AI Agents: The Most Valuable Skill You Can Learn in 2026 (Full Course)

0:00Every company needs to put AI at the

0:02forefront of all their operations

0:03[music] or they're going to lose to a

0:04company that does.

0:05>> That's Remy. He's a young non-technical

0:07[music] founder who's a master at

0:09explaining AI in a way you can actually

0:12understand [music] and implement.

0:13>> In order to make the most out of these

0:14agents and make them actually useful,

0:16you need to treat setting [music] them

0:18up like onboarding a real employee. And

0:20you do this through context, tools, and

0:22skills. He shares his screen and shows

0:23us exactly how [music] he builds AI

0:25systems, timeless principles you can

0:27steal, and the easiest way to get

0:29started today.

0:30>> If you're starting from zero, this is

0:31all you need. Just this episode, [music]

0:33and I think you're set.

0:44Guys, in this episode, Remy is going to

0:46demystify everything that's going on

0:48with AI, the terms, the tools, the

0:50models, everything. an a toz guide to

0:53know what's going on today in AI. He has

0:55spent the last two years living in this

0:57stuff and he is about to embed in your

0:59brain all things AI that he has learned

1:01in the next couple hours. He's not some

1:03technical engineer either. For the last

1:04couple years, he's been running an ad

1:06agency. This is all self-taught. And now

1:08he's running the AI agent team for one

1:10of the fastest growing ecom brands in

1:11the world today. And I know this AI

1:13stuff is very overwhelming. So, I want

1:15you all to take a deep breath and by the

1:18end of this episode, you're going to

1:19know exactly what the next steps are.

1:21Let's get into it. So,

1:22>> I guess let's just start with, you know,

1:24what are the kind of the two stages of

1:27using AI.

1:29>> So, stage one was chat, right? You'd be

1:33pretty hardressed to find anyone,

1:34especially founders and business owners

1:36that aren't using chat models. So, think

1:38like chat GBT or Claude Chat or Gemini.

1:42But stage two is now agents. So we've

1:45moved from chat to agents and the

1:48founders and employees that are now

1:50using agents are like no word of a lie

1:52like five to 10 times more productive in

1:53their day.

1:54>> Probably more probably more than 5 to 10

1:56to be honest with you.

1:56>> Probably. I mean even if you take like

1:58the the middle mark which is like seven

2:00times more productive. Think about that.

2:02That's a week in a working day and like

2:045 to seven weeks in a week. And then if

2:07you stack that up over like a month a

2:08year like it's toast. So, you're

2:11starting to just get this massive divide

2:12opening at the moment with the people

2:15who are using agents and the people who

2:16are stuck using chat and you want to be

2:19on the agent side.

2:21>> Yeah. I mean, it's unfortunate for some

2:23of the people listening out here that

2:24may or may not be employees, but like a

2:26big thing we talk about is just opex,

2:28your actual labor as a percentage of

2:29revenue. I think everybody is racing to

2:32just get this down the percentage of

2:34revenue. And I feel like people are

2:36doing it with just chat and they don't

2:39understand the true multiplier is

2:41actually with the agents and

2:43>> even like you're saying about it's tough

2:44for employees. I actually think that

2:47there's a great opportunity for

2:48employees right now which we'll touch on

2:49later. But I think there's a there's an

2:50amazing opportunity for employees that

2:52take the initiative to learn this stuff.

2:54You end up becoming like a 100x

2:56employee. And it's like you can now like

2:59one like go to market strategist can now

3:01take on like the work of a 10 person

3:02team and get paid for probably about

3:04five five person team now.

3:06>> Yeah. We're gonna dive deep into what I

3:08think and what you think the future of a

3:10team looks like. What are the necessary

3:12skills? What does that brain need to

3:13look like? And like I mean dude, we're

3:15seeing like I think there there was a

3:16company that was a billion dollars that

3:18was one person. Like I truly believe

3:20there's going to be one to three person

3:21teams doing billions and billions.

3:23>> Agreed. because there's so much of your

3:25time and day that's like taken up with

3:26just like stupid manual tasks. Like I

3:30give you an example yesterday someone

3:31from my team was asking for like some

3:32passwords and API keys and it's like

3:34before I would have had to go into one

3:36password like find them create a share

3:38link share it securely and like send the

3:40Slack message. So instead I just got my

3:43agent to just like go into one password

3:45like create the share links send them on

3:46Slack and it's like that would have been

3:49maybe like a 10 minute task

3:51realistically but it's like just 30

3:54seconds and it's like it just allows you

3:56as a founder to spend your time on the

3:57work that's fulfilling for you and

3:59creative and strategic the work that AI

4:01like can't do like the visionary stuff.

4:03>> Yeah, for sure. Let's dive into, you

4:05know, what does that agent look like?

4:07Let's really really deep dive into it.

4:08For anybody out here that's listening, I

4:10highly recommend that you either watch

4:11this on YouTube or Spotify so you

4:13actually can see the visuals, but let's

4:15break down, you know, agent a bit more

4:18and like what that kind of loop looks

4:19like.

4:20>> 100%. So the word agent gets thrown

4:22around so much. It's kind of like lost a

4:24lot of meaning now. You you see like if

4:26you open Instagram or X, it's just I

4:28build an agent that does this, agents

4:30this, agents that. But the easiest way

4:32to like actually think about like what

4:33an agent is is I like to think chat is

4:37just question to answer and an agent is

4:39just goal to result. So question to

4:42answer goal to result is just the simple

4:44way to think about it in your head. So

4:46for example with a chat you might ask it

4:49a question it'll reply and then you have

4:51to go and do the work. Whereas agents

4:53you're giving it a goal it's breaking

4:55that down to a plan and executing and

4:57then handing you like a finished

4:58product. And the word agent actually

5:01just comes from the word agency which

5:03means like ability to act and do things.

5:05And people think like agent as in like

5:08real estate agent or like travel agent

5:11as like a person like a guy that's like

5:12a sci-fi type guy that's like living his

5:15own life and then doing work for you.

5:17But in reality it's just like comes from

5:18that word agency which was like aentic

5:21AI which meant like AI that can do

5:23things for you and then people just like

5:25now call them agents. The way I think of

5:27it is like chat models is like a really

5:29brilliant consultant. Like you can go

5:31ask it advice and it will like give you

5:33give you the the answer but you have to

5:35actually go and take that answer and do

5:36the work. Whereas an agent's like hiring

5:39that brilliant person in house that will

5:41then go and do it for you too.

5:42>> Yeah. End to end like you what do you

5:45want? What does that result look like?

5:47They figure out kind of the workback

5:48schedule and then literally like when

5:50you press press go and we're going to

5:52talk about underneath the hood like what

5:53it looks like to obviously the piping to

5:54make that work. You just see it just

5:56typing and working and it just gets the

5:58end result.

5:58>> Yeah. Because I would argue that chat

6:00models question to answer like can't

6:03actually even take any meaningful work

6:04off your plate. Like maybe a research

6:06task, but realistically like chat models

6:08can't actually take work off your plate.

6:10>> Yeah. Let's talk about like a couple of

6:12those things. Let's like kind of

6:13categorize like stuff that maybe does

6:14make sense in chat. Like for me it'll be

6:17just like, you know, I'm staying at this

6:19hotel. I'm going to studio and then I'm

6:21going to

6:23book this hotel or this hotel and I'll

6:25screenshot a bunch of them and it'll

6:26just kind of optimize from a distance

6:29perspective how I get to places or like

6:31should I order this piece of food, this

6:32piece of food like little simple things.

6:34>> Yeah. Like yesterday I was on the subway

6:36and I took a screenshot on Google Maps

6:38and sent it to like Claude in just the

6:40chat and I was like yo can I pay Apple

6:43Pay on the subway cuz I'm from Australia

6:44and it was just like yep you can. I'm

6:46like sweet. Like just little questions

6:47to answers,

6:48>> little simple things, you still should

6:50use chat.

6:50>> Yeah. Yeah. Agreed. Cuz like sometimes

6:51using an agent to ask like how many

6:53calories in a banana is like killing a

6:55fly with a bazooka. It's like overkill.

6:57Like you don't need an agent for that.

6:58So chat is still like very useful for

6:59me. Like I just got called on my phone

7:01and like asking questions like that

7:02throughout the day. But any meaningful

7:04work

7:04>> straight to the agents.

7:05>> Yeah. So, I mean, like a good way to

7:08illustrate kind of this old way of

7:10working with chats versus this new way

7:12of working with agents is like a

7:14marketing example because obviously like

7:15I used to run an ad agency. It's kind of

7:17used to be my bread and butter. And if I

7:20was going to create a marketing

7:21campaign, right, let's just say you're

7:22like doing a Christmas in July campaign

7:24for ecom brand. The old way was you'd

7:27probably have the brief in notion or

7:29whatever project management tool you're

7:30using. You would copy that Christmas in

7:32July brief, paste it into chat GBT. Yo,

7:35we're doing this campaign. Like, create

7:37me some ads. Write me the email copy.

7:39You would then like take that copy and

7:41like paste it into Clavio or then like

7:42give it to your email designer. Get the

7:44ads and then like go into some other

7:45tool and make them. And you'd copy the

7:48content from chat into your content

7:50calendar, paste it, and then you would

7:52have to go into like ads manager and

7:54like manually input it all as well. Like

7:55once

7:56>> switching cost and a ton of like small

7:57optimizations. It's it's going to two,

7:59three, four, five different people.

