YT YouTube2Text

YouTube2Text → Transcripts → Build AI Agents Before They Replace You | PMAX Scripts and Automation | Mike Rhodes

Build AI Agents Before They Replace You | PMAX Scripts and Automation | Mike Rhodes — Transcript

by Niklas Buschner · 14,786 words · 2,005 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00So, welcome to a new episode of the
0:01Masters of Search Talks. I'm super happy
0:04to introduce today's guest, Mike
0:06Rhodess, who has become one of the most
0:09influential voices around PPC and AI.
0:13After building and selling four
0:14businesses, Mike has spent a lot of
0:17time, probably he will tell us today how
0:20much time, building products like the
0:22famous PMX inside script. I think like
0:25in PPC everybody knows this or at least
0:2890% of the people and teaching people
0:31how to do the same. So Mike, can you
0:34give us a quick intro about your journey
0:36and especially your move from agency
0:39founder or a business builder to AI
0:41builder?
0:43>> Yeah. Yeah. Um yeah, I fell into Google
0:46Ads by accident. I went to a conference
0:49in 2004
0:51um and Perry Marshall was in Australia
0:53for the first time speaking um ripping
0:56into Yahoo or Overture as it was back
0:59then and talking about this new thing
1:00that he was starting to look at called
1:02Google Ads and uh yeah well yeah a quick
1:0620 years later here we are I ended up uh
1:09doing all of the training for Perry's
1:11group and writing the book on Google Ads
1:13with Perry which I believe is the
1:15world's bestselling book on Google Ads
1:16but that was all Perry not me. Um, and
1:19then yeah, built an agent, fell into an
1:21agency by mistake, too. I took me sort
1:23of a year of helping people here and
1:25there and swapping advice for a case of
1:27beer or a round of golf. Um, and all
1:30these people kept saying, "Yeah, but
1:32mate, I don't care how it works. I just
1:35need you to do it for me." Because
1:37apparently my MO is just I like to
1:39explain things and I like to teach
1:41people how it works. So fell into agency
1:44land and all of a sudden turned around
1:46and my kids were still both in primary
1:50school a couple of years ago and I just
1:52thought you know what like this time
1:55I've always been very careful of my
1:56time. I was never working 70our weeks um
2:00maybe right at the beginning before
2:01kids. Um but I just thought you know it
2:04this is a time I do not want to miss. I
2:06know I'm never going to get this back.
2:08It's time for someone else to take the
2:09baton and and take this forward. Uh the
2:12plan was to basically do not too much
2:16and that lasted about three weeks and
2:19then I rediscovered how to code and it
2:22was right around the time that GBT4 was
2:25coming out and and growing and I just
2:28yeah the the the PMAX script as you
2:30mentioned just went gang busters. Um
2:34which gave me a lot of leeway then to to
2:36just explore. Um, and since then I've
2:39been building tools. I've always built
2:41tools for, you know, the the PPC
2:43community going back to 2008 I think was
2:46the first one that I built. I've always,
2:48as you said, tried to teach other people
2:50how to do what we did back in the agency
2:53world. Um, again, starting 2007208, I've
2:56had various sort of membership groups
2:58and, you know, spoken at conferences and
3:01so on. And, um, I just absolutely love
3:04it. I love building stuff. I'm not
3:07particularly good at managing um people
3:10and stuff and things. So like the the
3:12running of a mediumsiz business um
3:16definitely not my forte. I was I was
3:18definitely much happier once I had a
3:20sort of a proper general manager come in
3:23and and she managed all of the
3:24day-to-day and allowed me to to travel
3:27more and and speak on stages and so on.
3:30But um yeah, I just I just love building
3:32stuff. And so now it's just such a
3:35wonderful time to be a builder because
3:37you can go from idea to impact in the
3:41course of a day really and then get a an
3:43MVP out there and get feedback on it and
3:46and see if it resonates and and that's
3:48what I've really been doing most of the
3:49past two years.
3:51Before we dive into more of the PPC, the
3:54AI, the building stuff, I can imagine
3:57that there are a lot of people in the
3:58audience also either from maybe a
4:01freelance background or maybe also uh
4:04agency owners to just like to take a
4:06very short moment uh to recap um this
4:10journey when you were selling the
4:11business or when you were transitioning
4:13out of the business. Can you just give
4:15us a little bit of um an overview of
4:18things you were doing in the runup uh of
4:22you going out of the business? So, were
4:23there certain like key milestones you
4:26set for yourself that you have to
4:28achieve so the business is sellable in a
4:30way? Joe, just I'd like to hear a little
4:33bit of this this process.
4:35>> Yeah. Right. Um
4:37I I mean yes I suppose I so going back
4:40gosh more than 20 years I was an ETH
4:44consultant. I paid Michael Gerber a very
4:46large sum of money a long time ago to
4:48become an E-Myth consultant. I'm not
4:50sure if you've even heard of or read the
4:52book the ETH. It was the small business
4:54bible you know 25 years ago. It was the
4:58thing and it was the thing that really
4:59helped my first business be a success.
5:02And I didn't know there were these
5:04consultants. I went off to the States. I
5:06paid him a huge amount of money and
5:07trained and basically it's all about
5:09work on the business, not in the
5:10business. That's that's essentially it,
5:12right? Um, so I've always been a a fan
5:16of systems and processes that run the
5:19business and the people run the
5:20processes and a business is just a
5:22system of systems. So the business was
5:25was pretty systemized. I mean, it's a
5:27professional service business, so it's
5:28always going to be a bit messy and
5:30especially in the industry that we're
5:31in, there's a lot changing all the time.
5:34It's not we used to joke you know it's
5:35not like insurance where you could
5:37probably do the same thing that you were
5:38doing 10 years ago you know the halflife
5:40of information in the PBC world is two
5:43years so I don't know that there were
5:46milestones as much I mean getting a
5:48general manager like I said which was
5:50probably five years to selling but it
5:51wasn't like I was like oh I'm I'm going
5:53to sell in five years so I'll get a
5:55general manager now if you'd asked me
5:58then I would have said I'm never selling
6:00the way I pictured the business imagine
6:02a big sheet of paper in front of you
6:04like A3 sheet of paper. I've always said
6:06that it's easier to work out what you
6:08don't want to do than it is to work out
6:10what you do. So you picture all the
6:13stuff that you do as this huge sheet of
6:15paper in front of you. It's much easier
6:17to go, "Oh, I don't want to do that half
6:20of the page anymore. Let's let's spend
6:22the next year getting rid of that half
6:24of the page. Okay, now I'm down to a
6:25smaller piece of paper. Now next year I
6:27want to get rid of that half of the
6:28paper and then this bit and then this
6:30bit and then this bit." and you end up
6:32with just this little and that's where I
6:33was. I was just doing the little bit
6:35that I felt I was quite good at and that
6:38I really liked doing.
6:41But I don't know, it was always a um
6:43what I really struggled with, I'm sure
6:45many people will relate to this, but I
6:47always struggled when when a team member
6:50quit or when a client left. I was
6:53supposed to be kind of removed from
6:56that, you know,
6:57>> couple of layers away from that. um
7:00that's why you put in a proper general
7:01manager, you know, but I I could just
7:03never quite remove my my brain and and
7:07the feeling of that. It always felt like
7:09a kick in the guts and I thought, gosh,
7:11if I if I'm still feeling like this now,
7:14I'm always going to care and it's always
7:16going to hurt. Um
7:19ah yeah, I just like I maybe maybe I
7:22should try something different after
7:24after 17 years. Um, and yeah, I had a
7:28chat completely by chance really. I was
7:30actually considering maybe expanding and
7:33buying a creative agency. Like there
7:35were a lot of different sort of options
7:36floating around. And then I just, yeah,
7:38back to two kids in primary school.
7:41Actually, I don't think I want to do
7:44that. I really want to hang out with
7:45them and spend time with them. We worked
7:47out back then that my daughter had 134
7:50Monday mornings left in primary school.
7:53Um, and we've been counting those down.
7:55Um, you know, we don't actually have a
7:58chart on the wall, but we both know
7:59roughly where we are on that list and
8:01we've got about 55 to go at this point.
8:03So, you know, it's been counting that
8:05down and I was just very very aware of
8:07that number. So, yeah, in terms of
8:10milestones, I mean, not really. It was
8:11just had an amazing team. That was the
8:13worst part, you know, being away from
8:15your team and missing those
8:17conversations and missing the the
8:19problem solving part of it. That was
8:22definitely the hardest part of of
8:23dealing with it all. But um no, it's
8:26yeah, dive into a new thing and don't
8:28look back.
8:29>> Keep moving forward.
8:31>> Awesome. I once uh saw this video of
8:34yours on LinkedIn where you were telling
8:36an analogy or a metaphor about playing
8:39table tennis.
8:41um you probably remember this maybe and
8:44after that question we will definitely
8:45move into the AI and the PMGs and and
8:48the PPC stuff but maybe can you quickly
8:50share the story again with the audience
8:51because I think it's so it's so
8:53eyeopening.
8:54>> Yeah, I like this one. I wish I knew
8:56where this came from so I could give
8:57proper attribution. It's definitely not
8:59mine but um
9:01you always feel like you have a lot
9:03coming at you, right? And
9:07basically I get you picture that as
9:10you're playing table tennis and there's
9:12just constant there's a ball coming at
9:14you all the time and it's all you can do
9:16to just keep trying to pick up the bat
9:19and just keep smashing the ball back.
