your ideas for which ones deserve to be long-form. So you might say I, I've got all these little ideas and notes that you've gathered, you can do a bunch of short videos or even some short social media posts on LinkedIn, and how- The community responds to that, and if one or two of those ideas hits, then you go with that, and then you go long-form on that.

[00:36:59] So

[00:37:00] if, yeah, one idea seems like a really good idea, I'm gonna I'm gonna do a long-form video on this, or I might try to do an interview w- on this. I'm more in the camp of the first workflow, which is start with long-form and have a good conversation, and sometimes it takes a while for the conversation to warm up and the ideas start flowing. then the good stuff will come out, and then you're gonna clip that, and you're gonna isolate that. And then now with AI, AI can also take a big transcript, and it can really find the best content. It can find the information gain that it's seeking hone in on that. I think that's a great strategy.

[00:37:35] I know that you you reached out to me. I'm sure you're reaching out to other podcasts, and I think it's just a great use of time

[00:37:42] because you're starting with long-form video. You're gonna get great repurposed value. And then as you're gonna do this weekend with GrokBot, you're gonna discover that you can take this recording and give it to GrokBot, and with a few rounds of iteration, GrokBot will give you everything that you need.

[00:37:56] It's gonna it's gonna publish it everywhere. It can repurpose it

[00:38:00] and just be that workhorse for post-production,

[00:38:04] Gerard Holland: Yeah.

[00:38:05] Paris Childress: Repurposing

[00:38:05] Gerard Holland: Yeah, I,

[00:38:05] Paris Childress: and distribution

[00:38:07] Gerard Holland: and I'm a bit similar to you, like sometimes you, you do a, a, a chat like this and you end up saying some stuff, you're like, "Oh, that, that's actually pretty good, that. Like, why have I never, why have I not pushed that before? That's actually a really good concept or a good idea or a strategy or, or whatever it is."

[00:38:22] But it often comes out when you're having a good conversation with someone that, you know, you draw things out of each other, and I think that's where the best ideas come from

[00:38:31] Paris Childress: Yeah. My, my main format right now is I'm getting I'm getting AI ChatGPT-6 Astra is giving me this daily intelligence brief, giving me the top stories in my industry in GEO. And I'm looking through those top stories. I'm picking which ones I wanna go deeper into, and then I'm almost reporting on those stories and then giving my editorial opinion after. in that sense, I'm trying to really get my deepest expertise out there and my thoughts and

[00:39:00] opinions, and I record that, and it's about 15 or 20 minutes a day. In the same tool that we're using here, Descript, I stop the recording. I kick it over to GrokBot, and GrokBot has a 12-step workflow that kicks off. And and then I, yeah, I record. I go to lunch, and sometimes I get back from lunch and here's my-- there's my YouTube link. There's my Apple Apple s- and Spotify are live. The the show notes are live with the recap on the website. The LinkedIn article is in draft waiting for me to review. And it feels like magic really. I was never able to do it every day, but now I can do it every day in just a fraction of the of my day and still actually do other stuff.

[00:39:37] Gerard Holland: Yeah. I recall you telling me, Paris, that you've got a, a chief of staff as one of your agents, and, and that chief of staff is the one that instructs all your, all your other Grokbot agents to go and do the work

[00:39:48] Paris Childress: Exactly. Yeah. I think that's a great setup with GrokBot because you can build m-multiple agents, but if you have so many that it gets hard to communicate with them all and you just are overwhelmed by all the chat. You need to funnel all that

[00:40:00] through one single chief of staff and and then the chief of staff actually doesn't do any of the real work.

[00:40:05] It-- The chief of staff just communicates with you and delegates that work down to the team of different other agents. And yeah, setting that up it's a lot of fun actually. You can give them names and creative titles and things like that. But it's been working, it's been working pretty well. And I probably-- I don't have access to Muse yet. I think once I do, I'll try to use Muse maybe in my personal life to see if it can help me better organized. But th-there's gonna be... The next big wave is gonna be a-- There's gonna be a massive battle for your personal agent. W-- Is it gonna be GrokBot?

[00:40:42] Is it gonna be Muse? Google's gonna come up with something

[00:40:46] definitely and OpenAI will Claude will maybe. Amazon right now is blocking,

[00:40:53] Gerard Holland: Yeah, fucking huge

[00:40:54] Paris Childress: I-- That's amazing to me. There must-- that must be costing them a lot of sales because you're blocking agents that

[00:41:00] are coming to buy on behalf of their their human masters. And why would you do that? I guess they wanna force users to use the Amazon what is it called? Rufus. I doubt that. Nobody uses that, I think. But,

[00:41:14] Gerard Holland: yeah, but w- whenever a business or an industry blocks something, it's like when Uber came into every city, right? And, and the taxi industry were up in arms and they tried to block it. It's a hiding to nowhere. Like, it, it,

[00:41:26] inevitable. There's a better technology, and you can fight as much as you want and go to the courts and whatever you wanna do, but you're just delaying the inevitable.

[00:41:33] Because if you're not reinventing yourself, you know, all the time now, then you're at risk of another technology taking over. So, I think there's a bit of a shift happening for where the, you know, the power for the Mag 7 has been there for the last, you know, 10, 15 years, and for the first time ever they're being challenged, and there's some new kids on the block, and it's gonna be interesting to see how it plays out

[00:41:54] Paris Childress: Yeah. Yeah, it's, it's-- these are great times, actually. It's been a lot of fun. It's very

[00:42:00] disruptive, and the opportunity is bigger than ever before. And we probably should wrap on that, Gerard, but it's been a great conversation. I think we could go on a lot longer. But you've gotta, you've gotta get the weekend started.

[00:42:13] I know you've got some good plans. And good luck with GrokBot. I will, I'll

[00:42:18] Gerard Holland: Yeah

[00:42:18] Paris Childress: in a few days how that went, and I'm

[00:42:20] Gerard Holland: And, and I'll, and I'll get my chief of staff to get one of my... I'll get one of my chief of staff to get one of my agents to reply to you

[00:42:28] Paris Childress: Yeah. My, my agent will be in touch with your agent

[00:42:30] Gerard Holland: Yeah.

[00:42:31] Paris Childress: yeah that, that's where we're headed. Yeah. It's really becoming agent to agent. And that, that-- Yeah we're barreling towards that future now. Yeah. My agent will be in touch with your agent. We may never speak again, you and I,

[00:42:42] Gerard Holland: Yeah.

[00:42:43] Paris Childress: Right

[00:42:43] Our agents are gonna do a lot of business together, so can't wait for that.

[00:42:47] Gerard Holland: Exactly.

[00:42:49] Paris Childress: Yeah.

[00:42:49] Gerard Holland: Yeah.

[00:42:49] Well, thanks for having me, Paris

[00:42:51] Paris Childress: with me today. Hope you have a great weekend,

[00:42:53] Gerard Holland: You too.

[00:42:53] Paris Childress: and

[00:42:54] Gerard Holland: you

[00:42:54] Paris Childress: thanks everybody who, who have been listening. Okay.

African Talent Meets Open Source AI

Key takeaways

In this episode

Africa's talent secret and the Mandela spark

Looking forwards vs looking backwards

Leapfrog: M-PESA to open-source AI

Starlink and rural access

AI compulsory on every role

Knowledge base before content flood

SEO plus GEO, not SEO vs GEO

Founder podcasting and agent workflows

Full transcript

See how AI answers describe your brand

platform

Resources

About us

use cases

compare us