8:01>> And it's back to the thing about like

8:02more fulfilling work as well. It's like

8:04when I used to have to upload if you

8:05were uploading like a 100 ads to ad

8:07manager like the UI in meta ads manager

8:09is so awful. It would take you like an

8:12hour of like just data entry like

8:13putting them into ad manager and

8:15switching off all like the creative

8:16enhancements and all the all the

8:17[ __ ] So the new way of working

8:20would literally be you just tell Claude

8:21or whatever agent you're using

8:24kick off the Christmas in July campaign

8:26and it would go into notion. It would

8:27find like the the campaign outline about

8:29like the plan and like what specific

8:31discount you're going to give and the

8:32offer and stuff and it could go into

8:34Shopify like pull in actual product

8:36images. It have like your brand

8:38guidelines so the email would be like

8:39nice and branded and then you would like

8:43write the ads. You can even use a tool

8:44like Higsfield or an image gen to create

8:46the static ads and then it might just

8:48give you back like five emails and 100

8:49ads and you pick like your favorite

8:51email and like top 10 ads and then you

8:54approve and then it'll just literally go

8:55into Clavio, set it up and then go into

8:58ads manager, set it up and like it's

9:00done.

9:00>> Yeah.

9:01>> And and it could take it there and just

9:02leave it as a draft and obviously you

9:04can by hand optimize it before you click

9:06go. And for everybody out there

9:08listening and watching just to give you

9:09context like we're talking very macro

9:11right now. We are in this episode going

9:13to actually show you live how to do

9:15stuff like this because there's a lot of

9:16there's a lot of stuff like in between

9:18there to make sure that the piping is

9:19set up properly to get that end result.

9:21But yeah, man, I just feel like all of

9:23like the little small work is going to

9:24be absolutely done if and when you set

9:28up the proper piping. It's really just

9:29comes down to the proper piping and

9:31systems that you have underneath all of

9:33this.

9:33>> Yeah. Yeah, pretty much. So, you've

9:35given the agent a goal. It's it's done

9:37this amazing work for you and given you

9:39the output, but like what has it done?

9:40like what happens in that mysterious

9:42agent step where it's going and doing

9:44the work. I think that it's really

9:45really important to understand like

9:47what's going on under the hood here. And

9:48I'm just going to do a high level cuz

9:49I'm I'm not a technical person myself

9:51and I'm not going to lose anyone with

9:53like going into like the actual software

9:55engineering part of it. But I just want

9:57to touch on this loop that agents go

10:00through. So when you hand an agent a

10:02task, it's literally just going through

10:05three steps. observe, think, act over

10:08and over and over again in a loop until

10:10it has completed the task and then it

10:12hands you the final result. So when I

10:15say observe, think act, you can think of

10:17it as look at the situation, decide the

10:20next step, do it, look at the situation,

10:23decide the next step, do it over and

10:24over and over again.

10:25>> Optimizing, optimizing, optimizing.

10:27>> And then finally, it will output the

10:30result when it determines that the task

10:32is done. And then how it determines when

10:34it should stop going in loops is based

10:37on what you've told it in the goal. So

10:39if you for example said like to an agent

10:42as a simple example, research the top 10

10:45business podcast I should go on and put

10:48it in a 10page PowerPoint deck.

10:51The agent will not stop going in loops

10:53until it has 10 business podcasts in a

10:56PowerPoint to give to you. So if you're

10:58really vague with like what you want the

11:00final output to be then um be a little

11:02bit more subjective based on the agent.

11:04It's not going to be exactly what you

11:05want.

11:05>> Yeah. Exactly. So, it's important.

11:06That's one thing we'll touch on is

11:08clearly defining like what done looks

11:10like.

11:10>> Yeah.

11:10>> But, you know, when you're giving it a

11:11goal, it's just going in these the these

11:13three steps. Look at situation, decide

11:14the next step, do it.

11:16>> And I thought what would be cool just to

11:18like sort of illustrate this loop is to

11:19just give a simple example into Claude

11:21and I can show you this loop going in

11:23real time. Before we we show that to

11:25everybody out there, I think a key thing

11:26to touch on just because guys, I want to

11:28make sure that you're following along as

11:30we go in that observe loop that we're

11:32popping up right now. You see the

11:34context and skills. We are going to get

11:36very very deep into how to set up the

11:38context and the value of the skills.

11:41>> Yeah. Let me give this little example to

11:42Claude so I can show you this loop in

11:44action. Okay, so I'm in Claude Code

11:46right now and we're going to just paste

11:48this in.

11:48>> Welcome back, Rem Dog.

11:50Before you dive into this right now,

11:52guys, this is going to be awesome. you

11:53guys are going to see the whole entire

11:54loop just so people have like context. I

11:56don't even know if you even maybe want

11:57to zoom out from a macro perspective or

11:59we can do it after the fact of just like

12:02driving around Claude and just telling

12:04them like this is code, this is code,

12:06this is chat. Let's just make sure that

12:07we do that before the end of the

12:08episode.

12:09>> Yeah. Yeah, of course. And like I'm I

12:10want this episode to be core principles

12:13timeless. So I'm not going to go too

12:15deep into like right now the best model

12:17to use in cords opus 4.8 or Fable 5, but

12:20in 2 months time that's going to be

12:21different. But these core principles

12:23that I'm going to cover today about like

12:24how to make an agent actually useful and

12:26work for you and like how to build

12:28agents essentially that's not going to

12:30change for months. So for this example

12:33what I've told is I've said please give

12:35me a report of my best performing

12:36Instagram videos and a full breakdown of

12:38the angle the hook topic title hooks and

12:42why the video worked. I've given it my

12:43Instagram handle and I've said create a

12:45HTML report. So that's just going to

12:47keep going until it's done an analysis

12:49of my best performing videos. I could

12:51have even give it a number like 10 best

12:53performing and then it's going to

12:54deliver the HTML report and that's when

12:56it will stop the loop and give it back

12:58to me. So

13:01the first act step that it decided to do

13:03was go and search my computer to see if

13:05it can find more information about like

13:07title hooks and hooks and some of the

13:08things I've talked about. And it's then

13:11found the title hooks are exactly what I

13:12need. So now it's gone back to the

13:14observe step. Now it's evaluating the

13:17situation after the first loop. So

13:21second loop round. Now it's going to go

13:23and read. It's found some files. Now

13:25it's evaluating the situation. Great.

13:27I've found more files. What should I do

13:29next? Let's look at these files. So you

13:31can see it's read these two files. Back

13:33to the observe step. Great. This is a

13:35gold mine. There's a complete data set

13:36of 60 reels matched with play counts.

13:40And then it has I mean I think maybe I

13:43have my video my real performance

13:44already downloaded on my computer from

13:46another project that it's found. If it

13:48didn't find all these 60 reels with

13:50data, it would have instead of that, it

13:53would have gone on the loop and then

13:54probably called the tool to go search

13:55Instagram essentially. So, second loop

13:59round, it's it's read the files. Now,

14:01it's like, excellent, this is a gold

14:02mine. It's read some more files. Another

14:05loop. Now, it's gone. I have a complete

14:08real data set. All 60 reels matched with

14:10play counts, engagement rates, dates,

14:12patent tags, all that stuff.

14:14And then it said the one field that

14:16genuinely isn't in this data is the

14:17spoken hook first 3 to 5 seconds.

14:19>> That's the bottleneck.

14:20>> Yeah. So now it's gone. It's thinking

14:23it's like well we need to go find that.

14:24So now it's going to run a command and

14:26it is going to go launch a sub agent

14:29which we'll touch on later which isn't

14:31as crazy sci-fi as it sounds. It's just

14:34basically it's launching another version

14:36of itself to just go check. And then

14:40it's back in the observe step. Great.

14:42Now we've got it. The data is rich and

14:43complete. The spoken hook retrieval is

14:45running. Let me build the full report.

14:47And now it's running another act step to

14:49build the report. So, you know, we've

14:52got this loop here that the agent runs

14:54through. And a application or or a

14:57program that facilitates that agent loop

15:00just allows that agent loop to run is

15:02known as an agent harness. And then all

15:04of like the top platforms you see online

15:07now that you hear about like Claude

15:08Code, Claude Co-work, Codeex, Manis,

15:11Plexity Computer, those are all just

15:13different agent harnesses. They're just

15:15different applications to facilitate

15:17this loop. So when people talk online

15:20about I built an agent that does this,

15:24that would imply they've built a

15:26competitor to these like they've built

15:27their own cord code. They're not

15:28building

15:29>> probably not.

15:30>> Yeah. Yeah. They're not they're not

15:31building a loop. Like the better word to

15:33describe it is they tuned an agent. So

15:36out of the box they all go through this

15:37loop, right? But we can actually we have

15:40four levers to pull to make these agents

15:42better, which is kind of like the whole

15:44concept of this episode. We're going to

15:45just do a deep dive into each of these.

15:47But at the observe step, that looks like

15:50context, which is the information you

15:51can give it, and then skills, which is

15:54the processes that you've given it that

15:56it can follow. On the think step is the

15:59LLM model. So obviously these agents

16:02that step where they're you know

16:04deciding what to do the brain behind

16:06that is the large language model which

16:08you like the large language model is

16:10like GPT 5.5 or opus 4.8 like the actual

16:15model itself and then on the act step

16:18which is giving it more tools. So like

16:21out of the box like Claude Co-work or

16:23Claude Code or most of these agents,

16:24they have like a few tools. One of them

16:26being like a web search, but if you want

16:28to then have the agent like handle your

16:31email inbox, you need to tune that act

16:34step and give it more tools.

16:36>> Yeah. I think what's really interesting

16:37about that is it's almost like sports or

16:40like F1. Like you could have a

16:42Lamborghini and then you can have a

16:44Volvo. Those Let's just look at those

16:46and make the analogy that those are kind

16:47of the LLM models.

16:48>> Yeah. The Volvo could beat the

16:50Lamborghini if the context, the skills,

16:54and the tools are really [ __ ] tuned

16:56and super super dialed.