9:21But no matter how much you do it just
9:23keeps on coming at you. and the idea of
9:27systemizing a business really and and
9:29putting some automation in place and
9:31putting letting go a little bit and
9:33having other people do some stuff. It's
9:35all related. But you notice one day that
9:38down by your feet there's a pile of
9:40bricks. Oh, I haven't noticed those
9:42before. That's handy. The trick is that
9:45you have to bend down to pick up a brick
9:49to put it on the table to start building
9:51this wall at your end of the table. The
9:54downside is when you bend down to pick
9:56up that brick, inevitably a couple of
9:58balls go flying past you. I missed
10:00I missed one. And that's you feeling
10:03like you literally dropped the ball on
10:05something. You You've missed something
10:06in the business. Something has gone a
10:08little bit wrong. No worries. We'll
10:10repair that. We'll fix that. Now I've
10:12got to bend down and pick up another
10:13brick. And And if you keep on doing
10:15that,
10:17you keep picking up these bricks. You
10:18eventually build a wall such that the
10:20ball's coming over the net, bounce off
10:22the wall and go back. And to begin with,
10:25not very many bounce back. But over
10:27time, if you keep building the wall,
10:29yes, every time you pick up a brick, you
10:31miss a couple. But eventually, you've
10:33got this whole wall there, and all of
10:34the balls bounce back. And now,
10:37freedom, you know, freedom to do what it
10:40is that you want to do. freedom to do
10:42the thing that you set out to do in the
10:44first place. You know, um Gerber used to
10:47tell the story about the carpenter. He
10:48loves carpentry. He works with wood. He
10:50loves the smell of the sawdust. But now
10:53he's running a carpentry business and he
10:55hasn't touched a piece of wood in six
10:56months, but he's wearing 12 different
10:58hats and he's hiring other carpenters
10:59and he's dealing with invoices and
11:01dealing with clients and but he's not
11:03doing carpentry. He just wishes he could
11:05get back to the carpentry. That was why
11:07he got into this in the first place. And
11:09it's it's a similar sort of thing,
11:11right? But you've just got to you've got
11:12to build the wall. You've got to let
11:13other people do the stuff that they're
11:15great at and you're probably not. And be
11:18okay with missing a couple of balls
11:20going flying past. Yeah, I have no idea
11:22where that comes from, but I love the
11:24the picture that it builds in your head.
11:27>> Maybe at some point in time we will find
11:29the person that originally came up with
11:31the story. But yeah, I found it so
11:33inspiring. So now let's move over to a
11:35little bit more PPC and AI related
11:38questions. So your PMX script which we
11:41um obviously have already introduced has
11:44become some somewhat legendary. I like
11:46the term you use it went gang busters. I
11:49have to use this too. Um and like for me
11:53I feel like um all of the time
11:55especially in startup world people are
11:57talking about yeah the idea is one thing
11:59but execution matters a lot. Now we have
12:02this great new tools and obviously
12:04execution is still very important but we
12:06have stuff where basically our
12:08imagination
12:10uh can become reality in a much shorter
12:12period of time. So I just want to get a
12:14feeling of how did you come up with the
12:17idea for this product or for the script?
12:20>> Oh um oh gosh that's probably going back
12:22about three years. Um, it would have
12:24been a conversation with Stacy and
12:27Calvin and Brent and and the wonderful
12:30team that I had back at Web Sabi back
12:32then and um just really solving that
12:35problem. So, we worked out that you
12:37could work out how much of your total
12:39PMAX spend, how much you were spending
12:41on shopping because you could go into
12:43the product groups, but you couldn't
12:46just export that and sum it all up,
12:49right? Because each line was repeated
12:51two maybe three times. So, you couldn't
12:53even just sum it all up and chop it in
12:54half.
12:56If it was a very simple campaign setup,
12:58you could just about eyeball it and go,
13:00"Yeah, it's probably around about 23
13:02grand last month." But it was a bit
13:05painful to do and a more complicated
13:08setup with more product groups. It was
13:10just hard. And then, but we kind of
13:12lived with that and we eyeballed it and
13:14we moved on. And then we worked out one
13:16day that oh actually if you add the
13:20columns for the number of views and the
13:22CPV, you could multiply those together
13:25and you could work out how much money
13:26you're actually spending on YouTube. But
13:30um gosh, this is really testing my
13:32memory. But that worked for accounts
13:36that had a YouTube campaign running or
13:38had ever ran a YouTube campaign. that if
13:41that campaign had never turned on any
13:42YouTube campaigns, those columns weren't
13:45available. But they were with scripts.
13:48And so we've sort of found this magic
13:50hack, we thought of like, huh, well,
13:52let's save doing all of the maths for
13:55the product groups. Let's let the script
13:56do that. Let's let the script work out
13:59how much we're spending on video because
14:01it doesn't matter if they've got a
14:02YouTube campaign or not. We can work
14:04that out. And then whatever's left is
14:06search and display. That was version
14:08one. And then I ended up chatting to a
14:12wonderful bloke in Copenhagen. I
14:14actually met him in uh speaking at
14:16Bologna, John Jan Carlo's uh John
14:19Paulo's event uh the wonderful AdWorld
14:21experience, one of the best events in
14:24Europe. Um and Tobias was there. We had
14:27chatted maybe two and a half years ago
14:29and Tobias showed me how they had
14:32figured out how to get a rough
14:34estimation of that display spend. So now
14:37we could split apart display and search.
14:40And I said, "Oh, can I build this into
14:42of course you can go for it. It's not my
14:44thing. I just But he discovered that,
14:46not me." Um, and so built that in, gave
14:50away a free version of this 2 and a
14:53halfish years ago and still felt like it
14:57was a bit of special source um for us as
15:00an agency to to do, you know, the big
15:03version of this that we had. and that
15:04bucketed products um very similar to the
15:08what the labelizer script and product
15:10hero now do but we had sort of our own
15:11janky version of that back then um and
15:14when I sold the agency um I kept the IP
15:18for the script and so you know four
15:20weeks later I was speaking in Italy at J
15:23Paulo's event I thought yeah bugger it
15:26why not let's see if more people want
15:28this and are willing to pay couple
15:30hundred bucks for it that's all and it
15:33went gang busters Um, so I quickly built
15:35a little site and started selling it
15:38through that. Um, then built an MCC
15:41version because a lot of people like,
15:43"Yeah, this is great, but I have to do
15:45this for every single account." And
15:46every time you keep changing the script,
15:48which is all the time because I kept
15:50having new ideas and everybody kept
15:52telling me what else they wanted. Hey,
15:53could we build this in and could you add
15:55this? I'm like, yeah, here's a new
15:56version. Oh, great. Now I've got to
15:58update 50 accounts again. Um, so I
16:00created the MCC version and sold that
16:02and um, yeah, it just went it went went
16:04ballistic. It was great. Um, but yeah,
16:07it's it's the it's the community around
16:09it that I love. Like all these people,
16:10it was back to those conversations that
16:12I missed from the web savvy days. Now
16:14all of a sudden like I'm having
16:15conversations about, oh, how are you
16:17using this? Oh, that's really cool.
16:19That's really smart. And what would you
16:20like me to add to this genius idea? Hang
16:23on, let me just let me go and add that.
16:25So that's how it sort of built and built
16:27and built and we're we're up to version
16:2888 now. So keeps on going.
16:33>> It's a great story. Now if someone is
16:35listening and feels like hey I also want
16:38to get started with building
16:40automations, building scripts or maybe
16:42even building agents although like there
16:46are different uh specialtities to to
16:49these different um stages of automation.
16:51But what would you recommend someone
16:53that says, "Okay, Mike, I'm down. Um,
16:56next weekend is all like exploration
17:00time." How should you get started with
17:02building that stuff?
17:05Start with a problem. Start by thinking
17:08about what are all the tasks that I do
17:12rather than try and solve this for
17:14someone else. Think about the your own
17:15tasks that you do of those. Which ones
17:19are sort of data related? And is there
17:21anything in there that you're doing on a
17:23regular basis? Let's say more than three
17:25or four times a month. So maybe you're
17:27pulling the same data for your boss.
17:29Maybe you're creating some sort of
17:31client report and they want to see that
17:33every week or or every month, but you
17:34got a few clients and it's kind of the
17:36same, but you're doing this stuff
17:38repeatedly and you find yourself doing
17:40the same sort of things over and over
17:42again. That's the sort of stuff that a
17:45script is perfect for, assuming that
17:46it's Google Ads data. If it's some other
17:48type of data, you're going to need some
17:50other type of automation. Um, where to
17:54answer that question
17:56a year and a half ago, I would have said
17:58something completely different to what
18:00I'll say now. Because what I'll say now
18:02is just turn on your favorite AI chat,
18:05GPT, chord, whatever, and just start
18:08having a conversation with it. Here's
18:10this thing that I do and I get this data
18:12and it looks like this and I find myself
18:15doing this and this and this all the
18:16time. Like I've heard of this thing
18:19called Google ad scripts. I've never
18:20created one. I've used other people's
18:22never created one. How could I do this?
18:25What script could we create? So rather
18:27than just diving into chat GPT and
18:29saying I need a script that's going to
18:30pull this data and do this and maybe you
18:33don't know what you don't know yet. So
18:35have a conversation with an AI first. Be
18:37prompted by the AI. say, "Can you please
18:40ask me some questions so that you can
18:42figure out, dear AI, what it is that we
18:45need to build?" Great. Now, can you
18:48build that for me, please? Because guess
18:49what? It's seen millions and millions of
18:51Google Ads scripts, and it's really
18:53quite good at getting it right.
18:56Sometimes it'll get it wrong. But the
18:57wonderful thing about code, it's a bit
18:59like maths, right? There is a right
19:01answer. So, you take the script, as you
19:03well know, and you go chuck it into
19:05Google Ads and it either runs or it
19:07doesn't run. And if it doesn't run,
19:09there's usually some sort of messy error
19:10thing down the bottom. You don't even
19:12have to understand what that means. You
19:13just copy the whole thing, chuck it back
19:15into your favorite AI and say, "Your
19:17script got this error. Fix it for me,
19:20please." Um, and it does. And you
19:22usually only have to go around that loop
19:24these days with the powerful AI that we
19:27have now, maybe two or three times, and
19:29all of a sudden, you've got a working
19:31script that solves that problem.
