EPISODE OVERVIEW

Duration: Approximately 45 minutes

Best For: Trapped entrepreneurs who tried ChatGPT, got generic rubbish back, and quietly decided AI was not for them

Key Outcome: You will understand how to turn the knowledge that lives in your head into systems that run without you sitting at the desk

Roy sat across from a marketing agency owner with 14 people on his books. Then he showed him what one AI system was already doing more than all 14.

THE BOTTOM LINE

You built this business because you wanted freedom. Then you ended up spending more hours at your desk than you ever did working for someone else. Roy and Simon Coton both know that feeling. The thing is, Simon has spent the last two years working out how to take everything that sits in your head, your voice, your brand, your process, and hand it to a system that does the heavy lifting. Roy walks through exactly how he used it. A book he could never start, written. A website rebuilt in a couple of hours. A month of LinkedIn posts gathered, drafted and scheduled. Two and a half days of podcast editing done in 15 minutes. None of this is about replacing the human. It is about raising the human up, so the skill you spent 15 years building reaches ten times more people while you work fewer hours.

WHY THIS EPISODE MATTERS TO YOU

Every task you used to do by hand, finding a note, editing a video, writing a post, is time you will never get back. Simon shows how that time stops bleeding out the moment you build a system once and let it run forever.

You have 15, 20, 30 years of business knowledge in your head. That knowledge is now the most valuable thing you own. This episode shows you why the person who understands their own process wins, and the person who hands it all to a black box loses.

You are the bottleneck because nobody does it like you do. Roy reframes that. Your branding person, your developer, the people with the real skill can spend all their time on that skill instead of the busywork, and serve ten times more customers. That is how you stop being the centre of everything.

Carry on doing it the old way and you keep paying 600 quid to a developer for a job that now takes two hours, you keep watching the clock, and you keep telling yourself you will get to your own ideas one day.

KEY INSIGHTS YOU CAN IMPLEMENT TODAY

You do not need to learn everything anymore. Roy could never get his book started. So he talked it. Sixty life stories spoken aloud, fed in with his own voice and writing style, and the first chapter brought him to tears because he had never heard his own story told back to him. The lesson hides in there. The thing you keep putting off because you are not good enough at it might just need a different way in.

AI is a 60 percent tool, not a 90 percent tool. Roy keeps what he calls the architect role. He knows the vision, he knows the plan, and he keeps pulling the work back on track. The system does the lifting. He stays the one who decides. That is how letting go stops feeling like losing control.

Starting now is an advantage, not a disadvantage. Simon makes the point that you can pick up 80 percent of what someone learned two years ago in a month or two, and you get all the mistakes they made for free. Feeling late is the thing keeping most owners from starting at all.

The critique matters more than the praise. Roy's single most-used move in two years is feeding in an idea and asking what am I missing, what is wrong, go and research whether this actually works. We are all biased towards our own ideas. A system that flatters you is worthless. One that pushes back makes you sharper.

Build outside the tool, not inside it. Roy keeps everything portable so he can move between Claude, Codex and others when one of them slows down or changes its pricing. If everything lives inside one platform, you are stuck the day that platform changes the rules.

GOLDEN QUOTES WORTH REMEMBERING

"You do it all in the name of freedom. And then you spend more hours at your desk than you ever did with a job." - Simon Coton

"The most powerful people moving forward are those that have inherent inbuilt context that others don't have around their industry, around their niche, around their function." - Simon Coton

"If we can raise the human up, those are the companies that aren't going to be delivering slop." - Roy Castleman

"I read my first story and it actually brought me to tears, because I'd never heard my own story in that way." - Roy Castleman

"The tools are getting easier, but the processes haven't changed. Everyone can now create a meta ad, but not everyone can create a brilliant one that returns on investment." - Simon Coton

QUICK NAVIGATION FOR BUSY LEADERS

00:00 - Introduction: Why one in a hundred people ever build a business, and what that says about you

03:30 - Doing it for freedom: Simon on the moment the job became the cage, and why your desk feels different now

07:00 - The opportunity nobody is taking: everyone talks about AI, hardly anyone in the real world is using it

11:00 - The 60 percent tool: keeping the architect role so you stay in control while the system does the work

15:00 - Roy's book: how 60 spoken life stories became chapters in his own voice

20:00 - From two and a half days to 15 minutes: the podcast and content systems Roy built

26:00 - Doing more than 14 staff: augmentation instead of replacement, and freeing your team's real skill

32:00 - Where it is heading: autonomous systems, the danger of AI grading its own work, and why context wins

40:00 - The cost question: paying 300 a month instead of 20, and why value beats price

43:00 - Conclusion: where the Agentic Academy goes next and how to join

GUEST SPOTLIGHT

Name: Simon Coton

Bio: Simon helps business owners build agentic systems that run their business. He runs a paid community of over 1,500 members and has taught more than 100,000 people online how to put these systems to work. He sits at the bleeding edge of this technology and explains it in a way that keeps the human at the centre.

Connect with Simon:

Website: https://skool.com/scrapes

LinkedIn: https://www.linkedin.com/in/simon-coton-81608b98/

YOUR NEXT ACTIONS

This Week: Pick one task you do by hand every single week, editing, posting, searching for notes, and write down the exact steps you take. That document is the start of handing it over.

This Month: Get a proper subscription and spend a day talking, not typing, your knowledge into a system. Your voice, your process, your stories. Build one thing that runs forever.