16:57>> If you put like a V12 engine in

17:00>> I'm not a car guys. I don't know if that

17:01was I don't know if that was the right

17:02analogy, but it's just like

17:03>> well like this this is like one of the

17:04analogies I love to give is it's like

17:07today in this episode I mean you've

17:09already started but you're learning how

17:10to drive.

17:11>> Yeah. And all of these different tools

17:13on the market like Manis, Perplexity,

17:16Claude, Codeex, you can think of them as

17:18all just different cars. Like Codex is a

17:20Ferrari, like Claude, Claude Code is a

17:23Range Rover, Manis is like a G Wagon.

17:26And as long as you know how to drive,

17:28which is like just the the core

17:29fundamentals, which you know in driving,

17:31it's like steering, brakes, accelerator,

17:34you can pretty much hop in any of these

17:36cars and within 5 minutes like figure

17:38out how to work it. you know, like

17:40there'll be slight differences. Like the

17:41indicator might be on the other side or

17:43like one might have another cool feature

17:44like cruise control that another

17:46doesn't, but essentially they're all

17:48just the same thing and as long as you

17:50know how to drive, you can pop open any

17:52of these and and figure it out and start

17:54using it efficiently. Like

17:55>> hot take there. Everyone is just so

17:58bullish on a specific LLM model, but

18:00we're going to prove otherwise today.

18:01Let's dive into what does that loop look

18:03like right now on the IG side.

18:05>> Let's check on this on this little

18:07thing.

18:07>> Let's check on old boy.

18:08>> All right. All right. So, we can see we

18:09got a nice report here, which is what

18:10actually works on AI with Remy. And you

18:12can see it's analyzed all 60 of my reels

18:14with the top one being like over mill

18:16views. And we can see that it's like

18:18categorized them. It's like kind of put

18:21them into into categories. It's done a

18:23breakdown of each one with like the

18:25title hook like on screen like

18:27>> And if you guys look at this too, I

18:28mean, you can take direct inspiration

18:30from this. Like you see the right verse

18:32wrong way to use Claude. You can do

18:33another video. It says right verse wrong

18:35way to do X. It's saying don't ask

18:37Claude for business advice. Obviously

18:39contrarian in nature. You could just say

18:40don't ask X or don't do X to kind of you

18:43could literally just take all of these

18:45to make all of your new ideas.

18:46>> Exactly. I mean this was just a report

18:48to illustrate this example. Straight

18:49away I can think of a use case here.

18:50Like if you do content and brand deals

18:52like you might want a report like that

18:53generated for videos you did for the

18:56brand. So they can have like a a report

18:57of how the video performed for them as a

18:59company. If you're doing brand deals and

19:01stuff. There's so much stuff you can do

19:02here. You can see it's just gone through

19:03analyzed all my videos and like picked

19:05them apart like why it worked, what

19:07didn't. We can see here all 60 reels

19:09ranked by plays with the date, title,

19:11hooks, the durations, and then

19:14throughout this process, right, like I

19:16started to illustrate before how it's

19:17going through loops about like, you

19:19know, it's observing. It's decided to go

19:20find some files. It's fed that back in.

19:23Now it's decided like, all right, go do

19:24this. It it it couldn't find some data

19:28like the first spoken 5 seconds of the

19:31videos. So then it had to like call like

19:33make another couple loops to go find

19:35that. And hard to say, but I reckon this

19:37probably would have ran through about

19:38like 20 or 30 loops before it decided

19:40that the task was done. And you can see

19:42here it worked for like 20 minutes or

19:44whatever it was. So it just it will just

19:46keep going until it's decided like, "All

19:48right, this is done. I'm ready to give

19:49it to Remy."

19:50>> Yeah, it is amazing. And for anybody out

19:52there listening, you know, if you're

19:53running a company, if you want a report

19:55at 7:30 a.m. in your mailbox every

19:58single day, a beautiful PDF looking at

20:00social media analytics, e-commerce

20:02analytics, anything. Yeah, we're going

20:04to walk through how to do this. But it's

20:05crazy how granular this can get on, you

20:08know, that individual prompt. Quick

20:10second, Remy has shared so much in this

20:12episode. All of his principles and

20:13frameworks we put into a free PDF

20:15playbook. You can download it below,

20:17link it in the description. Enjoy the

20:19rest of the episode. It's really about

20:21what's what's underneath this and

20:22obviously the piping. So let's get into

20:25how to actually, you know, you call it

20:26tune to to optimize the agent to make it

20:29useful. So why don't you kind of walk us

20:31through

20:32>> those three big pillars and we're going

20:34to deep dive into each.

20:36>> 100%. So

20:38>> a raw like agent out of the box like if

20:40you open up Claude Code or or Co-work or

20:42any of these popular agents on the

20:43market and just like give it a task like

20:45it's they're really capable. probably

20:48not going to do it in the way you like

20:49and you're probably going to get an

20:50output you're not happy with. And that's

20:52why a lot of people like start using

20:53these tools because it's all the rage

20:54and then they just quit because it's

20:56like, oh, it didn't work for me. In

20:57order to actually make the most out of

20:59these agents and make them actually

21:00useful, you need to treat setting them

21:02up like onboarding a real employee. So,

21:05you can think of agents as like really

21:06really capable strangers. Picture that

21:08dude, you know, like we've all got that

21:10friend that's like really really smart,

21:11like good at solving problems, and we'll

21:13just like keep working until like the

21:14task is done. you now have the

21:16opportunity to hire this guy into your

21:18business for $100 a month in salary,

21:22>> $1,200 a year. Sign me up.

21:23>> And and he has infinite bandwidth. He

21:26can just like like there's no limit to

21:28what he can do. But no matter how great

21:31he is and smart and capable, unless he

21:33fully understands you in your business

21:35or if you're an employee, your role and

21:37what you do, unless he has the tools he

21:39needs to actually do the work, like an

21:41email account, a browser, a computer,

21:44and lastly, unless he's trained on like

21:46your secret source that makes your

21:48business unique and your processes, he's

21:50not going to be very useful at all.

21:52>> All right, so let's actually onboard

21:54this hire. And you mentioned context,

21:56tools, and skills. give us full context

21:58of how you look at onboarding the hire.

22:01>> Like I said before, this this really

22:02capable stranger you've now got working

22:04for you, he's not going to be able to do

22:05very much unless he fully understands

22:07you and your business, has the tools to

22:09do the work, and then knows how you like

22:11the work done. And you do this through

22:14context, tools, and skills. So context

22:17being like information to the agent,

22:19like what you do, your business, like

22:21your brand voice, things like that.

22:24tools being like, you know, a computer,

22:25a browser, Gmail, notion, Slack, and

22:28then skills being like your SOPs, like

22:30your special source, like literally like

22:32documents like you give a VA. And then

22:35if you get those three things right,

22:36that's how you like properly on board

22:38the stranger, and then you have the best

22:40employee you ever had.

22:41>> Sounds amazing. Let's go super super

22:44deep into each one of them. So, we have

22:46context, tools, and skills. Let's double

22:48click on context and tell us exactly

22:50what you mean by context. Yeah. So,

22:53context is literally like everything

22:55that the agent should already know

22:57before you give it a task. And for me,

23:00this looks like things like about me.

23:02So, all information about like me as the

23:04user cuz I want them to like I want my

23:06my new employee to understand like

23:07everything about like who I am and what

23:09I do. Even some things about like my

23:11story and like my come up and things cuz

23:13then it can use it if it's ever writing

23:15like a landing page copy or like a sales

23:17letter script. it can reference like

23:19real things in my life. It also looks

23:22like a document about your business

23:25information. A like a maybe another

23:26document of like your offer catalog

23:28where you talk about like all your

23:29services and offers and their prices and

23:30how they link together like upsells,

23:32downells. Another one might be like an

23:34ideal customer profile which is one of

23:36my like most important ones. So it like

23:38knows who your business like targets and

23:42quite literally

23:44like if you were onboarding this new

23:45employee, right? like a real employee,

23:47you'd probably like give them a PDFs or

23:49like some sort of documents, usually

23:50like Google Docs, Word Docs that teach

23:53them all of this stuff or like the key

23:54information you want them to know. You

23:56you are literally doing the same to the

23:57agent. So like just like you know you

23:59get get a new VA, you like give them all

24:01these docs, you're literally giving

24:03these docs to the agent and these are in

24:05the form of markdown. So for the

24:10listeners that don't really understand

24:11what markdown is, markdown's like it's

24:14it's basically one step above a .txt

24:17file. So TXT text file is just literally

24:20plain text. No headings, no bolds, like

24:23nothing. It's just text. Markdown is

24:25literally just one step above that. It's

24:27just a plain text file, but you can do

24:29like headings, boldings, like bullet

24:32points,

24:32>> simple formatting, stuff like that.

24:34>> Yeah, it's just like a slightly more

24:36formatted text file. And you know, if if

24:38any of you use out there use like a tool

24:39like Notion, the notion docs or like

24:41ClickUp, the ClickUp docs, they're all

24:44in markdown format. Just super simple

24:46like headings one, twos, and threes,

24:48text, bullets, and bolding and like

24:49italics. Just the super basic

24:51formatting.

24:51>> And to give you guys further context,

24:52it's basically like a simpler doc file.

24:54So, it's a text doc. It's somewhere in

24:56that universe.

24:57>> Exactly. And the reason like why

24:59markdown is so important is you could

25:02theoretically let's just say you're

25:03building you're following along and

25:04you're trying to build out your own

25:06agent. You could just give them all

25:08these in a docs file, right? Or a PDF.

25:11However, it it's a lot harder for the

25:14agent to access the information. Like if

25:16you give an agent your ideal customer

25:17profile as a markdown file versus a PDF

25:19with the PDF, it's going to have to

25:22unpack it and like try and extract the

25:24text cuz it's just the way that the PDF

25:26file is like built. It's hard for the

25:28agent to like ingest the information.