19:33Awesome. Use the time that you've saved
19:36by automating that thing. Even if it was
19:3810 minutes a week, doesn't matter. Use
19:41that time you've saved to go build
19:43something else. I have a little
19:44calculator on my website now that shows
19:46like if we only spent 20 minutes a week
19:50building some sort of automation. And
19:52even if it takes three weeks, it takes
19:54three lots of 20 minutes to build one
19:57automation that saves me 20 minutes. So,
20:00I'm going to spend an hour building the
20:01thing that's going to save me 20
20:02minutes. It just shows this exponential
20:05chart. If you keep doing that week after
20:07week after week after week and then you
20:09plug in what's my time actually worth,
20:13then you very very quickly find out that
20:15you can save or make thousands of
20:18dollars a year depending on how you
20:21value your time or dozens and dozens of
20:24hours per year if you prefer to think in
20:26terms of time saved just by continually
20:29going, "Oh, what next?" And just and
20:32once you get the bug for this, it
20:33becomes quite addictive, right? because
20:35you just start seeing problems
20:37everywhere and I just love solving
20:39problems. I love the the intellectual
20:42part of that of like how am I going to
20:43do this? And then you just keep pushing
20:46the edge of that of like, okay, what are
20:50some more pro? Oh, yeah. Not that, not
20:51that. Oh, that's that seems like a a
20:53hard problem to crack. How am I going to
20:55solve? Oh, yeah. Okay, let's just go
20:57attack that problem. And it's just
20:59addictive. Like I said before, going
21:01from idea to to impact so quickly is
21:05just so much fun. And yeah, you don't
21:07need to learn to code. It helps to know
21:09a little bit about code, but
21:13I think it's much much easier to learn
21:15that as you go. You don't have to. There
21:17are plenty of people using scripts that
21:19have no idea about code. That's totally
21:21fine. But if you do want to sort of
21:24understand a bit about code, it is
21:26infinitely better to start by trying to
21:29create something. Even if it's like a
21:31little 30line script that Claude writes
21:33for you and you use and then you start
21:36asking Claude like what does this part
21:38of the script do and why did you do it
21:39this way and and give me three other
21:42ways to do this part and you learn by
21:44doing rather than learning the way I
21:46learned to code a long long time ago. um
21:49which is like learning if statements and
21:51for loops and and yeah variables and all
21:54this kind it was so abstract back then
21:57learning from a textbook for 6 months
21:59before you could actually write any kind
22:01of workable code now it's like give me
22:03workable code and I'll follow along with
22:06you and learn as we go is a much much
22:08more interesting way to do it but of
22:10course you don't have to learn to code
22:11if you don't want to but you just have
22:14to get the bug for solving problems
22:16>> yeah I also remember myself 10 years
22:18years ago, uh, googling all the time
22:21and, uh, hanging out on Stack Overflow
22:23and always searching for the part of the
22:26code that would help me solve this, uh,
22:28nittygritty problem in a WordPress
22:31plug-in or WordPress theme. It was such
22:33a different time, but
22:35>> and it was so painful back then, right?
22:37Because you spend I remember when I was
22:39a kid type copying out computer programs
22:42from a magazine to create some computer
22:45game. It's like 18 pages of code. And of
22:47course, yeah, one semicolon out of
22:49place. Nothing's going to run. I'm like,
22:51"Mom, you're a touch typist. Could you
22:54just could you Yeah. And she'd do like
22:56two, three pages and then go, I'm done.
22:58Back to me." D. But it was so painful to
23:02to debug those things back then. And you
23:05got these weird error messages and you
23:06didn't know what they meant. And this is
23:07before you could go to Stack Overflow
23:08and actually look it up. But now it's
23:11just so simple. It's like, here's the
23:13error. You know what that means? And
23:16even if it doesn't get it right first
23:17time, do that enough loops, like I say,
23:20two, three, maybe four times, it's going
23:22to solve the problem and write working
23:25code. That's the wonderful that's why AI
23:27is so good at code because there is this
23:29kind of binary it works or it doesn't,
23:31right? We thought that AI would take
23:35years and years to write creative things
23:37like poems and songs. Turns out to be
23:39quite good at that too because it's just
23:41language. But running code and just does
23:44it work or doesn't it work? Great. Give
23:47it the problem. It just goes round and
23:48around that loop until it solves the
23:49problem. It's great.
23:51>> Now give us a little bit of an um behind
23:54thescenes look into the text stack
23:56you're using because I mean people I
23:58think see GPT are now releasing GPT5.
24:03They have they some people I feel like
24:06it's still quite unknown have used
24:07claude. Then there's Gemini. Then
24:10there's whatever all sorts of stuff.
24:12>> Rock and Quinn and Kira and
24:15>> so what are you using? What are you
24:17using or what would you recommend? Maybe
24:18even if you're using something that you
24:20feel like you wouldn't recommend for
24:21somebody that's just starting out. So
24:23where would you start? And maybe
24:24>> I I someone was just starting out
24:26exactly. I would not recommend the tech
24:28stack that I'm using today. But that has
24:30evolved to this point. But also I I see
24:34very much Google ad scripts as just the
24:36first step on that bridge of going from
24:39ads to AI, right? Of just that's the
24:42that's the gateway drug to go, oh this
24:46is possible. And the bit that I think
24:48might be useful for your audience, the
24:50bit that kind of got me over that hump
24:52by the beginning was was this
24:54realization that code can run in
24:56parallel. So, you take the time to you
25:01you found that 10-minute task and you're
25:03sitting there thinking, "Is it really
25:05worth Mike said it's going to take me an
25:07hour, so it's probably going to take me
25:08two. Is it really worth me spending two
25:10hours of messing around with something I
25:13do not understand to save 10 minutes?
25:16Let me just try it." Okay, it's going to
25:18take 12 weeks to pay back that 10
25:20minutes, right? Because we have lots of
25:2210 minutes. But it was that realization
25:26that I've built that little 10-minute
25:28thing and that little 10-minute thing
25:29and that little 10-minute thing, and I
25:32don't have to sit there and wait for
25:34them all to run one after the other. I
25:36can cram in at 9:00 on a Monday morning,
25:38hit the button, and by 10 9 50 of these
25:41things are all complete, and I've just
25:43done a day's worth of work, and it's
25:45only 10 9 and I haven't even had my
25:47second coffee yet. Oh my goodness, this
25:50is quite fun. And then you again you
25:52reinvest that time, right? So at the
25:54beginning, yeah, pick one. I think it's
25:57I think it's very useful to be aware of
26:00a few tools. It's kind of that T-shaped
26:03marketer type idea really, isn't it?
26:05like being aware of lots of different
26:08tools but probably going fairly deep
26:12into one because you get a bit of an
26:14intuition. You get a bit of a feeling
26:15for a particular model which is why last
26:18week there was all this um uproar about
26:21GP4 going away and GPT5 replacing it
26:23because people felt that they kind of
26:25they understood GPD4 that the way it
26:28spoke to them. It sort of it had a
26:30personality if you like. I know we tend
26:32to anthropomorphize things, but it does
26:34feel like these things have a bit of a
26:35personality. You get used to it. You get
26:37used to the way of working with it. Um,
26:40and then when that disappears and a new
26:42thing comes along, people just aren't
26:44very good with change usually, right?
26:46So, pick one. Yeah, I I agree. The vast
26:49majority of people, it's, you know, most
26:52stats are made up on the spot, but it's
26:53something like 95 90 95% of people are
26:56using Chad GBT at this point. Um Claude
26:59is a very very very distant second or
27:01third and then yeah Google have their
27:03Gemini. Gemini is really good at code
27:06and you would think oh it's Google ad
27:08scripts and Google Sheets so maybe I
27:10should use Gemini but their interface
27:12like the the website where you go to use
27:16this thing is just horrendous. I hate
27:18the Gemini website. So I personally
27:22would use Claude. I love the way Claude
27:25works but also the anthropic team are
27:27just they are shipping stuff like all
27:30the time. So I now use a tool called
27:31Claude code. Um I probably went from
27:35just writing code um see when you start
27:38and you wouldn't do this today but when
27:40you start you start doing it all in the
27:42Google ads interface and it's a horrible
27:44place to try and edit code or see what's
27:46going on. Def definitely don't do that.
27:48So today, if you were starting out,
27:50you'd probably start in chat GBT because
27:52you're probably listening to this and
27:53going, "Well, that's what I use." So
27:54start there. But if you're open to
27:58change, then have a go with claude. As
28:01well, and you do that for a bit. And
28:04then when you get to the point of, okay,
28:06I want a place to save all of these
28:08scripts that I'm downloading from
28:11LinkedIn and I'm making little changes
28:13to, and I just want to build up my own
28:14little library of these. I don't just
28:17want to save them in an account. What
28:19happens if I lose access to that
28:20account? Right? You want to be saving
28:22those? Don't save them as as Google
28:24Docs. That's when you sort of graduate
28:26to using a tool like Cursor. So, that's
28:29sort of the next step on from just doing
28:31everything in a browser like in a chat
28:33GBT. A tool like Cursor is fantastic
28:36because you can save all of your files
28:37in different folders down the left hand
28:39side. Your code is in the middle of the
28:42screen and then you've got this AI agent
28:44over on the right. So instead of copying
28:46and pasting code between all these
28:48different screens, you've got everything
28:49you need in front of you
28:52and then eventually you'll get bored
28:54with cursor's limitations and that's
28:57when you might use a tool like clawed
28:59code which I use inside of cursor
29:01because I like the visual aspect of
29:03cursor and being able to see all of my
29:05files and see the code changing but
29:07claude code is just the most powerful
29:10coding tool and it's kind of almost
29:12programmable. It's hard to explain and
29:14we definitely don't need to go there,
29:15but you can basically take huge big
29:18things that you do and automate them
29:21down to a word. And so by typing a
29:24series of a few words, you can do an
29:25amazing amount of work in a very very
29:27short time. It's just incredibly
29:30powerful. It's what the anthropic team
29:32use to build Claude. So, you know,
29:34they're quite good at it. They've made
29:36it basically freely available and it's
29:38just an incredibly powerful tool. But I
29:40wouldn't start there. I would start in
29:43cloud.ai or chatgbt. Or if you really
29:45really love Google, you might try and
29:48start in the Gemini app, but it's just
29:50it's a bit ugly and it's just a bit
29:52geeky and not very nice to use.