This Quarter: Look at your team the way Roy looked at that agency owner's 14 people. Ask what skill each person has, and free them to spend their time on that skill instead of the busywork. That is how you stop being the bottleneck.

EPISODE RESOURCES

Agentic Academy community (Simon Coton): https://skool.com/scrapes

n8n - workflow automation tool mentioned throughout

Claude Code and Codex - for building workflows and systems

Descript - podcast and video editing

WhisperFlow - voice dictation

Notion - connected knowledge base

Vercel, Supabase, Payload CMS - the stack Roy used to rebuild his site

Book Magic by Lucy McGrath

Key Person of Influence by Daniel Priestly

The Mom Test - on asking unbiased questions

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

READY TO ESCAPE THE TRAP?

Take the Freedom Score Quiz: https://scoreapp.atpbos.com/freedom

Discover how trapped you are in your business and get your personalised roadmap to freedom in under 5 minutes.

Book a Free Strategy Session: https://www.allthepower.co.uk/contact

Let's discuss how to build a business that works WITHOUT you.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

CONNECT WITH YOUR HOST, ROY CASTLEMAN

Roy is the founder of All The Power Limited and creator of The Owner's Thrive Method, a business coaching system for entrepreneurs ready to grow without burnout. As a certified Wim Hof Method Instructor and the UK's first certified BOS UP coach, Roy combines AI as a thinking methodology, wellness practices, and business operating systems to help trapped entrepreneurs reclaim their freedom.

Website: www.allthepower.co.uk

LinkedIn: https://www.linkedin.com/in/roycastleman/

YouTube: https://www.youtube.com/@allthepowerltd

1

Good morning, good afternoon, good evening, wherever you are in

2

the world. I'm here with Simon. Simon has helped me

3

so massively and we're going to share why. I've been

4

playing with AI a lot and I joined Simon's community,

5

the Argenti Academy online school, and yet was foundational.

6

So thank you, sir. Thank you and thank you for

7

joining me today. So tell us a bit about yourself.

8

Yeah, pleasure to be here. Thanks. Thanks for having me,

9

Roy. So I kind of got deep into AI about

10

probably 18 to 24 months ago. Probably a similar time

11

to a lot of people when they started playing around

12

with ChatGPT. And yeah, just found myself getting deeper and

13

deeper into AI, but understanding that there was this huge

14

gap between AI slop, as we call it, and content

15

that's actually produced that feels like human curated. So where

16

I sit and where my community sits is helping empower

17

business owners to use AI in their day to day,

18

but not to drop the quality. So it's to maintain

19

that high quality, increase output and yeah, use AI to

20

effectively empower you to do so. And you, yeah, we

21

speak to business owners. That's what I do. I speak

22

to business owners who go on this journey and yeah,

23

I want you to go back to the time when

24

you decided you weren't employable anymore. When could you not

25

work? So when did you want to step into your

26

own power, as it were? Oh, yeah, that's a really

27

good question. I think I've always been the kind of

28

person that I've always had my sights set on starting

29

something myself. And I'm not sure whether that's just the

30

freedom of being able to make your own decisions, etc.

31

But yeah, there's some distinct memories I have in the

32

past of when I've been told to do something by

33

other people that I felt was like, not where I

34

could add value the most. And I think it's those

35

kind of occasions that accumulate over time. And yeah, a

36

couple of those occasions where I've been like, hang on,

37

I think I could be more valuable if I step

38

out in my own thing and, and actually empower others

39

to do, to do greater things for themselves. So I

40

guess there's just an accumulation of those small times when

41

somebody tells you do it this way and you kind

42

of have your own vision on what it should be

43

done. Like, I think one of the things that really

44

inspires me about business owners, because we are a different

45

breed of people. Yeah, yeah. 1% of people in the

46

world will become a business owner. 1%. Oh, really? Is

47

it just 1%? Yeah. Yeah. And, you know, 1.6% or

48

something like that, in a given year, less than a

49

percent will actually start a company. Yeah. Wow. And the

50

difference between us and normal people, and I say normal

51

people because we're all a little bit ADHD and we're

52

all a bit. Yeah. Is that, you know, I don't

53

know how many people that you've spoken to that have

54

said, I've got this great idea for a business. Yeah.

55

Right. And they stay. Yeah. They stay in the comfortable.

56

Right. Yeah. The salary. Right. Or they think they need

57

a great idea. And actually when you look at most

58

successful businesses, they didn't start with this, like, original idea.

59

You. You of course have outliers like Uber, etc, but

60

most businesses are doing something that already exists, but doing

61

it in a better way. And like those business owners

62

already can see that they don't need something that's like

63

absolute revolutionary. They just need something that's going to add

64

value to a certain customer. Yeah. We see it. We

65

see a business owners see a problem in the world.

66

Yeah. And then know that they can fix it and

67

step out and do the fixing. And I love what

68

you said before. You did it for freedom. Right. And

69

yeah, for sure. That's the biggest motivation to get out

70

there and be free. And the problem is if you're

71

not careful, it becomes your biggest prison. Yeah, yeah, yeah.