25:30Whereas with markdown, it's just like

25:32that straight ingest. It can just read

25:34it really really easily. And we want to

25:35make things easy for the agent because

25:37that just means it's things are quicker

25:39for you and things are cheaper in terms

25:40of like cost for the model running. So

25:42markdown is just the best way to be

25:44giving information to your agent.

25:46>> To give people context as well, how many

25:47MD files do you have on your computer?

25:50>> Like hundreds.

25:50>> Hundreds.

25:52>> But like so the these you're literally

25:54building out like markdown files for key

25:58areas of like information that you want

25:59to give this agent.

26:01>> And as far as that goes too, you can get

26:03would you just say the more granular the

26:05better?

26:06>> Not necessarily. I don't really think

26:08there's like a right or or wrong way to

26:10do it. It's more like I I just build the

26:12markdown files for me almost like how I

26:14would want them formatted. Like I've

26:16just like literally got like a about me,

26:18like a business info, brand voice, idle,

26:20customer profile, like offer catalog. I

26:23could put them all in one doc, but then

26:25it's just like I don't know. I almost

26:27just prefer them having having them

26:28separate.

26:29>> Yeah, we're going to get into folder

26:30structure as well, too, so everybody

26:31understands kind of from a macro

26:33perspective how this brain looks and how

26:34it kind of all flows. I think for the

26:37people out there listening, you know,

26:38we'll get in, you know, what is the

26:39first thing that you can do to kind of

26:41start your journey. I feel like like the

26:42about MD file is like a pretty clear one

26:45of like

26:46>> great spot to start. And the thing you

26:48need to understand is these markdown

26:50files you're building are like assets.

26:52They're assets for you and your company.

26:54So like really like spending the time to

26:56build these docs like really really

26:57well. Like if you just like take the

27:00time to build a really wellthoughtout

27:02about your business doc and that like

27:04the ideal customer profile they're

27:06assets for your company because if you

27:08rush the ideal customer profile doc and

27:10it's just like a bunch of AI slop that

27:12you got chat here to put together and

27:13you haven't read over every line. There

27:14might be a line in there that says like

27:16my ideal customer like if you're a

27:18skincare brand a line in there in your

27:21ideal customer profile might say their

27:22problem point is like red skin or

27:25something. And if you don't check over

27:27that now in all your marketing copy,

27:29you're gonna have like it targeting

27:31towards the angle of red skin. And if

27:33that's not something that your product

27:34solves, then that's like a that's an

27:38issue. You know,

27:39>> the answer on the MD files is probably

27:41you want to minimize the subjectivity in

27:43the MD files. Make them all objective,

27:44like stuff that's kind of black and

27:46white. Like I definitely would use this

27:48as an input to make decisions.

27:49>> Yeah. Yeah. Agreed. And I and I get

27:51Claude to build all these files for me,

27:53but I read over every line before I like

27:55give it the okay.

27:56>> Guys, I got a little cheat code for you

27:58too on the about.md file. Like the

28:00craziest thing about all this is like

28:02I'm not saying have Claude do it, but

28:04you can say to Claude, I want to build

28:06an about.md file. How should I, you

28:08know, how should I process this? How

28:10should I think through this? Yeah.

28:11>> And that's where you I do think you can

28:12use something like chat because you can

28:14kind of go back and forth and build it.

28:16>> Yeah, that's like a a perfectly fine use

28:17case for like the chat models. Like

28:19that's how I built out all of these. I

28:20literally said to Claude, like in a new

28:22chat, I was like, "Yo, I wanted you to

28:24build out a bunch of markdown files for

28:26me, like an about me, and I had a custom

28:28profile, just listed out all the ones I

28:30wanted, and I just said, interview me,

28:32like to extract the information from my

28:35brain to build these files, and I sat

28:36there like using Whisper Flow answering

28:39all the questions for like 3 hours,

28:41>> just riffing, and then Whisper Flow just

28:43cleans it all up."

28:43>> Yeah. It just like sucks it all out of

28:45your brain into these files that are now

28:46assets for your company that you can

28:48then give to your agents. And like one

28:49of the key distingu like the key

28:52distinguishments to make is when you're

28:54using these chat models, they're sort of

28:56like a black box. Like everyone always

28:59goes on about like my chat GBT knows so

29:01much about me. I don't want to move into

29:02Claude because like Claude doesn't know

29:04me. ChatGBT knows me so well. And it's

29:07like, you're right, but what it's doing

29:09is it's like got pretty much just a

29:13markdown file like memory MD or like

29:16about user.md in the background and

29:20chatgbt is building this file on your

29:22behalf without you knowing and you can't

29:23access this file. So if you're like

29:25starting a chat GBT chat and you say in

29:26there like I'm I've just moved to New

29:29York City like what are some things like

29:31good cafes near me? JGBC in the

29:34background will take out that snippet of

29:36like I just moved to New York and save

29:38it in its markdown file and then now

29:40like it knows you've moved to New York

29:42and you just think that this model's

29:43like almost trained on you now but in

29:45reality it's just a markdown file where

29:47it's choosing what to save and the other

29:49thing about that is you can't control

29:52what it saves as well like your chat GBT

29:54like knows everything about you and then

29:56like you know you're writing like a

29:57landing page you're asking for advice on

29:59chat GBC like hey how do I break up with

30:01my girlfriend and then you're writing a

30:02landing page and it's like all of a

30:04sudden the landing page copies about you

30:05breaking up with your girlfriend. It's

30:06like you can't you have no control with

30:10these chat models about like what it

30:12knows and what it saves

30:13>> like it it does get mixed up sometimes.

30:15>> Exactly. So you want to make sure that

30:17you have full control. You want to own

30:18that context and that's like I'm talking

30:21about just in the format of these

30:22markdown files and then you can they're

30:24assets that you own now and you just

30:25bring them to whatever agent you're

30:26using. So like I've got my folder set up

30:29which we're going to get into with all

30:30the context in there and I can just open

30:33up that folder in claude code or codeex

30:36or manus and like straight away it knows

30:38everything about me. It's like it

30:40doesn't matter like I can just switch

30:42and stay sort of like mobile. Guys, real

30:45quick, everything that Remy is teaching

30:46you in this video is absolutely free. No

30:50gatekeeping as always. But for the

30:52people who want it implemented, Remy

30:53made a plugin that builds this entire

30:55system for you around your business and

30:57your life. The loop, the folder

30:59structure, all of it. I am not

31:01technical. I did it myself. 90 minutes,

31:04one sitting, done. Sign up for the

31:05newsletter below and get 50% off at

31:08launch. Enjoy the rest of the episode.

31:10So, the first step for onboarding agent

31:11is like building out these context

31:12files, right? And I'm going to show you

31:14like what this actually looks like. So,

31:16I'm going to open up Cloud Code again.

31:17Here,

31:19we're going to do a new session.

31:22Nice and clean.

31:23>> Hey, there we go. And I'm got my OS

31:25folder open now, which we'll touch on

31:26later. But the way that these agents

31:28work now is a lot of them run like

31:30locally on your computer as opposed to

31:32like the cloud. And what locally on your

31:36computer means is it's like instead of

31:37having like a project on your chatbt

31:39account, your that project folder now

31:41lives on your computer. And there's a

31:44lot of benefits to that which I'll I'll

31:46touch on later, but I'm just going to

31:47open up this one I've created now called

31:49demo. So this is just a brand new

31:52folder. You can see it's completely

31:53empty

31:53>> and that's just you just created an

31:55empty folder on your desktop.

31:56>> Yep. So it's desktop empty folder called

31:58demo. And we're just going to open that

32:00up. So now you can see in cloud code

32:02we've scoped the agent to just work in

32:05that folder. Just like you know when you

32:08have a a project in chat GBT, it's like

32:10any of the chats you do in that project.

32:12It's like scoping chat GBT to just work

32:14there.

32:14>> So now it's a new blank canvas and

32:17you're you're deciding what inputs go

32:19into this folder.

32:20>> Exactly. Exactly. So you can see just to

32:22demonstrate this the importance of

32:23context. If I'm opening up this brand

32:25new folder, so I've been working with

32:27claude code, right, for for like eight

32:30months and if I ask like you know who am

32:35I?

32:37So you can see like when I've asked it,

32:39there's no markdown files for it to look

32:40at. So, I've asked it who am I? I've

32:42been using CL code for eight months. It

32:44should know. But it doesn't. It's

32:45because all the context you own in those

32:47files. It's not a black box.

32:48>> So, if someone has an alter alias, they

32:50can they can just create different

32:51alteres in in in different folders.

32:53>> Yeah. Essentially, so you can see here

32:55it it tried to read the memory MD mark

32:58file, markdown file, but it doesn't

32:59exist. So, you can see here it's just

33:00like the f file doesn't exist.

33:02>> So, it went through that loop. It was

33:03searching, but there was no kind of raw

33:05data to extract.

33:05>> There was nothing for it to to do.

33:07>> So, how do they know the email and the

33:08likely name? because that was just from

33:10my claude account. It must just have

33:11access to that. That's literally just

33:12because that's the email associated with

33:14my claude account and name. Actually,

33:16no, no, no. Actually, it's it's pulled

33:17that from my computer because this agent

33:19lives on, you know, it's got you got a

33:21username on the computer. That's where

33:22it's gotten it from.

33:23>> Okay. So, what does this look like when

33:24you're on steroids? Let's dive deeper

33:26into kind of doing these MD files.