29:55>> Um, probably you're right. the session
29:57we have today is not enough to unpack
29:59all this and go as deep as you could or
30:02maybe we should also for the audience
30:05but you already dropped uh you already
30:08dropped an important keyword which was
30:10from ads to AI because you recently
30:12launched this membership program um I
30:15think it's quite new or at least fairly
30:17new um
30:19>> yeah so actually for me like let's let's
30:23maybe talk a lot a little bit about that
30:25because I feel like you
30:26a vast amount of knowledge and you were
30:29already uh telling the audience that you
30:31like to teach people stuff etc. So I can
30:33imagine this being a really really
30:35valuable um program but what's what's
30:38the vision here? What have people what's
30:40to expect?
30:42So I picture this this journey from ads
30:46to AI. I think that
30:49I think a lot of businesses a lot of
30:51agencies are going to be just fine but a
30:53lot of businesses are going to feel the
30:55squeeze of commoditization over the next
30:572 three years. I think that for many
31:00just offering the services that they
31:02offer today maybe ad management services
31:04I don't think that will be enough in
31:06isolation and I think there'll be
31:08pricing pressure on those services as
31:11businesses sort of start to say well
31:13isn't the whole thing automated you know
31:15pmax and all these things in meta like
31:18what what do you actually do they'll be
31:20saying they still they were already
31:23saying that um so I see this sort of
31:26path to
31:29offering
31:31new services to the same clients. I
31:35mean, 15, 20 years ago when I started,
31:38small businesses were crying out for
31:40someone. If you could spell AdWords back
31:41then, you got a client. And it's kind of
31:44the same now, right? All these small
31:46businesses, they they kind of feel like,
31:47"Yeah, we know met ads. We can do that
31:49ourselves. Thanks." But this AI thing,
31:52they're confused. They're overwhelmed.
31:54They're short on time like never before.
31:57They need help desperately. So agencies
32:00that are on the cutting edge here, and
32:02you don't really need to be too far
32:04ahead of everybody else to be on that
32:06cutting edge right now, have this huge
32:08advantage to solve those problems for
32:12the same clients they want to delight
32:14and surprise today, but they're just
32:17going to have to transition their
32:19services a little bit. They may still
32:20offer those ad services, but there's
32:22probably some adjacent services there
32:24teaching those small businesses how to
32:27get the most out of AI. That the
32:28question I hear all the time is, "Yeah,
32:30but where do I start?" And so I think
32:32for those agencies, they need to start
32:34on themselves. You know, what Americans
32:36would call dog fooding. You've got to be
32:38doing this to yourself and with your
32:41team and upskilling and lifting up your
32:44whole team, completely changing probably
32:47how your agency approaches things. And
32:51it's not just about efficiency. It's not
32:52just about cramming more work into the
32:54same number of people. But it's
32:56definitely not about, oh, how can I
32:58replace all of these warm bodies with
33:00some cold, hard code and save more
33:02money, Mr. Burns. It's not about that at
33:05all. I don't want to work with those
33:06agencies, but the agencies that are
33:08like, how do we take the amazing skills
33:11that we have in front of us and use
33:13these incredible tools like AI and
33:15automation to augment all of this to
33:18make these guys so much better at what
33:21they're capable of doing for our
33:23clients. do that to yourself first and
33:27then go provide those additional
33:29services, maybe extra products to those
33:32small businesses because I think those
33:34businesses, they're not stupid. They're
33:36going to be crying out for that. And if
33:38you're going, "No, we we just do Google
33:40Ads management because we're really
33:41really good at it." And they're going,
33:42"Yeah, thanks. I'm going to jump over to
33:44this guy who can do all of this stuff
33:46and teach us how to use AI and
33:48completely revamp the way we do content
33:50and change how we do all of these
33:52things. build these little tools for us
33:54internally. We used to spend 20 grand a
33:57year on this SAS tool and this agency
34:00over here just built us a little version
34:02of that in a weekend exactly to our
34:04specs. Um it's the same with the agency,
34:07right? You don't then have to if you
34:09know how to do this stuff. You don't
34:10have to do kind of things to like 80 90%
34:14of the way you want to do them because
34:16you're trying to fit your system into a
34:19tool. Just build your own tool and now
34:21you've got it doing it exactly the way
34:23that you want and you have the skills to
34:26keep changing that and evolving that.
34:27And as your team keep coming up with
34:29great new ideas of how you can use it,
34:30you're like, "That's a great idea. Let's
34:32build that in. I'll just spend half the
34:34weekend on that." And now we've got an
34:35even better tool that everybody can use.
34:37And so that for me is that journey. I
34:39picture it as a bridge from ads to AI. I
34:42think the scripts are just the first
34:44step on there. Like I said, the the
34:46gateway drug. That's the bit that makes
34:47you go, "Oh, wow. This is fun. This is
34:50possible. I save all of this time." And
34:53I get to solve problems and and build
34:55stuff. It's figuring out who on your
34:58team. It might be you. It might be
35:00somebody three levels down. It might be
35:02somebody three levels up. Figure out who
35:04the people are in your agency, in your
35:07business, or if you're a freelancer,
35:09it's probably going to be you. Figure
35:12out who those people are and invest in
35:14them. I have this line I use on stage of
35:185% or 100%. So, think about the amount
35:22of income, top line, not bottom line,
35:23topline number for last year and take 5%
35:28of that. Or if you're a freelancer,
35:29think about your income for the last
35:31year. Take 5% of that. That is the
35:33amount I believe you need to be
35:35investing right now in your AI literacy
35:38in your education because if you don't
35:42you're risking the 100% of that number.
35:44You're risking being obsolete and being
35:46gone in two three years. So I think
35:49that's a pretty easy thing to weigh up
35:515% or 100%. But I still see so many
35:55people just sort of head in the sand of
35:56like no it's it's all a fad. I don't
35:58need to learn anything about this. It'll
36:00all blow over. I'm going to keep going
36:02the way I'm going because it's the way
36:03we've always done it. I think those
36:05businesses are going to be in trouble.
36:06I'm not interested in working with
36:08those. But the ones that are really
36:10interested in like we know we don't know
36:12but we really want to learn. That's my
36:14crew. And I'd love I love working with
36:16those people.
36:17>> And what can people expect from the
36:19program? Like how does it work? Because
36:21I mean I can imagine a lot of people
36:23thinking like, "Hey, this sounds really
36:24great, Mike. I where can I join? How
36:28much does it cost? What does it include?
36:31costs about two bucks a day. That will
36:33go up at some point. So, I I love
36:34pricing that goes up in tiers. So, I had
36:36the first 100 people got it for 500 euro
36:39a year. Um, currently it's 650 and it's
36:42about to go up to 800 for a year or you
36:44can pay I think it's about 300 bucks a
36:46month. So, it's clearly aimed at people
36:48that want to be with me for a year
36:50because I want to massively overd
36:52deliver on that year. So, I have weekly
36:55videos where I talk about tools and show
36:57off how I'm using various tools and how
36:59small businesses are using tools. We
37:01have a monthly sort of office hours Q&A
37:04webinar. Um, just had one a couple of
37:06nights ago. It was fantastic. Um, where
37:08people can come in, ask questions, maybe
37:10show off a little thing, but just get
37:12unstuck. Um, and then I'm creating a new
37:15course every quarter as well. So, I drop
37:17my courses in there. Um, they get
37:20courses before anybody else. they get
37:22the courses for free and those courses,
37:24yeah, maybe I go on and sell elsewhere.
37:25So, I've redone just recently my entire
37:28scripts and sheets mastery course for
37:302025. Completely redid that. Uh, last
37:33week I just dropped a Claude code course
37:36showing how to use Claude code inside of
37:38Cursor, which has proven to be quite
37:40popular in the past week and a half that
37:42it's been out. Um, I've got so so many
37:45plans. I'm in the process of creating a
37:47a basically an AI literacy course not
37:51just for them but for them to be able to
37:53use with their clients so that they can
37:55take it white label it pass that on to
37:57their client and say here's a course
37:59that will teach you and your team in
38:02small business world how to use AI and
38:04they just get that for free they get the
38:05transcripts they get the slides they can
38:07tweak it and change it they can use my
38:09videos or they can re-record their own I
38:11just yeah I just I love overd delivering
38:14on value. I hate feeling ripped off and
38:16so I'm obsessed with overd delivering on
38:19value. So yeah, build the agent stuff.
38:21So, you know, done a few build the agent
38:23courses. All of that stuff will be in
38:26ads to AI in the nearish future by the
38:29end of the year. That'll be the next
38:30course that I do in Q4. And then yeah,
38:33loads and loads of plans for for for
38:34more stuff to add for yeah, couple of
38:36bucks a day basically.
38:38>> It's I think it sounds really good. So,
38:40I think we'll just drop the link uh to
38:43the landing page in description of the
38:44video below so everybody can can check
38:46it out.
38:49>> Pretty easy. I found the shortest domain
38:51I could and it doesn't really matter if
38:52you type in from ads to aai t or the
38:56number two. I've got them all. I like
38:57buying domains.