72

Don't I know it. Yeah, it's super difficult because you

73

do it all in the name of freedom. And then

74

you spend more hours at your desk than you ever

75

did with a job. But it feels very different. I

76

don't know how you feel about it, Roy, but it

77

feels very different. It's like every moment I'm at my

78

desk, I'm actually super passionate about what I'm doing and

79

I'm trying to. Yeah. Do something to help others, whereas

80

when I was at desk for somebody else, it didn't

81

feel like that at all. It felt like I was

82

watching the clock to leave the place. And then when

83

I do want to take a day off, I can

84

take a day off. And that's. That's what's important. Even

85

if I don't, I can. Sure. Yeah, you have, you

86

have the option. And I think that's. Yeah. Such a,

87

such a, such a good point is that you feel

88

empowered all the time. Right. I'm just busy going through

89

vision statements and things with some of my clients. And

90

once you have this powerful vision. Right. That vision drives

91

you through all that hard work. Yeah. You got to

92

really keep on going to that vision, towards that vision,

93

towards that vision. And yeah, if a vision is not

94

strong enough, then you become one of the 78% of

95

businesses that fail in the first year. Yeah, yeah. So

96

that's a super powerful thing to do. Let's just flip

97

onto this new world of ours. Right? Yeah. And wow,

98

what an opportunity there is for everybody. Right? Yeah. What

99

do you. The thing I see the most, and I'm

100

guessing for you, it's the same because of where you,

101

where you look on social media, all these people saying

102

all these amazing things about AI and what it can

103

do, but when I go out into the real world

104

and talk to people, hardly anyone is doing anything. Yeah,

105

yeah, it is exactly like that, actually. Like, I guess

106

we're at the very early stages and some people think

107

because of all the hype around there, that they're late

108

to adopting this tech. But actually you can pick up

109

the foundations of this technology extremely quickly. It can take

110

you a month, two months with just a bit of

111

practical application. And you can pick up the same knowledge

112

or 80% of the same knowledge that somebody has from

113

two years ago when they were doing ChatGPT. And it's

114

almost better to be doing it now and starting now

115

because there's all that accumulated knowledge and learnings of what

116

hasn't worked in the past and what things need to

117

be improved going forwards. Yeah. I remember the first I

118

was in the game from when ChatGPT came out. Yeah.

119

And the thing that really blew my mind with my

120

accent, with my mumbling, with my not being as clear.

121

Yeah. I've worked on it, was that I could actually

122

talk into this chat window and get something comprehensive out.

123

Yeah, right. And I was like, wow, okay, great. Yeah,

124

yeah. And I think fairly fast, but I can't type

125

very fast. Yeah. So that just basically meant that I

126

always felt like I was not at my potential. Right.

127

Are you now using something like Whisper Flow to dictate

128

or. I am using Whisper Flow, yeah, yeah, yeah. But

129

you used to just Talk directly into ChatGPT, is that

130

right? Actually, probably about a year ago, I built an

131

11 labs little program to do. To do that. And

132

yeah, I've literally only just. Only just gone over to

133

Whisper Flow. It's one of those ideas. Right. I was

134

this great idea. I can just do this. And so

135

I built out a script and I gave it to

136

a whole bunch of people. And yeah. Then somebody came

137

along and is now making millions. Yeah, exactly. So many,

138

so many ideas to chase, actually. Like when, when you

139

first start out, it's really hard to, to just have

140

that core focus. But as soon as you start to

141

have that core focus, that's when you start to see

142

the accumulated learnings and growth. Right. And, and going back

143

to the point around lots of people talking about these

144

business ideas, that is the view that a lot of

145

people have. Right. Oh, I've missed that idea now. Yeah.

146

But it's not like that actually. If you can find

147

an area of that specific niche that you can do

148

a lot better than Whisper flow for example, then that

149

is a business in its own right. Yeah, for sure.

150

The, the part that you mentioned there, one of the

151

things that I spend a lot of time talking to

152

people about. Yeah. Is that we have 30 years,

153

20 years, you probably have. Yeah. 15 years worth of

154

accumulated business knowledge of the things you've done. Right. And

155

for me, AI is a 60% tool. Yeah. It's

156

not an 80% or 90% tool, it's a 60% tool.

157

And I have to keep the architect role all the

158

time. Right. I have to be powerful in that. I

159

have to know what my vision is, I have to

160

know what my plan is and I have to keep

161

on bringing it back on track again. And what we

162

can do as companies, as company owners is we can

163

allow everybody in our company to bring the human to

164

the top. Yeah, yeah. To really raise the human up.

165

Because if we can raise the human up, those are

166

the companies that aren't going to be delivering slop. Yeah,

167

yeah, yeah. What does that look like? Yeah, like it's

168

a really good point because the people who are going

169

to be most powerful in my opinion going forwards are

170

those functional knowledge owners. They understand their processes inherently. And

171

now the tech can build itself or like, like you

172

said, get to a 60, 70, 80% version. So if

173

you can manage your context and your logic, that's 80%

174

of the battle. 20% is now the tech. And it

175

used to be the opposite. It used to be the

176

hard part. And why you'd hire an expensive engineer was

177

because we couldn't build the tech quickly. And then loads

178

of tools came out to build quick landing pages, quick

179

automations. But this of wraps it all in one. And

180

now the most powerful people moving forward to those that

181

have inherent inbuilt context that others don't have

182

around their industry, around their niche, around their function. Yeah.

183

And I think as we, as we step through this.

184

Yeah. There's. This is what's going to come to the

185

top. Right, Yeah, I see it. So even Claw

186

Chat Rock Whatever I'm using. Right. The context window is

187

actually really small. It doesn't really work. Right. Yeah, yeah.