33:28>> Yeah. So, let's let's continue

33:29onboarding our agent, right? So, the

33:32most important markdown file, like, you

33:34know, we've got our markdown context

33:35files. The most important one when

33:37you're dealing with agents is what's

33:39called an agents.mmd file. And I mean

33:43claude have just decided to be difficult

33:44and in in their systems they call it a

33:45claw.md. But in pretty much all the

33:47other harnesses you use like codeex and

33:49open claw and stuff it's called an

33:51agents.mmd. And essentially what happens

33:53is is every time you're like starting a

33:56new session and by new session I just

33:58mean like if we go back to that goal to

34:00result thing it's like every time you're

34:01going to like give it a new goal you're

34:03kind of going to click like new session.

34:04the same way you start a new chat in

34:05chat GBT.

34:07>> Each session has an amount of like

34:08space. It's kind of like you can think

34:10of like the agents like brain power and

34:12as the session starts to like fill up

34:15the you have what's called context rot

34:18which is basically like have you ever

34:20been in a in a chat on chat GBT and you

34:22it's like so long you've been in there

34:23for hours and it starts to just like go

34:25crazy like it literally just starts to

34:26just like hallucinate forget things. So

34:28what what's happening there is every

34:30time you start a new session like a

34:31fresh like chat window I guess it's got

34:34like a certain amount of of I guess

34:36capacity you could call it like free for

34:38the actual work and what your sessions

34:40probably look like is something like

34:41this where they're just like like so

34:44like so much information that the stuff

34:45you started telling it at the start like

34:47fall out the top

34:48>> that was V1 Mark bro I'm not on V1 Mark

34:50anyone I mean I think 99.999% of the

34:53world is in probably V1 Mark one endless

34:56chat to be frank with you.

34:57>> Yeah. Yeah, probably. So, every time you

34:59start a fresh session, the way that this

35:01agent like knows about you. Same way

35:02that chat gbt's memory works is if you

35:04start a new chat in the background, it

35:06started the chat with that memory. MD

35:09loaded into it, but you can't see that,

35:11right? So, you give it a prompt like uh

35:14write me an email in chatgbt and to you

35:17it just shows as like that one prompt

35:18you've given chatg, but in the

35:20background it's also loaded in that

35:21memory markdown file. So instead you've

35:23got like like a hundred lines of

35:26information about Mark and then write me

35:28an email. So it's got all that context

35:30in it. Same thing happens with agents,

35:32right? So we've started a fresh session

35:34here. What this actually looks like is

35:36any file, any markdown file in that

35:38folder titled claude.md gets

35:41automatically loaded in. So your

35:44claude.md or agents.mmd

35:47that's like your your northstar markdown

35:50file. That's the one you need in every

35:51single project you're working in. So to

35:54illustrate my point, I've created this

35:56folder here with a claude.mmd in it,

35:58right? That I've pre-prepared. And we

36:01can actually just open this up. So you

36:02can actually see what a markdown file

36:03looks like. And I got Claude to build

36:05this for me. So we can see Rey's AI

36:08operating system. Who am I? Rey's the

36:10founder of AI with Remy. The business,

36:13the brand voice, the ideal customer, the

36:15offer, the values, how I like to work,

36:17preferences.

36:19This is like our you know northstar

36:21context file and that's the one that's

36:22going to load in to each new session by

36:24default

36:25>> to give macro context on how you're

36:27thinking.

36:27>> Yeah. To give macro context essentially.

36:29So that same example right if we go back

36:32into claude here we just asked it in

36:34that folder when it was empty who am I?

36:35It had no idea. Now if we go new session

36:39same folder but it's now got that

36:41claude.mmd in it. And we can say who am

36:43I?

36:44>> Heyo.

36:46And you can see here now it knows

36:49exactly who I am. It knows exactly like

36:51what I do, my background, the brand, the

36:53three products.

36:54So that's like a great illustration

36:57there. Do you see how how it goes from

36:59knowing nothing about you to now having

37:00that context file that knows everything?

37:02>> That's why it's so important to be lean

37:04and intentional on what files you upload

37:06obviously in there.

37:07>> Yeah. Yeah. Exactly. So that's that's

37:09your pinnacle markdown context file.

37:11That's like the first one that you need

37:13to build. the the part about the

37:15claw.mmd made sense, right? That gets

37:17slotted in and it's like the agent wakes

37:19up to do a task and he's got that loaded

37:21into his brain, right?

37:22>> Yeah. And that's something that's you

37:24should constantly be updating if

37:26something big in your life changes.

37:27Correct.

37:27>> Yeah. Yeah. Correct.

37:28>> Yep.

37:29>> So, you're probably like wondering now,

37:30okay, great. I get that claw.md

37:32agents.mmd. Where do the other ones fit

37:34in? Like the about me and the customer

37:37profile, like where do I put those? So

37:39those ones, the other markdown files,

37:41they don't get loaded into that session

37:43window by default. So we need to we've

37:45got our CLMD that gets loaded in by

37:48default. And then we need to tell it in

37:49that that it will always read, hey, go

37:52check here as well. So then what I like

37:54to do, how I have this set up is we got

37:58our claw. MD here, and I'm going to

38:00paste in these other ones I've

38:01pre-prepared.

38:03So now you can see we've got a couple of

38:05other files taking shape here. So we've

38:08got our claw.mmd.

38:10We've got this context folder with all

38:12these markdown files in it. My values,

38:14my offer, my item customer profile,

38:16brand voice, about me, and then this

38:18memory.mmd, which we'll touch on in a

38:20second. So by default, claude is any any

38:23project you open up claude in to work,

38:25it's always going to read that

38:26claude.mmd. And then we're going to

38:28point to the context folder in the

38:30claw.md. So you can start to like

38:32compartmentalize and like modulate your

38:34your project, if you will. making like a

38:36whole like a like decision rubric like a

38:38decision tree of you know if it goes

38:39here it opens up this folder and stuff

38:41like that.

38:42>> Exactly. And then this is like all these

38:44agents you build they're super like

38:45modular. It's like you hear other ones

38:47really popular like Obsidian is a really

38:49popular thing that people give to their

38:51agents as like a second brain

38:53essentially like there's loads of

38:55different like frameworks like that. And

38:57instead what they're doing like if if

38:59you had this folder and you wanted to

39:00use obsidian instead you would instead

39:03delete this context folder and you would

39:05in your cloudmd point to the obsidian

39:07vault instead. That's kind of how people

39:08are doing that like you can kind of

39:10modulate the the context.

39:11>> So in short you can kind of redirect it

39:13whichever route you want to go.

39:14>> Yeah. Exactly. And it all stems from

39:16this file here. So now this new setup

39:18we've got is kind of like evolved. So we

39:21just the first one was just the

39:22claw.mmd. And that's probably the most

39:23basic one that if you're trying to like

39:24start using agents for yourself. That's

39:26probably the first step I'd say is just

39:27create a folder and then just build a

39:30cloud. Mmd just ask claude, hey build

39:32this claw.mmd file for me and just

39:34include in it like an overview on who I

39:35am and what I do just to give it like

39:37that base context. Then v2 is now we're

39:39adding in like this extra folder for

39:41like we're expanding out our our context

39:44here. And if we open claw.md now, now

39:47I'm going to point to that context

39:49folder. So you can see if we open it up,

39:50it's like who am I? My context read

39:52before any task. Everything you should

39:55know about me and the business lives in

39:56the context folder and then it's just

39:58given an overview on the files.

39:59>> Wow.

40:00>> Load the relevant files in before

40:01starting any tasks. Assumptions are the

40:03enemy. If the answer isn't there, ask

40:05me. And then everything on continues

40:07like normal. And then the memory I'll

40:09touch on in a sec.

40:10>> And then as you keep developing more

40:12folders, you just keep updating. The

40:15most important thing is the claw empty

40:16file cuz that is going to create the

40:18infrastructure like you could have like

40:20marketing, finance,

40:22>> logistics.

40:23>> Yeah.

40:23>> And then underneath that you can have

40:24multiple folders.

40:25>> Exactly. So when I asked Claude before

40:28like who am I? It knew straight away.

40:29That's cuz that file automatically

40:30loaded in. But if I deleted that claw.md

40:33and then asked it who am I again?

40:35>> Back to dummy. Back to dummy.

40:36>> Well, it's got the context folder.

40:38>> Oh yeah.

40:39>> If I just deleted the CL MD and left the

40:41context one, it wouldn't know straight

40:42away. It would then go through a loop to

40:44think like, oh well, I don't I don't

40:46know who who he is. Uh, let me think. I

40:49can let me let me check for what folders

40:51I've got. Oh, context. Then it would

40:52read it. Then it would load it in. But

40:54it's just like it's slow and it's got to

40:56go through another loop. So like that's

40:57why you have the claws that MD there.

40:59>> Yeah.

40:59>> Why why don't you everybody listening

41:00right now just ask it something very

41:01very basic under the notion that now you

41:03have context based in like something

41:06about any of those kind of raw folders.

41:08I know you had under there like brand

41:09voice like anything.

41:11>> Yeah. Yeah. Well, what I what I might do

41:12is I'll just say like write me a cold

41:15email to

41:16>> how about to Mark to come on uh to come

41:18on open residency.

41:19>> Actually, know this. I'll just say like

41:23>> I'm going to whisper flow it to demo.

41:25>> Yeah. Amazing.

41:25>> Makes sense.

41:27>> Write me a cold email to Mark from Open

41:30Residency podcast to try and get me a

41:33position on the show.

41:34>> Wow. Whisper for the win, baby.

41:36>> Whisper flow is so good. Like talk about

41:39like agents stacking up the

41:41productivity. You can even stack on top

41:43of the agents when you start using

41:44Whisper Flow because it's like you can

41:45just just get things down onto the page

41:49at the speed that you can think.

41:50>> For the people out there listening, I'm

41:51going to give you guys a little cheat

41:52code. One of our episodes a bit before

41:53this is of Chris Voss, which is a

41:55negotiation expert.