38:59>> Really, really great. Really great. Um,
39:02you already mentioned um, build the
39:04agent and I also saw you doing this
39:06in-person workshop in I think it was in
39:08Amsterdam, right?
39:10>> Yeah.
39:10>> Yeah. So now we we were talking about
39:13scripts, we were talking about um,
39:16writing code, we were talking about
39:18code, automations, agents, all that
39:21stuff. Can you give us a little bit of
39:23your perspective and your mindset around
39:26the differentiation for example between
39:29automation general automation maybe also
39:32automated workflows and then agents
39:36right yeah I think so firstly the
39:39definition of agents has changed a lot
39:42over the past 12 months um there was a
39:44quote I heard once technology technology
39:46is is the stuff that doesn't work yet
39:48you know it's the stuff that's just out
39:50of reach And so the definition of an
39:53agent 12 months ago
39:55net today we look at that and go of
39:57course it can do that but an agent
39:59should be the thing that does this. It's
40:00the thing that's just over the hill just
40:02out of reach. Um I think that well when
40:07I first got into AI I'll put it this way
40:09when I first sort of went deep into AI
40:11about seven eight years ago I came away
40:13thinking you know what most businesses
40:15don't need AI at all they really need
40:18automation. and I wrote this whole
40:20business plan out for like an automation
40:21agency and then realized I didn't want
40:24to do that because automation kind of
40:26bores me. Um, and there were lots of
40:28those. Um, there's a great guy here in
40:31Australia, good mate, who's who's built
40:32the automation agency. That's literally
40:34the name of his business and it's really
40:35good and it does really good stuff. So,
40:37we just would send people his way. Um,
40:41where am I going with that? Um,
40:43automation I think is the place to start
40:47because most of the time that will solve
40:50the problem. You don't need the latest
40:52shiny object. You don't need to be like
40:55sending a whole bunch of stuff off to a
40:57large language model to get a response
40:59in text back. Actually, you kind of need
41:02something that works every single time
41:03in exactly the same way. and just
41:06looking for those opportunities, looking
41:08for things we can go automate
41:11um is a really good place to start.
41:13Really easy way to do that by the way if
41:14someone's listening. Um
41:17next time like so first of all think
41:18about the tasks that you do and think
41:20about those that maybe you do fairly
41:24often and don't require huge amounts of
41:27human judgment. So you can sort of plot
41:30all your tasks out on that scale of like
41:32really frequent to infrequent and loads
41:34of judgment to not very much judgment.
41:36Find that quadrant on that little chart.
41:39Stuff I'm doing three, four times a
41:40month at least. Not massive amounts of
41:42judgment. Think about all the tasks in
41:45that quadrant and then record yourself.
41:47Turn on Loom and record yourself doing
41:49that task and just talk your way through
41:52it. Maybe it takes you 20% longer than
41:54it usually would, but you just talk out
41:56loud as you're going through. When you
41:58get to the end, Loom will not only
42:00create the transcription of that for
42:02you, there's a button next to that
42:04transcription that turns that into an
42:06SOP or a standard operating procedure.
42:08Then you take that SOP over to chat GBT5
42:12and you turn on thinking mode up at the
42:14top and you say, "Here's a task that I
42:16do all the time. Can you see any
42:19opportunity to help automate any of
42:21this? Or are there any steps that we
42:23could eliminate or are there any steps
42:24you think I'm missing?" But basically,
42:26how can you help me automate this task?
42:30And it'll give you a bunch of ideas.
42:31Some of them will be made up and wrong.
42:33Yes, but most of the time a lot of it
42:35will be right. And if it's Google
42:37Adsbased stuff, you might even nudge it
42:39in the right direction and say like, are
42:41there any ways to use Google Ads scripts
42:43to help with any of this stuff or Google
42:45Apps scripts, which is a very close
42:47cousin? But if you're working with
42:49spreadsheets a lot, you definitely want
42:50to look into apps scripts. They're
42:52incredible. They're powerful. So, you
42:54start there.
42:56AI, you mentioned AI workflow. I think
42:58most things in AI world right now are
43:01workflows rather than truly autonomous
43:04agents. There's lots of arguments in the
43:06in the AI world about what actually is
43:08an agent and how autonomous is it and
43:10how agentic is it. I really don't think
43:12that matters too much. What matters is
43:14does it solve problems for you for your
43:16business? And we've all seen those
43:19flowcharts on, you know, LinkedIn and
43:22drop your thing down below in the
43:23comments to get the thing that gives
43:24you. Yeah. All that rubbish.
43:27Those flowcharts, something like a make
43:30or a Zapia or today's flavor of the
43:32month is N8N, which I know you know very
43:34well. Maybe one or two of the boxes on
43:38that big flowchart are large language
43:41models and AI doing its thing. So you
43:43get to that step and you take all of
43:45that information to that point and you
43:46send it off to Claude and it comes back
43:48with some text. You go great, thanks for
43:50that. And then you use the output from
43:53Claude as the input for the next step.
43:55So you just oh okay. So that's the bit
43:57where a human maybe made some sort of
44:00decision about something and that was
44:02the point where we had to go is it is it
44:03more this or or more that. Is this email
44:07really urgent or can it wait till
44:09tomorrow? If it's really urgent, then do
44:11this step next. Otherwise, let's go off
44:14on this other trail. Or is this customer
44:17review really positive, really negative?
44:19Do we need to respond to this urgently
44:22or not? Do we need to thank them? Do we
44:24need to send them a gift? So, I think
44:26thinking about it in those terms, that's
44:28exactly the right phrase that you used
44:29of an AI workflow. So, it's it's just an
44:32automation workflow, but some of those
44:34steps might be AI, but I I think they
44:38get conflated and I think that what a
44:41lot of people call AI is really just
44:44automation. And that's totally fine. Um,
44:46and more people should be using a lot
44:49lot more automation. I've heard some
44:51horrendous stats recently about, you
44:54know, just the the average number of SAS
44:57tools that a small business uses.
45:00Apparently, it's on 50 on average, you
45:02know, and they're copying data from a
45:05sheet into zero to create some invoice.
45:08And then they're copying something from
45:09Gmail into Asana and they're copying
45:12from here and it's just all copying and
45:13pasting stuff. And if you find yourself
45:16copying and pasting or if you find
45:17yourself saying things like, "Oh, that."
45:19No, look, to do that, we go over to this
45:22spreadsheet and we look over here to do
45:24like, "Look for those little phrases
45:26being used around the office or by your
45:29workmates or by your clients." Like,
45:31aha, right? that that's a really good
45:32automation opportunity because if you're
45:34going over to there to get some
45:35information and then going over to there
45:37to get some information and then okay
45:39maybe you stay being the human in the
45:42loop there synthesizing all of this
45:44together maybe we don't try and use an
45:47AI for that yet but hey it worth trying
45:50but but please please it's 2025 please
45:54don't be like going and fetching all of
45:56this data bit by bit by hand every time
45:59you do that task have everything brought
46:01to you in the format that you want so
46:04that it's always there at the ready
46:07whenever you need it and then sure keep
46:12making the decision yourself for now but
46:14also in parallel be testing that you
46:17know be testing can we use an AI to make
46:20this decision that I make at this point
46:21before we bounce it over to the next
46:23step in the automation let let's run it
46:25in parallel I'll make the decision 10
46:27times let's get the AI to do it 10 times
46:29and let's compare results
46:30And when we get to the point that the AI
46:32is kind of like 9 out of 10 good, maybe
46:35that's something else that we can get
46:36off our plate and we reinvest that time
46:38into working on the next bit and on it
46:40goes.
46:42>> I remember so well like eight years ago
46:44I was working as a student at a big
46:46internet company in Germany and I felt
46:48like okay I want to so besides all
46:50freelance work I was already doing I
46:53wanted to get an idea about how uh
46:56actual business and internet business
46:57works. I remember this colleague and I
47:00liked him a lot, but I always felt like
47:02when I was looking at his screen, he was
47:04literally copying um like impressions
47:07and click data from one Excel sheet to
47:09the other Excel sheet which was then
47:11connected to a PowerPoint um slide deck
47:14which was then manually presented to a
47:16client and like sent out. And this is
47:19exactly like even at that point, I mean
47:22AI for me was not a thing back then. So
47:26eight years ago I was like hey this
47:28can't be the case and then when I for
47:31example stumbled upon uh Zapier and felt
47:35like oh cool you can exactly what you
47:37were telling just uh right now you can
47:39connect these tools and send data over
47:41there and there and I feel like a lot of
47:44people just have to train this muscle
47:46more right this um sensitivity for
47:50>> it it's been either for me it was kind
47:53of a boring thing I could work out how
47:55to do it was boring. But but I think for
47:57a lot of people up until now, it's been
47:59very hard to do that. You've obviously
48:01got the brain for it, right? You see it
48:02and you go, "Oh, I can see what the
48:04pieces of that are. I can see how they
48:05fit together." And tools like that, like
48:08M10, there's a bit of a lift there. And
48:10and a lot of people don't want to spend
48:11the 10, 20 hours to learn that up to a
48:14decent level. It's the same with
48:16Photoshop, right? Like you've got to tip
48:18in the time to learn how to use the
48:19tool. The wonderful thing is now is so
48:23many of those tools just run on code in
48:25the background, right? So Claude can
48:27create or catch GBT if you want to use
48:29it or Gemini, bless you. Um can create
48:32that JSON file or that code that builds
48:35all of those little boxes on the screen
48:37that tells you how to automate the
48:39thing. Maybe you need to go in and tweak
48:41it a little bit. So you still need to
48:42know how to use the tool, but it's it's
48:44never been easier and it's going to keep
48:46getting easier and easier. So, I think
48:50it's about like finding the way to solve
48:51the problem that that you enjoy. I just
48:55love solving the problem. I like the
48:57intellectual challenge of that. I like
48:58seeing people's face light up and when
49:00they get the bug and it's like, could we
49:04really we can do that? Oh my god, that's
49:06so cool. And that's that's why I did
49:08build the agent because it was just like
49:10sheets were getting a little bit in
49:13straining, you know, like the the UI of
49:15sheets having to do it cell by cell. I
49:17would come up with a new version of the
49:18pmac script and that was the e bit and
49:21then I'd spend like three days fiddling
49:23around in the sheet getting all the
49:24formulas in the perfect place. That's
49:26why I kind of moved to to doing stuff
49:29inside of an app. And I I needed to
49:31build the app first before I ran a
49:33program called Build the Agent before I
49:35tried to teach that. And it's just oh
49:38it's it's just so much fun like having a
49:41a slider. Have we got a minute? I can
49:44give you example for your listeners. If
49:46you go to 80 at 8020agent.com,
49:48you do not have to sign up for an
49:49account. Accounts are free, but you
49:50don't have to sign up. I've put loads of
49:52dummy data in there so you can play with
49:54the tool just to experience what it's
49:57like. So, go to ad20agent.com
50:00and then you can use the sample data.