188

And one of the simplest things that I did in

189

the last year was to take probably eight months ago,

190

I took my clawed and I connected it to my

191

notion. Yeah, yeah. It's such a simple thing to me

192

now, but as I'm working through my day, I'm just.

193

Oh, yes, just do that in the notion for me.

194

Just do that in ocean for me. Just do. Do

195

this. And. Yeah. And suddenly. Yeah, my notion is quite

196

messy because there's a lot of stuff in there, but

197

I don't even have to look into it because I

198

can just say I did this thing the other day.

199

Can you help me find it? Oh, yeah, yeah. It's

200

under here. Yeah, yeah. And yeah, the. The thing I

201

hear so often as well, AI is going to make

202

you dumb. And I found it's entirely the opposite. Yeah.

203

It's just. It's just opened up my potential. It's opened

204

up what I'm. What I'm capable of doing. Yeah. Because

205

all I need to know now is what's the problem?

206

Yes. You're spending your time now on the more creative,

207

harder tasks, I imagine, than what you were doing before,

208

which was searching for where you saved that note, which

209

is the task that we shouldn't really be doing, arguably.

210

We should just be actually focusing on the next biggest

211

leverage thing for our business. And AI helps you to

212

do that because you don't need to search for the

213

note anymore. You can just say you have access to

214

my notion. Go and find the note that I wrote

215

the other day. I can't even remember the exact words,

216

but you can search by meaning. So you can go

217

and find that it was something about finances. And then,

218

you know, it goes and finds the right. The right

219

note. Yeah, yeah, for sure. It was something about finances.

220

I think I was speaking to Bob, and as we

221

go through now, like the. I'll just go through how

222

I've used your process. I used NA10. Yeah. For a

223

long time. So when I finish this podcast, I take

224

the audio file, I put it into descript, or do

225

some, you know, remove ums and R's and stuff. You

226

know, just make sure it sounds okay. Take that file

227

and drop it. Drop it into a folder and it

228

goes through. Boom, boom, boom. Basically does what would have

229

taken me two and a half days in 15 minutes.

230

Right? Yeah, yeah, yeah. And I was like, wow. But

231

it took me six weeks to build that NI10 Whiffler.

232

Yeah. Wow. Yeah. About a month ago, I realized you

233

can connect the API directly to Claude. Yeah. So now

234

I'm just saying I need a workflow that does this.

235

Go build it. Yeah. You leave and do something else.

236

And how long did it take to replicate that? 20

237

minutes now. Yeah. Does it work as consistently as N8N?

238

Like, did you have to do many iterations? Well, I'm

239

getting Claude to build the N8N. Right? Yeah. Oh, okay.

240

So it's actually just building the flow. Yeah. Yeah. So

241

I've got the flows in N8N. I'm like, okay. And

242

then I'm like, okay, I need to sort these flows

243

out. I need to do this and this and this.

244

And then I need this flow to go into that

245

flow. I need a human checkpoint here. I want you

246

to go to my notion. Like, now I got a

247

video that I've done. Let's say I'll take that video,

248

I'll drop it into a folder. Folder. It'll go off

249

to Gemini, it'll check the video, it'll give it a

250

scoring for Hook, for this, for that, the next thing.

251

And then it'll go, okay, here's the transcript. Now let's

252

get a good caption for it. Let's get a title

253

for it. Throw that all into Notion for me. And

254

then using your system, I'm now going to write, I

255

now need to run my, my LinkedIn scheduling for the

256

month. Yeah, go and go and find out what people

257

are researching or talking about on Reddit. Find out the

258

most useful things and build me out of schedule for

259

the month. Yeah, go and check the analysis for last

260

month. See what they did. Yeah. And see what's working,

261

what hasn't worked. See if the time of the days

262

are working. Come back and talk to me about it.

263

And then, yeah, you need to change this, you need

264

to change that. Okay, I change those. Off you go,

265

send it. And then I've done a month with the

266

posting. Right? Yeah. And what I'm hearing from when you're

267

talking through it is these are things you used to

268

just do manually yourself or not have the time to

269

do. Right. A month worth of posts. How long would

270

that take you to do if you were actually, obviously

271

you go through and you do the human elite checkpoint,

272

but like to gather the ideas, to pull the content,

273

to rework them into, you know, content posts in the

274

right format, ready, in the right place. Yeah, all these.

275

It's just so much time. And yeah, I actually, I

276

shared my book with you and yeah, the, the, the,

277

the fun Part of that was that I was like,

278

how am I going to do this? How am I.

279

I couldn't. I could never get going, right. Yeah. Because

280

I knew I had it in me and I put

281

some more in me, but I couldn't get it going

282

right. Yeah. And then I could talk. So I'm like,

283

okay, I'm not going to type it, but I'm going

284

to talk it. So I started talking it and then

285

I was like, okay, that's okay. Get my. Claude. I've

286

talked in six stories now. Get my, my voice, right?

287

Yeah. Okay. Oh, do I say that? Do I say

288

this? And okay, now I need a writing voice, right?

289

Yeah. So I'm like, cool, I'll have a writing voice

290

to create the new writing voice. So then I'm like,

291

okay. And I just. I spent a day just talking

292

in like 60 different life stories that I've been through.