41:56>> You can make some MD files with him to

41:58speak in his language. So it probably

42:00would be a bit more persuasive on this

42:02email that you're going to send me.

42:03Let's let's let's see how good it is.

42:04Let's see your sales pitch right now.

42:05>> Okay, great. So it's gone through a

42:06couple loops, right? It's got decided to

42:08go and research who you are. It's gone

42:10to go and like find

42:11>> that is me

42:12>> open residency

42:13>> and now it's written right. It's like

42:16literally said here like like we haven't

42:18given any other context. We just said

42:19write the cold email and it's got cuz

42:21it's got our our clawd and then this

42:24context folder and that context folder

42:25is so rich with like my values, my

42:27philosophies, my like thought processes

42:29on things and my offers. So you can see

42:32like the subject lines, right? Operating

42:34a 20 person team without hiring 20

42:35people. guest idea for open residency.

42:38But the email right here, hey Mark, been

42:40working through open residency. The

42:41operator as last line of defense framing

42:44is the realist take on running a

42:45business. Terrible cold email by the

42:47way. Terrible. But it's basically you

42:50can see here it knows everything about

42:51me. It's like I teach founders to stop

42:52drowning in tools.

42:54>> This is where we're going to get into

42:55like the additive layers of like an

42:57example is like that negotiation

43:00>> skill that you know how you speak skill.

43:02Like you can get way deeper to make this

43:04thing absolutely flawless. Clearly, it

43:06did an amazing job.

43:07>> Yeah. I mean, the key point demonstrate

43:08was right that it just knows about me.

43:10It it can write an email with all the

43:12context. Like, it doesn't like the

43:14email's terrible, but like and you can

43:16tell it knows about like what I do and

43:18my philosophy.

43:19>> Amazing. You're booked. Let's go.

43:21[laughter] We on. We're here.

43:22>> Okay. Great. So,

43:24>> and just to take a step further, just to

43:26kind of paint the picture of where we're

43:27going to go with this.

43:28>> Yeah.

43:29>> This could be this email is drafted. It

43:31could also be automatically sent as well

43:33too, which we're going to get to to

43:34connectors and MCPS as well too, just so

43:36people know like this could be fully a

43:38toz. Find all the business podcasts,

43:41draft the email, and send the email.

43:42>> Then you schedule it. So every week it's

43:43going to or every morning it's going to

43:44go find business podcast, cold email

43:46them to get you book calls, like all

43:48this stuff. It's crazy.

43:49>> You're going to need to dial that email

43:50in a little better for me to say yes

43:51though, but it's okay. Don't worry. Keep

43:52going.

43:53>> So So that's that's what I'm going to do

43:54now. I want to display like so

43:56understand like we need these context

43:58markdown files. The most important one's

44:00the claw.md that then points to the

44:02others. And now like when we open up a

44:04new session, it it it straight away

44:06knows everything about us, right? But

44:08the next layer of context is memory,

44:11which is still kind of under the

44:13umbrella of context, right? Memory and

44:15lessons. So off the rip, it knows all

44:18this stuff about us. But what if we ask

44:19it to do something differently? So you

44:22can see I've got in here this memory.mmd

44:26file, right? If we open this up, this is

44:29just this has got memory learned rules,

44:32lessons Remy has taught me. I check

44:33these before every task and it's got

44:35never write formally, just a couple of

44:37like random preferences. So the agent

44:38like knows everything about us out of

44:40the box, but we want it to be like be

44:42able to improve over time and like if we

44:44correct it, we wanted to save that

44:45somewhere, which also counts as context,

44:47right? So I like to build this little

44:49like kind of de facto memory system. I

44:53think like when I started putting this

44:55together, it was because none of these

44:56agent harnesses had memory built in. So

44:58if you said to it like right this this

45:00email right here, if we said like can

45:03you please sign off the email with

45:05cheers Remy and like I always want you

45:08to sign off emails with cheers.

45:09>> It's not auto updating unless you put

45:11>> Exactly. It's not so like before the

45:13last maybe in the last month they've a

45:15lot of these harnesses have now added in

45:17memory as a feature but before that you

45:19had to do it yourself cuz otherwise it

45:20would say like if you didn't have that

45:22in place it would say like yep got it

45:23but then next session in a week time

45:25when you write an email it wouldn't do

45:26it.

45:27>> Show them a memory MD just in case

45:28they're driving the Kia and not the

45:29Lamborghini and maybe that LLM does not

45:31have it built in. Show them a little bit

45:33of context.

45:33>> So we can just show in here like the

45:36memory

45:36>> MD. No shame on the Kia either. I've

45:38seen some nice Kia by the way too.

45:39[laughter]

45:40So, we got the the memory learn rules.

45:42These just a couple I put in here

45:43before, like never write formally, like

45:44always use real examples, never generic

45:46placeholder ones. And then same way that

45:49we're pointing Claude at the context

45:52folder, we're also going to point at the

45:53memory folder. And I've just basically

45:55said, you know, your memory lives in

45:56this file. When I correct you, write

45:58lessons into this memory file as a new

46:00rule and then check it before every

46:01task. This is how you go from good

46:03assistant to know exactly how I like

46:04things done. And I have one little tip

46:07is if you use the at@ symbol if you tag

46:09a file like that then it like it's

46:11almost like a a sim link. I don't know

46:13if you know what that is on a desktop.

46:15It's like

46:15>> I don't

46:16>> if we for example if I tell it

46:18>> is it like a lighter load on the brain

46:20like that it zips right into the file or

46:22something?

46:22>> Yep. Zips right in the file. It's it's

46:24like as good as the file being loaded in

46:26if that makes sense. Cuz if we just tell

46:28it like go use the memory MD and don't

46:30add it'll then have to make a loop to go

46:32and find the file and bring it in. But

46:34if you just have like the app, it's

46:35automatically loaded straight in.

46:37>> Guys, we're going Lamborghini full

46:38steroids here.

46:39>> Hell yeah. So

46:42Claude like has this this built in, but

46:44even still, it's still almost a little

46:46bit of a black box. It will save it to

46:48like Claude's desktop app. So now if

46:50you're trying to work with this folder

46:51in like codecs where you want to try and

46:53like use other harnesses, you don't own

46:55the context file. Claude kind of owns

46:57it. So like that's still why in all my

47:00projects I still kind of add this memory

47:01system in. So we can just test this out

47:04here, right? And we can say

47:06please in future when creating websites

47:09I want you to always use dark mode and

47:12never ever use light mode.

47:17>> Here we go. It's done it. Bang.

47:19>> Okay, great. So you can see it's now

47:21gone and said add I'll add this to your

47:23memory rules so I never forget it. It's

47:25read the memory md and then you can see

47:28it's added it's edited and it's added

47:30this line. When building any websites,

47:32always use dark mode. Never use light

47:34mode.

47:35>> Love that. That's obviously in that auto

47:38updated in that folder that you showed

47:39us.

47:39>> Yep. So, if we go back here into this

47:41demo folder, we go into our memory.mmd

47:43file and open that bad boy up. You can

47:45see when building any website, always

47:47use dark mode, never use light mode. And

47:50then now this is like we own that that

47:52preferences, right? This will grow over

47:53time. And like I've had the question

47:56before, it's like what happens if that

47:57just keeps growing? you've got like

47:59hundreds of different lines in there and

48:00they all start to like contradict and

48:02get messy. Like I don't think that will

48:05happen. Maybe in like 6 months time you

48:07can just do a spring clean. Just

48:08>> open a session with Claude and go like

48:10>> just audit my memories and just take out

48:12anything that's contradictory or

48:13conflicting and cleans things up.

48:16>> So that is pretty much like context 101.

48:19Just to recap, we've got the claw.md

48:21which Claude autoloads in every session.

48:23We want to give it more context. we add

48:25that extra little file and then point

48:27claude to that folder and then we've set

48:29up our little memory like

48:30self-improvement system as well.

48:32>> So from a loop perspective we're going

48:34to start with the about.mme the claw.md

48:36is going to be kind of the the macro

48:39infrastructure which is then going to

48:41lead into more folders which we'll

48:42obviously get into and then the memory

48:44MD is just a self-fulfilling loop that

48:46just keeps remembering all the

48:48preferences and such.

48:49>> Yeah. Yeah. Exactly. So now like every

48:52every session

48:54we start that observe step now has all

48:58the context and skills loaded in. So

49:00every time we're giving it a goal

49:01straight away in the first observe step

49:03before it does anything we've already

49:05got like now I mean skills will touch on

49:07but now you can see we've really tuned

49:09up that context step. So when it starts

49:12to work on your tasks it has like all

49:14the information that it needs. And I

49:16think one last thing just to just to

49:18finally like drive home this point. We

49:20can run in Claude code this little

49:22command. Don't worry about all the

49:23others. We'll touch on that later. But

49:24this is a one that comes built in with

49:26Claude. You can go slash context and it

49:29will actually pull up that real bar

49:31chart for you. So you can see here

49:33context window. We've used 6% of that

49:36bar.

49:37>> Are you talking about 6% of like the

49:38memory within that individual chat?

49:40>> Yeah, within that individual chat. So

49:41you can see it's already like our bar.

49:44used up 6%. And if we look at like we

49:48we've given it one task here, right?

49:49Like please in future tiny little

49:51message and it's 6% which is a fair bit.

49:53It's almost like it's a fair bit of of

49:55the total space used. So you're starting

49:57each session even if we just did like a

49:59blank thing with no messages sent you'll

50:02always like have a little bit of your

50:04window chewed up because it's feeding in

50:06all that context before you start any

50:08task. That's why you could also argue

50:09going back to the beginning is keep the

50:11MD files like lean and cut, you know,

50:13only use what you need to use.

50:14>> Yeah, exactly. Because if you have these

50:15like really bloated context files and

50:17like you have, let's say you have 100

50:18markdown files, you start a session and

50:20the context window is already like 20%

50:22full. It's like you you want to Yeah.