50:01You go up to the top and you find the
50:02keyword page. I think it's under analyze
50:04account. There's a few pages on there
50:06now and there's a keywords page and it
50:08shows you a a tree of keywords. It's
50:11just a for me a really intuitive way of
50:14thinking about campaigns that branch out
50:16into ad groups that branch out into
50:19keywords. And as you move the slider for
50:22cost or impressions, you move the slider
50:24up and it shows you, oh, only keywords
50:26that have spent more than $10 last
50:28month. Move the slider up. Oh, let's
50:29just show only the keywords that have
50:30spent more than 50, more than 100, more
50:32than 500. And so it's a really intuitive
50:34way to say, what are the most important
50:36keywords in my account? Now, yes, you
50:39can go into Google Ads and you can set a
50:41filter there, but it's just feels a bit
50:43more cold and impersonal. I just I love
50:45the the the UI and constantly trying
50:48different ways of displaying data and
50:50turning it into useful information. And
50:53then, yeah, you can try chucking AI over
50:55the top of that and turn useful
50:57information into insight into actionable
50:59insight. We're kind of right on the
51:02border to finally answer your earlier
51:04question about agents. We're kind of
51:06right on the border between insights and
51:08action. And I think when we're really in
51:12agent mode, maybe next year is actually
51:14the year of agents, we'll be trusting
51:17the agents to then take those actions
51:20and and do much more of the stuff for
51:21us. But right now, we're at the point
51:23where the AI can help us surface those
51:25insights and sort of solve the blank
51:27page problem and give us an 80% good
51:30enough, hence the name 8020 agent. give
51:32us an 80% good enough insight, but then
51:36it's kind of still on us at that point
51:37to go actually let me let me bring my
51:40human brain to bear on that and
51:42synthesize it with some other data and
51:44think about that and then I'll go take
51:46the action for now. I'm not at the point
51:48where I'm ready to trust the agent to do
51:52everything from where to go, but I'm
51:55sure Yeah, I'm sure that that's going to
51:57come.
51:59>> I think the keyword tree example is so
52:00cool. I also obviously tried the 8020
52:03agent. I also so I can just recommend
52:05everybody signing up for an account and
52:07connecting their or one of their uh own
52:10Google ads accounts or a client Google
52:12ads account. I did it and I found it
52:14really useful and I think this is
52:16something that I just want to do a plus
52:17one for the data visualization part
52:20because people I think a lot of people
52:22still feel like they have to just
52:24tussles through these spreadsheets etc
52:27and uncover stuff whereas
52:30A well done data visualization can help
52:33you see stuff that is just impossible to
52:36see from just like campaign campaign
52:39campaign some costs some conversions
52:42like it's really hard especially if you
52:44take in a dimension of time like
52:46developments over time. So if you think
52:48about like okay hey my campaigns went
52:51down the last month why did it happen
52:54probably there's something that happened
52:56either to the keywords to the ads uh
52:58maybe something like in a particular
53:00week and like trying to to track those
53:04changes down and then revert it or come
53:06up with optimizations. I think this is
53:08where people confuse their job. They
53:11think their job is to hunt down this
53:13stuff whereas the the job is to actually
53:15uncover this and then come up with ideas
53:18to solve it or what's your take on that?
53:20Do you find this reasonable approach?
53:22>> I love that. I hadn't thought of it in
53:24those terms. I it's it's adjacent to one
53:27of those things I was saying before
53:28about you shouldn't have to go get the
53:32data from there. I want the data brought
53:34to me. I want those insights presented
53:36to me in such a way that I'm going to
53:39going to get it straight away and then
53:40then we get to action and then we get to
53:42what clients really want which is how is
53:45the business going to change what's
53:46going to happen to my profit when you do
53:48all that stuff. They don't want a report
53:49with a whole bunch of insights. They
53:51definitely don't want a dashboard with a
53:53whole bunch of data. What they really
53:54want isn't just the list of actions.
53:57What they really want is like what's
53:59going to happen to my business? How much
54:00more profit is this business going to
54:02do? How many more people are we going to
54:03be able to serve? How many more people
54:05am I going to be able to hire when I
54:06grow this business? That's what they
54:08care about. And yet, you know, Avanash,
54:10who was Google's uh digital evangelist,
54:13I think was his title for many, many
54:15years at Google. He's left Google now.
54:17Wonderful guy. If you ever get a chance
54:18to see him speak, please do. He's a
54:20wonderful, wonderful speaker. But he
54:22talks about the data vomit. You know,
54:24like all these all these agencies are
54:26just vomiting data over their clients.
54:28Clients don't want that. What they want
54:30is expected business outcome. what's
54:32going to happen when we do all of these
54:35things. But yes, it it it's very very
54:37similar like have it brought to me,
54:40make the the getting of the insight
54:43way easier and way faster so that you
54:46can as you said then go do something
54:48about it, make the change, run the
54:50experiment. I think most agencies would
54:53be greatly served trying to outdo the
54:56number of experiments that they ran last
54:58month. that that should be the the
55:00leaderboard of like no we run 136
55:03experiments. That's the thing I think
55:05that clients want to hear too. Like we
55:06have a a system for running experiments,
55:09a system for always trying new creative,
55:12trying new things. Sam Tomlinson talks
55:14about this a lot in his wonderful Sunday
55:17newsletter. Like the the velocity that
55:19you can test new creative at is going to
55:22be one of the the key things that that
55:24matters. Not do you have a really good
55:28way of you know looking at engrams that
55:31helps but you know it's it's about speed
55:33really and and and we're all under
55:36pressure to be more efficient. That's
55:38that's what I love helping people with.
55:41We we will get to the point where like
55:42okay so then everybody's using the same
55:45tools and everybody's 30% more efficient
55:47than they were last year. Yeah true.
55:49That becomes the new baseline and then
55:51you just have to keep on going to the
55:53next bit. I mean, this is this is the
55:55bit I don't like saying, but it's true.
55:57This is the slowest it's ever going to
55:59be. It only gets faster from here.
56:02Sorry. This is the slowest the world's
56:03going to be for the rest of your life.
56:05But it's also this is the worst AI we're
56:07going to have. And yes, I'm completely
56:09red build on the AI. But I I honestly
56:11believe if it never got better from
56:13here, we would spend the next 5 years
56:16putting in place the AI that we have
56:18available today because businesses tend
56:20to move slowly. It's incredibly
56:22powerful. It's stupid, too, and it does
56:25weird stuff and it hallucinates. I get
56:26all of that. It's not all singing, all
56:28dancing, brilliant at everything.
56:31But your job, if you're still listening
56:32to me drone on an hour later, your job
56:35listening to this is to figure out what
56:37it can help with and what it's not good
56:39for. And you do that by playing with it,
56:41by experimenting, by testing it out. Go
56:44spend 10, 20 hours with it. every single
56:47thing that you do throughout the week,
56:49every task that you do, you're like,
56:51"Hm, how could AI help me with this? I
56:54wonder." And if you're really unsure,
56:56turn on voice mode and chat GPT. Go for
56:58a walk and have a chat with it. Ask it.
57:01Explain what you do and ask it where it
57:04thinks there might be ways to automate,
57:07ways that AI can help. I love chatting
57:10to the the So, I use the chat GBT mode
57:13on the on the phone for for a
57:14conversation. I use Claude mostly for
57:17for code, but I love that chatbt app.
57:20It's um it's quite amazing to listen to.
57:23My kids love talking to it as well. It's
57:25quite bizarre. What do you know about
57:26Mike's daughter Frankie?
57:29>> I was having a chat to it.
57:31>> I did it too. I did it too. I um had a
57:34walk in the park and just was talking
57:36like half an hour uh to to chat about
57:39some I don't know random problem or or
57:42challenge in the business. And I felt
57:44like it's it's just it also helps
57:46structuring your thoughts. And I think
57:48there's just this this key insight. It
57:51goes back to an article that a German I
57:54don't know if it is was it philosophist
57:57or doesn't matter but Hinri Kle is his
58:00name and he wrote this article about the
58:04yeah I don't know the English term but
58:05basically the idea is um that while
58:08talking you have to structure your
58:10thoughts in a way if you talk to another
58:12person you have to structure your
58:13thoughts in a way that the other person
58:15can understand what you're saying
58:18>> and if you do this and it also applies
58:20to public speakers, for example, you
58:22have this you have this um this wild
58:26idea about where you want to go with
58:28your thought, but
58:30>> like on the go, you have this energy of
58:32just your thoughts structuring like the
58:34unstructured thoughts in your head
58:36structuring around this this really
58:38cohesive message. And this this thing I
58:40just I can't stress this enough. If you
58:43have a complex problem in your head and
58:45you feel like, hey, this stresses me
58:47out. It's like so confusing. Talk to
58:49someone. Talk to a person that at least
58:52like generally gets the context. What is
58:54this? Okay, you have to tell me what
58:56this is in a second. But or talk to AI
58:58because it also helps.