293

Yeah, yeah, yeah. And I was like, okay, cool. Then

294

I was like, ah, I'm doing Daniel Priestly. I don't

295

know if you know him. Yeah. So I'm doing his,

296

his course on key person of influence. Yeah. And one

297

of the people in there is the lady called Lucy

298

McGrath and she's got a book called Book Magic. It's

299

hard to write a book. Okay. I have to read

300

that book, put the book into Claude. I'm like, right,

301

I need a book. I need a book magic skill.

302

Okay. Analyze this book. So I've now analyzed the book.

303

Yeah. And I have a skill. So, okay, take these

304

stories and create me a database in notion where I

305

can put the, put the stories in there. Cool. And

306

it now says, okay, this book, this story will fit

307

into this chapter. The story will fit in that chapter.

308

This is about this and about this and about this,

309

about this. And it gives me this whole, this whole

310

flip, right? Yeah. And then I met this amazing

311

guy called Dennis Ross. If you haven't heard of him,

312

you want to look him up. Yeah. He's just a

313

wordsmith of amazing capability. He's setting this call with 200

314

of us and he's like, has us in tears just

315

the way he's talking. Oh, wow. Yeah. Yeah. Just amazing.

316

And I need to learn that. I really need to

317

learn that. No, I don't. Go and find out all

318

about Dennis Ross. Find out about what his methodology is.

319

Yeah, he's got 14 point methodology. Okay. Taught me a

320

skill, right? Yeah. So, yeah. So it's my voice and

321

it's my words and my stories. Right. And the

322

same way that you'd go out. And you'd go and

323

get yourself a ghostwriter who would then write it for

324

you. You know, I've got my ghostwriter. Right. Which is.

325

Which was a Claude code machine. And this was before

326

I came on board with you. And. Yeah. So we

327

went through this process and I was like. And, you

328

know, I went to Florida. I did a book writing

329

retreat in Florida with some friends. And by the time

330

I'd finished going through this process. Yeah. I was like.

331

And I read my first story. Yeah. It actually brought

332

me to tears because I'd never heard my own story

333

in that way. Really? Wow. Yeah. And that's when I

334

knew, okay, this is doable. This is. Yeah, yeah, yeah,

335

yeah. It's like conveying those emotions in you, yourself, who

336

already knows the stories then. Yeah, yeah, that's. That's very

337

impressive. Yeah. And for me, that was this piece that.

338

Yeah. This is why it's thinking outside your brain. You

339

don't need to go and learn all this stuff anymore.

340

Yeah. We've been conditioned that we have to learn something,

341

we have to do a course, we have to do

342

this, we have to do that, and then we have

343

to practice it and get good at it. And those

344

days aren't clear anymore. They're gone. Yeah. 100%. And we're

345

not saying that you like going back to the point

346

of it's not making you dumber. What you're doing is

347

just like using your time in a more valuable way.

348

You're still extracting the key points from those different frameworks

349

and books and understanding the 80% of what you need

350

to know, but it's taking you 20% of the time

351

instead of having to sit down, read the book, extract

352

the frameworks, think how they apply to your case, and

353

what you can do is, again, spend your time on

354

quality, vetting the output. Does that framing work or should

355

I try a different framework? And therefore you put in

356

the second framework that you know and you know, you're

357

basically improving it through these. These proven bits of context.

358

Yeah. And then. So then I joined your community. Yeah.

359

I saw your YouTube stuff and I was like, yeah,

360

yeah. I really love the way that you brought the

361

human to it. Oh, thank you. Yeah, yeah, it was

362

okay. Yeah. He gets marketing, he's got an understanding. You

363

knew what an ICP was. Yeah. So many people that

364

don't even know what that is. Right? Yeah. Yeah. So

365

I joined up. And then it's ah na 10 and

366

Claude code. And I hadn't started Claude code yet. I've

367

done. I've done a little bit. I've done the book

368

part of it, but I hadn't done much of it.

369

But I. I started Claude code. Did the book thing

370

burn through $20 of credits in no time? Bent through

371

$100 of credit, burnt through $200 of credit. Right. I'm

372

gonna have to have five of these things to actually

373

do any work, right? Yeah. So I jumped on yours.

374

You did NHN first, and that was pretty quick because

375

I've been doing that. Then I default code. Then that

376

was pretty quick. And then you're like, okay, I'm bringing

377

out this agent operating system. Oh, that's what I need.

378

I need one of them. And we put that in.

379

I put it in. The day you put it out,

380

I was like, let's get it going and put it

381

in. What I love about that is that on the

382

onboarding process, it captures all of the different. Yeah. Information

383

that you need to have. Right. There's this. This process

384

that you go in and step by step by step.

385

Yeah. Your brand guidelines and your voice and your. Yeah.

386

So if you have those things, you just import them,

387

and if you don't have them, it'll. It'll ask you.

388

So I got this layer first of all of things

389

that were there. So I'm like, okay, cool. And then

390

started realizing that, because I know NH. Because I know

391

how APIs work and because if I have a whole

392

bunch of other stuff, I can do a whole bunch

393

more. Yeah. So I was like, okay, I want to

394

build my website. Right. And I'm not a web designer.

395

I never have been. Never been visual. Yeah. I can.

396

I can talk tech all day long, but how do

397

I do user experience? How do I do CRO, how

398

do I do all these things? So, okay, I needed

399

to go and find the top five people in this

400

field, this field, this field, this field, in this field.

401

So we can find the file right now. I want

402

to. Skill here, skill here, skill here, skill here. Yeah,

403

cool. And I was like, now I've been hearing about

404

Wix and I've been hearing about WordPress, and I've known

405

about these things. What's the new way of doing it?