50:24100% to your point, you want to keep

50:25them nice and lean and like really like

50:27potent with like high density

50:29information.

50:30>> Amazing.

50:30>> I use Whisper Flow for hours every

50:33single day. I think way faster than

50:35[music] I type. I speak about 220 words

50:38a minute and type only 45. Whisper Flow

50:41closes that gap. It's a voicetoext tool

50:44that works anywhere on your computer or

50:46phone. You talk and clean formatted text

50:49shows up wherever your cursor is. All

50:51right, let me show you. I'm going to

50:52answer a real email right now without

50:54typing. You can see here Tiger is asking

50:56for feedback on our intro video for the

50:58next episode. Watch how I respond. if

51:00you could just make a few changes. Uh,

51:02make the red a little bit brighter,

51:04[music] the typography a little bit

51:05smaller. Uh, for the music, if you can

51:07just give me some other options, and

51:09[music] that first photo, some more

51:10options as well. Uh, please give me

51:12three options for each. And, uh, make

51:14sure you send it to me by end of the day

51:15Friday. Thanks.

51:17So, as you can tell, the text is way

51:19cleaner. [music]

51:20I kind of speak in half sentences and

51:22sometimes mumble on. This makes it very

51:24clean, formatted perfectly, and avoids

51:27all the small mistakes. This is a

51:29product that I cannot live without. I

51:31promise you that you will love it. If

51:33you guys want to get one month free of

51:35Whisper Flow Pro, click the link below

51:37and use code open residency.

51:41So, yeah, context. We've we've we've

51:43taken our really capable stranger that

51:45we've employed and we've told him

51:47everything he needs to know about us.

51:49He's ready to start working for us, but

51:51he's still going to be pretty useless if

51:52you don't give him like tools to work

51:54with. Like imagine like you've just got

51:55this this dude that's working for you,

51:57but it's like he doesn't have an email

51:58account or a computer or a browser or a

52:00phone. He's like still going to be

52:02useless. So we need to give him tools to

52:03be able to actually complete the work

52:05for us. So the way we do this is what's

52:08called a thing called MCP. It stands for

52:10model context protocol. And I don't want

52:12to lose anyone here. It's it sounds

52:14technical, but like it's it's super

52:16simple. And and this all MCP is is just

52:18a way for your agent to connect to

52:21tools, things like

52:22>> third party tools.

52:23>> Yeah. It's a third party tools. Yeah,

52:24exactly. Things like, you know, Slack,

52:26Gmail, Calendar, like you need a way for

52:28your agent to to connect and you do that

52:30through what's called MCP.

52:32>> To give people more context as well, I

52:33think a lot of people out there use

52:34either N8 or Zapier is a big one that

52:37people use to kind of extract

52:38information and and connect it.

52:40>> And we're in the era now. I'm so curious

52:42to see. I think all the principles here

52:44are evergreen,

52:45>> but it's like it's a big deal now to

52:47have an MCP. And like some of these big

52:49companies don't have MCPs

52:51>> and like it's it's they're crazy. We're

52:53going to get into the BI of MCP

52:55>> that it's what I make purchasing

52:56decisions. Now, if if I'm choosing,

52:58let's say I'm choosing like a marketing

52:59email marketing software, for example,

53:01if like if I've got five options and if

53:04if they don't have an MCP for a way for

53:07me to connect my agent to it, I it's out

53:09of the question. I won't use it. I don't

53:11use front ends anymore of apps. Yeah.

53:13like like the future of this is and

53:17everyone that I know that works the way

53:18this new way of working this stage two

53:20which is agents they don't really use

53:22like frontends anymore like I complete a

53:25full workday from with inside like my

53:26cloud code or or agent harness that I'm

53:29using

53:29>> by front ends just so people out there

53:30listening like you're not in platform in

53:32behind on the dashboard you're just

53:34extracting from there and then working

53:36within your workspace

53:37>> y like I use Google calendar I use Gmail

53:40I use Slack I use Stripe for payments

53:42cal.com for bookings And I can't even

53:45tell you the last time I've opened any

53:46of those apps up. They're just all

53:47connected into CLA. It's this new it's

53:49it's a completely new way of working

53:51where you've got

53:52>> Claude or whatever AI harness you're

53:54using that sits on this as like an

53:57intelligence layer on top of all your

53:59tools and then you're just interacting

54:00with like this this intelligence layer.

54:03So, you know, I've alluded to MCPS being

54:05just this this way for your agent to

54:07connect to tools. But super quick on

54:09like what it actually is, right? So

54:10before MCPs existed, if you wanted to

54:13connect your agent like Claude to a

54:15tool, it was it was difficult. It was

54:18possible, but required lots of custom

54:19development. And the way to think about

54:20it is like Claude speaks English, right?

54:22And all these different tools speak

54:23different languages. Like Gmail speaks

54:25French, Notion speaks Spanish, Slack

54:27speaks Chinese. So if you want Claude to

54:29talk to to Gmail to like get it to do

54:31things, it was required a lot of custom

54:34development. It required time and effort

54:36because Claude had to learn

54:37>> heavy, slow, expensive, everything.

54:39>> French.

54:39>> Yep. And shout out Ross Mike. This is

54:41explanations from him. He was on like

54:43some part I heard him. So this is taken

54:45directly from him. This is what landed

54:46like really nicely in my brain. MCPS are

54:48just like this translator that sits in

54:51the middle between Claude and your and

54:52your tools. So Claude does not need to

54:54learn French and Spanish and Chinese. It

54:57just has this little translator in the

54:59middle. So it just makes calls and then

55:01it can be relayed really really nicely

55:02to the tools. This is an

55:03oversimplification but quite frankly

55:05like no one really needs to know what

55:07happens here. All you need to understand

55:08is the MCPs are just your way to connect

55:11tools into your agent.

55:12>> Uniformity, simplicity, and that just

55:14streamlines everything for speed as

55:16well.

55:16>> Same way you can think of it like an

55:18API, right? For those of you who know

55:19what an API is, like it's just a way for

55:21tools to talk to each other.

55:23>> And again, that's an oversimplification

55:25of explaining MCPS and connectors. And I

55:27think the big thing to take out of this

55:28episode is the ability that you can ask

55:30the LLM like, oh, how do I set up

55:33connectors and how I do it? And I don't

55:35want to encourage people to be lazy, but

55:36you want to just quickly just show over

55:38the screen of just where you go into

55:39Claude to do it because it's very very

55:41simple.

55:41>> It's so simple. So the way like if we

55:43open the Claude desktop app

55:45>> and we go into customize, you can see

55:47straight away connectors.

55:48>> Boom.

55:49>> And we have like we go up here to

55:51browse.

55:53There is so many connectors here. Like

55:55pretty much all of the major tools that

55:57you would use have a connection here.

55:59And it's as simple as this, right? So

56:01let's just say we want to connect up

56:03ClickUp. So I can literally just go

56:05ahead and go plus and then I'll just

56:08oneclick login to my ClickUp.

56:12>> Guys, for context, we are power users of

56:14ClickUp. That is the harness the project

56:16management tool that we use that creates

56:19all the workflows, all the dependencies.

56:22>> So we can go in here. Literally, I've

56:24selected my workspace connect

56:27open claude. It's connected. We go back

56:29to the desktop connected to ClickUp. And

56:30you can see here like we've got all of

56:32these different things that we can do in

56:33ClickUp. Create tasks, create lists.

56:36Claude can now pull information from

56:37ClickUp if you're planning like

56:39marketing campaigns. It can update it.

56:41And we've got all these like by default

56:43set to needs approval. So it's always

56:44going to ask you before it takes

56:45actions. But as you start to build up

56:47trust with Claude and like

56:48>> you know you you get more comfortable

56:50with allowing it to have more access.

56:51You can just do always allow. Like all

56:53my tools I just have always allow pretty

56:54much. But in Claude like that's as

56:57simple as it is to connect these. But

56:59that is using MCP by the way. So you

57:02don't really have to deal with any of

57:03the technical [ __ ] But that when you

57:05click the plus and log in that is it's

57:07it's got an MCP server basically. That's

57:09what it's using.

57:10>> It's crazy that you say that like you're

57:12picking vendors based on if they have an

57:13MCP or not. It's literally like 5 to

57:1510x. We're going to dive deeper on a

57:17specific use case onto how much more

57:19granular and strategic and faster and

57:21optimized it is with this.

57:23>> Yeah. I mean, it begs the question of

57:25just like,

57:27are a couple LLMs going to win and then

57:29just the world is just going to be a ton

57:30of MCPS and connectors into it?

57:32>> Like, if you're building a software

57:33product, it's like I I had a buddy come

57:36to me and he's like, "Yo, I'm building

57:37this like travel agent like AI agent

57:40that helps people like plan their trips

57:42and he's got this like nice website

57:43where people log on and you you say,

57:45"Okay, I'm going to New York for 10

57:46days. Like, help me do this." And like

57:48has an agent that plans your trip. Like

57:51I I said to him, I'm like, I I don't

57:53think that's the right way you should be

57:54building products for the future. I'm

57:56like, you'd be better off building this

57:58travel agent as an MCP that you then get

58:02hooked up into Claude and then now

58:04people who use Claude can now have you

58:07show up as a connector here, click plus,

58:10your, you know, travel AI and then add

58:13your product as an MCP and then now you

58:15build a purposebuilt MCP for Claude that

58:18helps that makes Claude be able to plan

58:20your trips better that helps Claude like

58:23gives it some extra tools it needs like

58:24it can search, you know, like Sky

58:26Scanner and some other websites a bit

58:28easier.