59:00>> There is a concept in computer science
59:03called rubber ducking. The idea being
59:04that you have this object on your desk
59:07and many people have an actual rubber
59:09duck on my desk. I really should have a
59:10rubber duck sitting here for this
59:11example, but I have this little 3D
59:12printed metal thing. I've had it for
59:14years and years. It's one of those kind
59:16of shapes that you can't make, but it's
59:18been 3D printed. And I just have it here
59:20and I sort of sit with it. It's not that
59:21I talk to this, but yeah, you're dead
59:24right talking out loud helps. So, the
59:26idea with the rubber duck is you explain
59:28your problem to the rubber duck. And the
59:30act of talking out loud and explaining
59:32it solves like 50 80% solves most of the
59:35problem. AI is a rubber duck that talks
59:38back
59:40and it can ask you questions and yes it
59:42can be a bit sicopantic sometimes and it
59:44can blow smoke up your ass as we would
59:45say in Australia but you can also prompt
59:47it and say I want you to play devil's
59:49advocate. I want you to push back on me.
59:51I want you to think critically about
59:52this. Don't just say no to everything
59:54but think critically about this and and
59:56maybe poke me a little bit where where
59:59my argument is weak or yeah you can use
1:00:01it to roleplay. If you're going for a
1:00:03job interview, you can use it. If you're
1:00:05interviewing someone, you can say like,
1:00:07"What are they likely to say? And what
1:00:09are my questions? What what are a bunch
1:00:11of good questions for this particular
1:00:13candidate?" Not like, "Give me five
1:00:15questions that I'll ask everybody.
1:00:17Here's their CV. Here's their work
1:00:19history. Here's everything, you know,
1:00:20about our business and the role and the
1:00:22and the PD and the job description and
1:00:26what should I be asking this specific
1:00:28person?" It's wonderful for stuff like
1:00:31that. It's it's so personalized and
1:00:33tailored and yeah, having a chat to it
1:00:35to get clear around your thoughts. I
1:00:37think it's one of the most underrated
1:00:39uses of AI. I'm just in the process. I'm
1:00:42trying to build this new app just this
1:00:45week. Just been playing with it a bit.
1:00:46So, I've no idea when this comes out and
1:00:48if it'll be ready on time. Um, but I'm
1:00:50going to keep going with it even though
1:00:52it's doing my head in as a way to show
1:00:54people the power of just adding that
1:00:58little bit more context of telling the
1:01:00AI a little bit more about yourself,
1:01:02about your business or about the
1:01:03products that you sell or the services
1:01:05that you offer and just how much better
1:01:08the response can be if you take the time
1:01:11up front, that upfront investment to
1:01:15educate the AI. It's it's the old um
1:01:17what's the saying? I don't know remember
1:01:18who it was, Abe Lincoln or Benjamin
1:01:21Franklin or somebody. You know, if I had
1:01:236 hours to chop down the tree, I spend
1:01:25the first four hours sharpening my axe.
1:01:27It's that upfront investment. And I get
1:01:29that nobody has the time to do that, but
1:01:31oh my god, if you do, the payoff is is
1:01:35massive.
1:01:36>> Yeah. We also use this concept uh at our
1:01:39business that and also for our clients
1:01:40that we call a super prompt and it's
1:01:42exactly uh what you were just
1:01:44emphasizing. So we have this document
1:01:46where we just say hey this is the
1:01:48business this is the target group these
1:01:50are key demographics these are the
1:01:52products these are the services these
1:01:54are the the core objections that people
1:01:56have about the service this is how this
1:01:58differentiates and this is like it it
1:02:00depending on the client it ranges from
1:02:031,500 words to something like three 3k
1:02:07words so there might also be like text
1:02:09examples in there
1:02:11>> yes
1:02:11>> and the quality of the output that you
1:02:13get out especially like
1:02:16>> around strategic thinking it's already
1:02:18great but especially if you write like
1:02:20content for example if you work with
1:02:21landing pages and all that stuff like
1:02:23it's just it's a night and day
1:02:25difference. Yeah, it's you mentioned
1:02:28strategy, right? AI has basically read
1:02:31the entire internet. So, it's seen every
1:02:33slide deck that McKenzie and Bane and
1:02:35BCG and every consulting firm has ever
1:02:37put out online. It's read all of that.
1:02:39As you say, it is a genius management
1:02:42consultant. But if you don't tell it
1:02:45about you, about your business, about
1:02:47the stuff that you do, you're just going
1:02:48to get bland average answers because
1:02:51it's just predicting the next token.
1:02:53That's all it is. It's just a big bag of
1:02:54words and it's just pulling words out
1:02:56the scrabble bag and it's just
1:02:58predicting the next word. But the more
1:03:00context you give it, the more
1:03:02information about all of that stuff
1:03:04around you, then that helps sort of
1:03:06ground it. It helps move it away from
1:03:08that that that average has this really
1:03:10strong gravitational pull that that
1:03:12average. And if you can push it away
1:03:14from that center and push it out to the
1:03:17edges by explaining all of that
1:03:20information that you're talking about
1:03:21and drag it away from the boring average
1:03:23center, that's where the gold is. That's
1:03:25where the really interesting answers
1:03:27are, where it can be the best management
1:03:30consultant in the world, but on your
1:03:31specific business, on that particular
1:03:33problem at that moment in time for that
1:03:35person, it's incredible. But again, it's
1:03:38it's an upfront investment. It's a bit
1:03:40of a heavy lift. So I'm building this
1:03:41tool to sort of say like here's some
1:03:42dummy data again. Play with it. See
1:03:46what's possible with and without this
1:03:47context. This is the answer you get
1:03:49without. This is the answer you get
1:03:50with. Try it with different models. See
1:03:52which one works best for you. Now bit by
1:03:55bit. Add your information to that
1:03:58because nobody wants to sit down and
1:03:59write a 30,000 word document in one go.
1:04:03But you can add that bit by bit. Oh,
1:04:05okay. Now I can see the benefit of doing
1:04:07that. I'm willing to do the next little
1:04:09bit. And yeah, I use a tool Whisper
1:04:12Flow. That's part of my tech stack that
1:04:13you were asking about before. I I love
1:04:15Whisper Flow. Roads.ai if you want my
1:04:17affiliate link. Um, it's a wonderful
1:04:20tool to just you hold a button down and
1:04:22you just talk and you can get so much
1:04:25more nuance out that way. Yeah, it's
1:04:27four times faster than typing, even if
1:04:30you're a good typer. And I'm pretty
1:04:31fast, but I can just ramble a bit. As
1:04:34you've noticed today, I tend to go on
1:04:36tangents and I tend to ramble a bit. But
1:04:38just just doing that for three, four,
1:04:40five minutes gives the AI so much more
1:04:43to play with than just I'm got this
1:04:46problem and I'm typing it out and I'm
1:04:48going to stop as soon as I can because
1:04:49I'm typing and I'm not very good at
1:04:51typing and go. But if you just ramble
1:04:53for 3 minutes and just talk to your
1:04:55computer, gives it so much more to play
1:04:57with and the answer will be so much
1:04:59better. I'm so glad we got to that bit
1:05:02that you reminded me about that because
1:05:03it's just it it is night and day. I
1:05:05totally agree. And people need to
1:05:07experience that. They need to see that
1:05:09and go, "Oh my god, really? I I did not
1:05:12know. I thought it just wrote Dr. Zeus
1:05:13poems. I didn't know it could do that.
1:05:15Holy Okay, I'm in. What else can
1:05:17it do?" And that's the best question you
1:05:19want to get to. Like, what else can it
1:05:21do? Not where do I start? Just start by
1:05:24playing.
1:05:25>> Yeah. Couldn't agree more with that. So
1:05:27now to be uh respectful of your time and
1:05:30wrapping um our obviously very
1:05:33insightful, very great conversation up a
1:05:35little bit, you already mentioned speed
1:05:37and velocity uh velocity and uh a big
1:05:41struggle I hear from either friends or
1:05:44colleagues and also other business
1:05:46owners is the the speed of change in the
1:05:50AI world. So there's all the time a new
1:05:52model coming, a new feature release.
1:05:54Then there are people obviously on
1:05:56LinkedIn and you have so you have the uh
1:06:00the confidence to say yeah yeah this is
1:06:02an end and a workflow and it yeah it's
1:06:04rubbish I go past that but I feel a lot
1:06:06of people have uh fear of missing out
1:06:09they feel like maybe I should get that.
1:06:12So how do you handle this or what's like
1:06:14your mindset around all this news and
1:06:18information that's just coming at you?
1:06:22It's a really good question.
1:06:26Answer it this way. Um,
1:06:29so I love books. I've got a thousand
1:06:31books over there. I've read 80% of them.
1:06:35Um, we tend to read stuff or even watch
1:06:38a YouTube video just in case. Yeah, it's
1:06:42just in case learning. We'll read this
1:06:44book because it's come out. We've heard
1:06:45good things about it. it just in case
1:06:47maybe one day, 6 months from now, two
1:06:49years from now, I've got a problem where
1:06:51I can use some of that information and
1:06:53apply it and yay, feel better about that
1:06:55problem.