406

So I said, well, you need Vercel, you need a

407

Supabase, you need maybe payload, cms. Okay, let's do that.

408

And without knowing anything. Yeah. Yeah. I was able to

409

go through and set this up literally in a couple

410

of hours. I had my site rebuilt and up using

411

my branding, using my voice. Yeah. Okay. I need some

412

content. Where's the content gonna Come from. I know I've

413

got 70 blogs, right. Take my VTT file from a

414

blog, ingest it with my code. Yeah. Find out, you

415

know, do my keyword research. I know what I'm doing.

416

Right. Yeah. And make me 50 blogs. Cool. Done 50

417

blogs up in. So, yeah. And since then,

418

that was three months ago. Since then it always go.

419

I've just been layering on top and layering on top

420

and layering on top. Yeah. To the point that I'm

421

now able to take a video file of something I've

422

recorded and drop it in a folder. And that does

423

all the bits and pieces. That hook should be in

424

the front and it cuts it out and puts it

425

in the front, you know? Yeah. Strips it all together,

426

puts it back in again, throws the captions on it,

427

puts some B roll on it and does everything I

428

need. Yeah. Yeah. And this is, this is what you've

429

done. And you've built that over time. Right. You've basically

430

been like, okay, what's my first goal that I want

431

to achieve? Let's get that done. And then actually that's

432

going to run forever. So then it's like time saved

433

and that continues to accumulate. So then you're like, okay,

434

I actually also want it to post on LinkedIn for

435

me. So I'm going to then focus on that for

436

a week or a day, however long it takes you

437

to build. And then you're. You want blog posts off

438

the back of it and then you want, you know,

439

it to come back and optimize the blog posts for

440

internal links and things like that. So, yeah, everything is

441

an additional amount of time saved. You've obviously got to

442

invest a little bit of time in it. But, yeah,

443

yeah. I mean, I sat down with this new client

444

the other day and he runs a marketing, a full

445

service marketing agency. Yeah. So I'm like, cool, okay, show

446

me what you're doing. And we went through a whole

447

bunch of stuff and I've got this AI system for

448

it. So I'm like, okay, show me. And then we

449

got, you know, one of the things I do is

450

look at how they're running their budgets. And so let's

451

look at the team. And he's got 14 people in

452

his team. And I'm like, okay, what are these 14

453

people doing? Right. Yeah. And this, this is a key

454

thing because what we can do now as we go

455

forward is not the same as what we have done

456

in the past. Yeah. And when I showed him what

457

I was doing, yeah, I was doing more than his

458

14 team members. Wow. And that was a real eye

459

opener for me and for him. Yeah, yeah. And he's

460

like, well, what about these people? I'm like, yeah, but

461

those people all have a skill. Your branding person has

462

the skill. Instead of spending 70, 80% of her time

463

going and doing stuff, she can spend all of her

464

time taking that branding skill and seeing 10 times more

465

customers. Yes. Oh, yeah. That's what we want to do.

466

Okay. There's this developer. Hey, let's get them working in

467

a way that they can be way more efficient and

468

bring that human to the front. Right? Yeah. And 10

469

times it. And let them work six hours a day

470

and pay them more money. That's my view of the

471

AI world in the future, is that it actually frees

472

us. Yeah. Augmentation instead of replacement. It's like, how can

473

we make somebody enjoy their work more? Focus on more

474

of the right things that add value to your business

475

whilst freeing up more time and getting more customers and

476

growing your business. Where do you see it going? I

477

mean, you're really at the sharp end of this. We

478

call you at the bleeding edge. Yeah. So

479

where do you see it going even in the next

480

couple of months? Oh, it's really hard to tell because,

481

you know, if you'd asked me this 12 months ago,

482

my answer now would have probably been way off. But

483

like, when you look at the micro things of what's

484

happening is like there are a lot of patterns in

485

new tools that have been released like openclaw and Hermes

486

that have been adopted like tremendously quickly, like at historic

487

rates. Right. Like GitHub stars are coming in at historic

488

rates for these personal assistants that can do everything for

489

you, but they're very much a black box approach. You

490

don't know what's going on underneath the hood. And then

491

you see companies like Anthropic and OpenAI basically trying to

492

replicate all the great logic, but have a user almost

493

like understand what's happening. So. And that's effectively what the

494

agent operating system that you talked about earlier is. It's,

495

it's context management. But I know exactly what's going on

496

under the hood and I teach others exactly what's going

497

on under the hood. So I see like those companies

498

are building features that make it way more autonomous and

499

way more hands off. I definitely see it being more

500

autonomous and hands off, but like you said, I hope

501

it's in the direction of not complete automation and instead

502

in augmentation. Like how do we improve the quality of

503

our outputs but also Improve the quantity of our outputs.

504

How can we 10x the meta ads that we're running

505

so that we understand which meta ads are actually performing

506

better without, you know, 10x the time Google Ads.

507

Right. I don't know how much you hate Google Ads.

508

I've gone through the process of trying to set them

509

up and get them working and doing all this kind

510

of stuff and yeah, yeah, it's painful. Yeah. Painful isn't

511

even. Yeah. And then two weeks ago, three weeks ago,

512

I was like, okay, you know, API this, put it

513

straight in. Yeah. So now that my agentic operating system,

514

your system connects to my Google Ads account. Yeah. And

515

then because I've managed to sign up for a Google

516

Ads developer account, I can go and search or do

517

the keyword research. Yes. Nice. And all of this, it

518

doesn't work unless you know who your client is. Unless

519

you know what they're saying. Yeah. Unless you understand you

520

need SEO and GEO and aeo, you

521

need all these things. Yeah. Otherwise it's not going to

522

work. Right. And it all starts with understanding what your

523

SOPs are for your customers. Yeah. The tools are getting

524

easier, but the processes haven't changed. Like the functional knowledge

525

you need to run something well. Like everyone can now

526

create a meta ad, but not everyone can create a

527

brilliant meta ad that, you know, returns on investment. And

528

that's the difference. That's where that functional or niche knowledge

529

is actually going to play a part. Because yeah, one

530

set of people can do it really well and then

531

99% of people can do it, but not well. And

532

then how do you use the tools to learn? I

533

love that about your system as well. The tool is

534

always learning, right? Yeah. It's all about like, how do

535

you feed back both like with the human feeding back

536

and also things that it can validate and test by

537

itself. How does that feed back into the learning cycle

538

so that it is getting better and isn't just getting

539

more biased towards its own view? So for example, I

540

was doing some research this week on Hermes and Hermes

541

Agent and one of the inherent problems in the way

542

it's set up is it builds out these process documents,

543

these skills automatically. So you tell it to do something

544

and then it basically creates the skill for you. There's

545

a couple of inherent problems that people have noticed with

546

that. One of the biggest ones is that it's grading

547

its own work. So it's supposed to be a self

548

improving skill. It basically goes through, creates a process and

549

then because it's research, it's reviewing its own outputs it

550

basically always tells itself you've done a really good job.

551

So you effectively get this self learning loop that's getting

552

more and more biased, potentially changing the underlying logic and

553

process because it's. Yeah, it's telling itself, it's doing well

554

rather than getting that external validation from a human or

555

from another model. And that was the. I think being

556

that we've been in this game for a long time

557

now. Yeah. People still might not understand this about the

558

large language models is. Yeah. The context is so important.

559

Yeah, yeah. What's the context? Who's your audience? What's the

560

requirements? What's the expectations? I've got a CARE framework and

561

the biggest single tool that I've used in the last

562

two years has been, right, I've given you this idea,

563

what am I missing? What's wrong? What else could I

564

do to improve it and go and do some research

565

online and, and find out if this will work or

566

not. You know, because if you go out and you

567

say to ChatGPT, Hey, I've got a great idea, I'm

568

going to sell, I'm going to sell soap on the

569

side of the road. It's amazing. Yeah. I've got people

570

that walk past you, there's at least a few a

571

day and they're going to buy my soap. You know,

572

Chat. Yeah, that's a good idea. Yeah. This is how

573

you could do it and why don't you think about

574

getting yourself a banner and. Yeah, but if you go

575

and put that the whole contest and you can get

576

much more out of it. Right. 100% and you definitely

577

need that critique as well. And that's a really good

578

way to use it is to say, you know, we

579

all have our inherent biases, we're all biased towards our

580

own ideas and you know, that's why the Mum test

581

exists as a book. Right. It's about going and asking

582

questions in a non biased way to get an accurate

583

reflection of actually what people think. And we need to

584

do the same with like Chat, GPT and stuff is

585

like we almost want it to criticize our work so

586

that we can understand the limitations and we can improve

587

our ideas. Yeah, yeah. I've had a couple of conversations

588

this week with people saying I want to move from

589

chat to Claude but Chatgpt knows too much about me.

590

Yes. Yeah. And 100% I'm like let's.

591

That's quite hectic. Yeah, yeah. It's a big barrier actually.

592

And I'm sure like this is one of the directions

593

that all of these Platforms are moving in is like

594

easy migration, because you do get stuck inside a tool

595

and what you want to be building, if you're building

596

something, is something that is almost like technology agnostic. You

597

want it just to be a portable framework. And that's

598

why it's so important to understand the underlying structure and

599

folders and logic, because that can all be ported over

600

to any tech. But if you're opening openclaw, for example,

601

everything's stuck in openclaw. You can change model, but you're

602

stuck with the framework that they've built. You can't just

603

grab those files and put them in somewhere else. Or

604

you can, but it might take a very long time

605

to unpick them all. Yeah. And I think there's another

606

shift that's happening. Claude's doing this at the moment where

607

they are moving away from allowing the

608

model to just use its own 20 bucks a month

609

or 100 bucks a month pool to do the building.

610

So you're going to be limited. And yeah, for me,

611

having everything outside of the tool means I can flip

612

around. I'll just set up a Codex. So I'm just

613

trying both things and Obsidian Brain to try and have,

614

you know, the ability to go between things because it's.

615

Remember when broadband came out? You know, you're old enough

616

for that. Yeah, yeah, yeah. And when it first came

617

out, when we first had broadband out away from the

618

dial up. Yeah, it was so fast, right? Yeah, it

619

was so fast because you were the only one in

620

the road and you were getting the full 50 megs.