1:06:56And I think we're moving to that world
1:06:58of more just in time knowledge. So
1:07:02that's certainly what I do with YouTube
1:07:04now. I'm not just watching video other
1:07:06than all the sports highlights and stuff
1:07:08that I watch there. It's mostly like
1:07:09I've got this particular problem right
1:07:11now. I'm trying to figure out this new
1:07:13thing and I can watch something that
1:07:15solves that problem right now. And so my
1:07:18use of books has changed dramatically
1:07:20over the past two years. And so to turn
1:07:23that around to answer the really good
1:07:25question, I would say like focus on the
1:07:28problems. Don't don't get caught in the
1:07:31headlights and looking at all of the
1:07:32shiny objects and thinking that you need
1:07:34every single one and you need to type
1:07:36checklist into LinkedIn 15 times a day
1:07:38to get all of those things sent to you.
1:07:40Maybe
1:07:42instead just go like what is the problem
1:07:45I need to solve in the business right
1:07:47now? I I picture businesses like a pipe
1:07:50and there's always a thin point in the
1:07:52pipe somewhere, right? It might be over
1:07:53here and it's fat pipe over here and it
1:07:56goes thin again. It's fat over here.
1:07:57There is no point making a fat part of
1:08:00the pipe even fatter. There's no point
1:08:02taking this part of the business and
1:08:03making this part better if there's a
1:08:05really thin part of the point. It's the
1:08:07theory of constraints. It's the
1:08:09bottleneck in the business. You have to
1:08:11identify the bottleneck in the business
1:08:13and work on that. There is no point
1:08:15working on other parts of the business.
1:08:17You could have three times the leads
1:08:19coming in, but if you have one person
1:08:21calling people back, it ain't going to
1:08:23matter. You need to work on the
1:08:25bottleneck first. So then, okay, what
1:08:28specifically can I do right now to work
1:08:31on the bottleneck? And yeah, okay. I
1:08:33suppose I'm at the point now where I am
1:08:36consuming a lot of AI stuff because I'm
1:08:38fascinated by it and I want to stay on
1:08:41the cutting edge and I have a community
1:08:42of people now to serve and I feel that's
1:08:44sort of part of my job is to and I've
1:08:47always loved this. I've always loved
1:08:49running four or five hills out into the
1:08:51future and then coming back and not just
1:08:53reporting on it. never been really
1:08:55interested in that. But like showing
1:08:57what's possible and building stuff to
1:08:59say like this and making it really
1:09:01useful and really practical and then
1:09:03running off again into the future. I
1:09:04kind of live my life six months in the
1:09:06future, four or five hills that way and
1:09:08then every now and again pop back into
1:09:10the real world and go like look at what
1:09:12it can do. It's amazing. I I just love
1:09:14that. So I like consuming a lot of
1:09:16stuff. But I get it. You're wearing 12
1:09:17hats. You're running around like a
1:09:20chicken with its head cut off and just
1:09:22like you're super super busy running a
1:09:24business. Where the hell do you start?
1:09:26Pick the bottleneck and exactly what you
1:09:29were saying before. Talk it through.
1:09:32Have a chat. It may feel ridiculous the
1:09:34first time you do this because why am I
1:09:36talking to a computerized voice that's
1:09:38just a big bag of words apparently? It
1:09:40doesn't actually think. It doesn't
1:09:42really know. And yet it's actually
1:09:45really really good at that. So talk the
1:09:47problem through and like give me four
1:09:49ways. I have this thing of 141. Here's
1:09:52my one problem. Give me four possible
1:09:55ways. Make them as different as
1:09:57possible, as diverse as possible. Give
1:09:58me four different ways that we might be
1:10:01able to solve this problem. You're not
1:10:02saying like, tell me how to do it, and
1:10:04I'll follow what the AI says. Saying,
1:10:06give me four different ways that I might
1:10:07be able to explore to maybe try and
1:10:10solve this problem. And then give me
1:10:11your one recommendation. So one, four,
1:10:13one. and just pick a couple of
1:10:16bottlenecks and try that and then go,
1:10:18"Okay, Let's do it. Let's try and
1:10:21solve this problem. Let's ask AI for
1:10:23help. There's probably some automation
1:10:25in there to free up some time." And
1:10:27also, if you're listening to this, it's
1:10:28not just you. Like, lean on your team. I
1:10:31think I think it's absolutely an all
1:10:33hands- on deck moment for most
1:10:35businesses. It's it's it's got to come
1:10:37from the top. It's got to be like this
1:10:40is important and yes, we are going to
1:10:41fund this. We are going to give you the
1:10:44resources to do this. Again, back to 5%
1:10:46or 100%, you need to invest in teams
1:10:49right now, but also those people on the
1:10:52front lines, the people dealing with
1:10:53customers all day. Those are the people
1:10:55that are going to have the best ideas of
1:10:57how to implement these, they're all
1:10:59using these tools anyway, boss, sorry to
1:11:01tell you. They're just not telling you
1:11:02about it because they're worried that
1:11:03you're going to tell them off or you're
1:11:05going to fire them or they're going to
1:11:07end up creating the system that replaces
1:11:10them. They're worried about that.
1:11:11They're worried that they're breaking
1:11:12some AI policy and maybe they shouldn't
1:11:14have pasted all of that client data in
1:11:16and it's against some policy that your
1:11:18lawyer wrote for you and so they're
1:11:20worried but they are using it in their
1:11:22jobs already. They're just not telling
1:11:24you about it. I think you got to get the
1:11:26whole business together and say right we
1:11:28are taking this seriously. We are
1:11:30investing in to start with just basic AI
1:11:33literacy. We just all need to just be on
1:11:35the same page and then let's invest in
1:11:37right who's going to be that sort of
1:11:38like the advanced team who's going to
1:11:41run one or two hills into the future
1:11:43with me as a business. But it has to
1:11:46come from the top. There has to be a
1:11:47vision of where we're running to. We're
1:11:49not just brushing off in different
1:11:51directions over different hills. It's
1:11:52like here's the mountain we're going to
1:11:54climb. This is what the business is
1:11:55going to be like two or three years from
1:11:58now if we all go do this now. Now, I
1:12:00need a few like scouts to come with me
1:12:04one or two hills into the future. Like,
1:12:06who's going to come with me from base
1:12:07camp up to camp one and help build the
1:12:10trail for everybody else that's going to
1:12:12follow along behind? Like, this is the
1:12:14view from the mountaintop, but now let's
1:12:17focus on just getting from here to here.
1:12:18Let's just get to camp one and then
1:12:21let's go build that path and then bring
1:12:22everybody else up to camp one. Okay,
1:12:24this is the new version of the business.
1:12:26Okay, take a moment. Let's have a big
1:12:28party. Right, scouts, you come with me
1:12:31now. We're going to build the path up to
1:12:32camp 2 and we're going to keep on and it
1:12:35just keeps on going and going and going.
1:12:37But yeah, I I get it. Most particularly
1:12:40small business owners who are doing 12
1:12:42different jobs, it is utterly
1:12:45overwhelming. They don't have a spare
1:12:47minute, plus they're trying to be super
1:12:48parents, plus they're trying to be
1:12:49friends, plus they're trying to do
1:12:50everything else. It is really hard. So,
1:12:53just don't get distracted by the f the
1:12:55uh the shiny object. Focus. And focus
1:12:58stands for follow one course until
1:13:00successful. That's what focus stands
1:13:01for. Follow one course until successful.
1:13:04Fi figure out that thing, that
1:13:06bottleneck, fix that bit. Get that part
1:13:08of the pipe wider and then find the next
1:13:11bottleneck and don't get distracted and
1:13:14just keep on going. The game is just
1:13:16staying in the ring. You will get
1:13:18knocked down. The whole game is getting
1:13:19back up again and staying in the bloody
1:13:21ring. My mentor told me that once. He
1:13:23said, "There are many, many days where
1:13:24you just want to quit and walk outside
1:13:25the ring and just go, I'm done." No, you
1:13:28get back up and you stay in the bloody
1:13:30ring.
1:13:32>> Mike, I can't tell you this has been
1:13:34such an enlightening and insightful
1:13:36session. So actually I want so everybody
1:13:38that has listened until now I can
1:13:40actually recommend you just listen the
1:13:42whole thing again because there has been
1:13:45so much to unpack. So actually the this
1:13:49metaphor or this um image of going like
1:13:52two three hills into the future I will
1:13:56like
1:13:57If you ask my team, you will hear this a
1:13:59lot. Now,
1:14:00>> I've got a whole document I've written
1:14:02around that. Happy to send it to you.
1:14:03Link in the show notes. Whatever you
1:14:04want to do.
1:14:05>> Yes,
1:14:06>> that's that's how I picture it. I'll
1:14:07send it to you.
1:14:08>> Let's do it. So, Mike, thanks so much
1:14:11for um this insightful session. It has
1:14:15been a blast. Um, everybody that wants
1:14:18to follow you, um, see more of the stuff
1:14:22you're doing, uh, try some of the tools
1:14:24that you're doing, what's best to follow
1:14:26you, where to find your stuff. Um,
1:14:30I'll make it easy for you. Nine
1:14:31characters 8020 agent. So 8020agent.com.
1:14:37Um, all the courses are there. The link
1:14:39to the ads to AI community is there.
1:14:41That's the best place to find me. I
1:14:43don't really do social media. I'm on
1:14:44LinkedIn a bit, but I don't do anything
1:14:46else. But yeah, adagent.com.
1:14:49Everything's there.
1:14:51>> Awesome, Mike. Thanks so much. And uh
1:14:54yeah, everybody else, go explore. Try to
1:14:57use some of the concepts and some of the
1:14:59ideas and some of the thoughts Rogers
1:15:02that Mike has given us today and just go
1:15:04try it out uh today or tomorrow.
1:15:08>> Thanks very much, man. Appreciate.
1:15:09Thanks for having me. Been great fun.
1:15:11>> Thanks so much.

About this transcript

This page contains the full transcript of Build AI Agents Before They Replace You | PMAX Scripts and Automation | Mike Rhodes by Niklas Buschner, generated from the public captions YouTube serves with the video. The transcript has 14,786 words across 2,005 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.