621

Yeah. And. And then everybody else on the road got

622

broadband and suddenly it was backbone to normal bloody

623

dialogues. And that's kind of what's happened over the last

624

year with this, with AI. As soon as more and

625

more people get on and they need more processing power

626

now, how do we pull this back again? So we're

627

potentially in for a bit of. A bit of a

628

ride. Yeah. So you see like, feedback on models like

629

Opus 4.6, where they say, hang on, this is totally

630

regressed from where it was six months ago. And it's

631

probably because of those limitations. Right. They're buying up computing

632

power from everywhere they can and yeah, they just have

633

to sacrifice something. Yeah, yeah. And Claude just

634

did a. Did a nasty. And everyone there said, yeah,

635

sorry, we did that. We're trying to undo. Yeah, yeah,

636

100%. 100%. I mean, it's the same with the 20,

637

$20 of credits. Right. Was this where they

638

recently announced that basically. Or they've Done a few of

639

these announcements, but one of them was autonomous tasks that

640

you do in the background were on subscription plan. Yeah.

641

Here's some credits which was packaged up as, you know,

642

free credits, but under the hood it was actually, you're

643

no longer able to use your subscription plan for these

644

types of tasks. Anyone wanting to do the NAN style

645

event based workflows in the background, you're now going to

646

have to pay API credits which inherently are 10 times

647

more expensive. Like actually 10 times more expensive. Yeah. That's

648

coming out 15th of June. That's going to be hooked

649

into it. So there's going to be a lot of

650

people suddenly having to pay 15 times more for the

651

same thing. And that's another kind of thing that we

652

need to understand. Right. Yeah. I'm now building out agent

653

operating systems for all my clients. Right. And great. Build

654

them out, off we go. If I do this on

655

the basis that the clients are now able to have

656

the same staff doing 10x, then I should be happy

657

to pay for that service. If that means I need

658

to pay 300 bucks a month for it instead of

659

20 bucks a month. Yeah. Then, then, yeah. What's the

660

value that we're getting compared to the function that we're

661

doing? Yeah, yeah. And so many people I'm like, get

662

yourself a cloud subscription. I've already got ChatGPT. That's going

663

to be another 20 quid a month. Yeah. But yeah,

664

see the big picture in the value it's delivering back

665

to you. Yeah. How much more can you do with

666

that when you actually use a property? Yeah. So that's

667

a, that's a question we're going to have Austin answered,

668

I think. Yeah, I think so. Like it becomes evident

669

when you're trying to, like everyone will reach for it

670

when they're trying to do things they can't do. So

671

for example, when you said you can't like inherently by

672

yourself, build a website that's got your, you know, copy,

673

it's got a nice style, etc. So actually you'd be

674

more than happy to pay $20 a month to get

675

a website up and running in a day. But how

676

much would you have paid, you know, five, 10 years

677

ago to get a website up and how long would

678

it have taken? But yet people aren't willing to commit,

679

you know, $20, $100 a month to even just test

680

the thing and give it a trial run and see

681

how powerful it actually is. Yeah. I had a. Put

682

something out on LinkedIn the other day which was talking

683

about one of my clients, you know, going to go

684

and redo his website and he had built us the

685

HTML for it and he was going to send that

686

across to his developer and pay 600 bucks to developer

687

to put that onto his WordPress program. And I was

688

like, those days are gone, we don't do that. So

689

I put that onto Lentin and I got 48 abuse

690

answers from everyone developing. It was

691

really insane. I was like, okay, this is the problem,

692

guys. You need to get with the latest tools because

693

this is the new world we live in. Right? This

694

is where it's coming. Yeah, 100%. So let's

695

thinking about wrapping up and thinking about what you're doing

696

at the moment. The agentic operating system and I'll share

697

this whole link below for people to come and join

698

you. Where do you the next six months? Where's the

699

agent academy going? Yeah, so we started like the agentic

700

operating system, as we've talked about a couple of times,

701

is basically how do you manage the right context into

702

the models at the right time? So you can compare

703

it to when you used to use something like ChatGPT

704

and it basically everyone told you a better prompt is

705

when you include examples. We're doing that at a larger

706

scale. We're including your brand voice at the right time

707

because it's a better example of how you write. We're

708

including the visual identities at the right time because it

709

knows your fonts, your colors, et cetera. So everything feels

710

like you. So we've kind of built that infrastructure around

711

pushing in the right context and the right examples at

712

the right time. The next level for us is building

713

out those process documents. So you know, you mentioned LinkedIn

714

posts as an example. A lot of businesses need to

715

post on LinkedIn. Some posts on Instagram, some posts on

716

Twitter. Like all of Those have an 80% version that's

717

like the same out the box. You need to gather

718

research, you need to actually refine that in your voice

719

into content. You need to create visuals and then you

720

need to post them. That's like the 80% version. And

721

then the 20% version is the bit you customize to

722

make it slightly more nuanced to you. Your visual is

723

slightly more yours. So what we're doing is building systems,

724

and we call them skill systems that combine multiple of

725

these processes into a box that you just one line

726

in store and basically you then have LinkedIn carousels straight

727

out the box. One line in store you can post

728

from that day, basically. So we're focused on and we

729

did a vote recently inside the community what skill systems

730

to build out. So we're focused on going down the

731

priority list of what people want built out. So that's

732

social carousel systems. We've already released a version of that.

733

We we've got like meta ad creation, and we've got

734

a bunch of those blog creation. A lot of these

735

universally recognized good things to do to grow your business

736

and grow your traffic. Kind of the heavy hit is

737

the top things we focus on next. Magic. Well, thank

738

you very much for joining. It's been amazing chatter and

739

I'm looking forward to more great things in the community.

740

Yeah, thank you so much for having me. Appreciate the

741

time. Thanks for listening, guys.