58:28>> Yeah. I mean, I feel like Yeah. I feel

58:30like the future is like as opposed to

58:31offering like a hundred things, it's

58:33like be really really dialed on 10 20

58:35things, then it becomes almost like an

58:36external data play in conjunction with

58:39that vertical.

58:40>> Yeah.

58:40>> Like,

58:40>> yeah,

58:41>> be really really smart and understand

58:43and know less things and just feed it

58:45>> into Claude. I completely agree. We talk

58:47all the time about like mid-market B2B

58:49SAS is absolutely smashed. like some of

58:51these companies that got sold like I

58:53don't even want to call some of them out

58:54but like basically what I was doing on

58:57>> these SAS tools

58:59>> I'm literally just doing it inside

59:01Claude right now.

59:02>> Yeah. Yeah. Exactly. I mean like

59:05>> Yeah. The the the SAS market is just

59:07getting absolutely decimated. And I

59:08think another thing is like a lot of

59:10these software is built on like

59:11headcount. Like they're built on like

59:12seats in the software and now it's like

59:14you just people are like reducing

59:16headcount. They got agents in there and

59:17like they can't bill for like agents.

59:20That's bad. Yeah. So, you mentioned kind

59:22of the always allow or kind of ask

59:24first. Like that begs to ask the

59:26question of like security. A lot of

59:28people are very concerned about like

59:30connecting a lot of this stuff. Like

59:31what's your just overall thesis on how

59:32people should approach security?

59:35>> You know,

59:37you just have to weigh up the trade-off.

59:40And for me, it's like, yeah, there's a

59:42risk of things going wrong. Like maybe

59:44sending an email to someone who it

59:45shouldn't without my approval. Of

59:47course, there's a risk, but it's like

59:50I've I've weighed up carefully like that

59:52risk compared to these productivity

59:54gains that I'm making. And to me, it's

59:55worth it. And I guess you just have to

59:56do that in your own head. But like

59:58there, of course, there's so many edge

59:59cases with this. It's like you hear like

1:00:01horror stories about agents doing things

1:00:03for people. And it's like you hear the

1:00:04same thing about people getting hacked

1:00:05on the internet. It's like just cuz

1:00:07someone got hacked or you've been hacked

1:00:08before doesn't mean you're not like

1:00:10going to stop using the internet, you

1:00:11know?

1:00:11>> Yeah, definitely edge cases. And I think

1:00:13like the last line is like data is one

1:00:16thing. I think like connecting your

1:00:18something that has like your credit card

1:00:19and it could like go buy things. I feel

1:00:21like that is kind of the last line of

1:00:22defense, but like

1:00:24>> you could just do like for a bunch of

1:00:25things like draft only or take it to the

1:00:28finish line and don't click the button

1:00:29at the end.

1:00:30>> Yeah. So I would say like if you're

1:00:31really like conscious about it like give

1:00:34it these tools but give it what's called

1:00:36a readon access. So it can't do actions

1:00:38it can just read the data. For example,

1:00:39read emails but it can't send anything.

1:00:41And that's just a nice way to start out

1:00:43and then you can like increase the

1:00:44permissions as you start to like feel

1:00:46more trustworthy. I give Claude full

1:00:48access to everything. Stripe everything.

1:00:50It's it's got full access. I'm

1:00:52comfortable doing that and it's a risk

1:00:53I'm willing to take on. I've been in

1:00:54power using it for like 8 months and

1:00:56I've never had any issues. Some of those

1:00:58other harnesses like Open Claw and like

1:01:01Hermes and stuff, they are built in a

1:01:03way where the agents like not going to

1:01:05ask you for permission as much. The way

1:01:07that like these big companies like

1:01:09Anthropic and and and OpenAI have built

1:01:11like Codeex and CL code. The models are

1:01:13very good at asking you for permission

1:01:14for stuff. Like pretty much anytime

1:01:16sometimes it gets annoying for me. It's

1:01:17like I'm trying to get it to answer

1:01:19emails, but it'll ask me for permission.

1:01:20Hey, should I send this? I'm like, yeah,

1:01:21send it. Like I've given you

1:01:23>> like allow access, allow access, allow

1:01:25access all the time. Yeah.

1:01:26>> Yeah. Yeah. Yeah. Yeah. Exactly. So, I

1:01:28I've never had any issues personally,

1:01:29but it's just something you have to

1:01:30weigh up. I mean, like there's two sides

1:01:32with the security question. There's like

1:01:34obviously giving it your tools and then

1:01:36it going and like taking actions. That's

1:01:39that's probably my least cons like my my

1:01:41least concern is it taking actions on my

1:01:43tools cuz that's like not that big of a

1:01:45deal. The other security side of things

1:01:47is like the models and like putting

1:01:49sensitive data through it. And this is

1:01:50like a big thing and like a big problem

1:01:52I'm trying to like solve with working

1:01:53with some of these large like enterprise

1:01:54companies is it's like some of them have

1:01:56like literally said like we don't want

1:01:58any personal information company IP

1:02:00running through a clawed model because

1:02:02they're not convinced that like the claw

1:02:05stealing their data.

1:02:06>> Yeah. So that I mean that's a whole

1:02:08whole another question but for like the

1:02:09personal level just like getting the

1:02:11tools connected and getting it to start

1:02:12doing things for you.

1:02:13>> Yeah. just start with read only access

1:02:16if you don't feel comfortable and then

1:02:17you can escalate it up as you go.

1:02:19>> Nice. Um let's talk about a couple of

1:02:21the tools that are like musthaves for

1:02:23you on your side. I want to share some

1:02:25of the stuff on our side as well too.

1:02:26>> 100%. So if we think back to like our

1:02:29little agent loop of observe, think act.

1:02:31We've tuned up that observe step with

1:02:34context and now we're going to tune up

1:02:35the act step. Right? These are like the

1:02:38top six things I reckon that everyone

1:02:40should have connected up to their agent

1:02:42or claude, whatever you're using. First

1:02:44one is Appify. So Appify is essentially

1:02:47a library of online scrapers and these

1:02:50are like scrapers that people have

1:02:52built. Let me just go into here.

1:02:53>> So it's almost like a third party like

1:02:55sub agent that helps you scraping.

1:02:57>> Yeah, exactly. So it's like a

1:02:59marketplace. There's thousands of

1:03:00different scrapers you can use here. Oh

1:03:01wow. So you see on Ampify, right, we've

1:03:04got Instagram profile scraper, YouTube

1:03:06transcript scraper, Reddit post scraper.

1:03:08>> Wow. This eliminates like some people

1:03:09are using like Grock as an LLM under the

1:03:12notion that it could scrape Twitter/X,

1:03:14but now you can use Aify.

1:03:16>> Exactly. We've got LinkedIn post

1:03:18scraper, Google Map Scraper. There's

1:03:21like Craigslist scrapers. You can get

1:03:23like get lead scrapers, email scrapers.

1:03:25>> These are all individual MCPS.

1:03:27>> No. Oh, it's just a

1:03:30>> and can use what the check. [ __ ] I

1:03:33never heard of this one. Is it? It is.

1:03:35Everyone needs it. Like literally

1:03:37there's there's almost nothing that

1:03:38Claude like can't scrape off the

1:03:39internet for you with Ampify connected.

1:03:41>> Damn.

1:03:42>> So, this is one that no matter what

1:03:43tools you use, everyone I think needs

1:03:44this. They've got a pretty generous free

1:03:47plan. It's like a five Yeah, like $5 a

1:03:50month worth of usage.

1:03:52>> Yeah, we're going to throw another zero

1:03:53too behind that. This is crazy. Holy

1:03:55[ __ ] So, Ampify, everyone needs to have

1:03:58that one set up. The next one is Fire

1:04:01Crawl. So, by default, like pretty much

1:04:02all these agent harnesses as part of the

1:04:04axe step, one of the tools they all have

1:04:06baked in is a web search.

1:04:07>> But this is super like simple and

1:04:09primitive. It's like if you were to ask

1:04:10the agent like, "What's the price of

1:04:11Bitcoin right now?" It can just web

1:04:13search and get the price. It's like more

1:04:14information gathering.

1:04:17when it comes to like things like, "Hey,

1:04:19scrape the branding for this website or

1:04:22like go look at this website and tell me

1:04:24like why they're or like do a CRO audit

1:04:27of my website."

1:04:28>> Less subjective, more subjective, more

1:04:30like context strategy type stuff. Not

1:04:32like what's the this is the question.

1:04:33This is the answer. It's more

1:04:35>> No, no. I I would actually say it's it's

1:04:36more that like the web search it can't

1:04:39look at the website. It can't actually

1:04:41like Does that make sense? We

1:04:43>> can't extract raw data almost.

1:04:45>> Yeah. Yeah. Exactly. Like web search is

1:04:46just to like find like facts off the

1:04:48internet.

1:04:49>> Yeah.

1:04:49>> Whereas fire crawl allows it allows it

1:04:52to actually like read the page not just

1:04:55search it like look at the page. So it

1:04:56can extract like it can tell what the

1:04:58branding colors are. You could say like

1:05:00do a CRO a order of the page and then it

1:05:03can use firecrol.

1:05:04>> That's the first thing my brain went to

1:05:05from a CRO and merchandising perspective

1:05:06from an e-com store. That probably is

1:05:08big.

1:05:08>> Yeah. Exactly. So, you know, if you if

1:05:10you said if if you if you didn't give

1:05:11Claude Firecrol and it just had default

1:05:14web search and you're like, "Here's my

1:05:15ecom store. Like, run a CR order." It

1:05:17can't it can't see the store. It can

1:05:20just like maybe like search some

1:05:21information about your website, like

1:05:24what you sell or like a little about

1:05:25section.

1:05:27With with with Firecrol, it can actually

1:05:29like look at your page, scrape the

1:05:31entire thing. It can see like what

1:05:32buttons you have and where, and like,

1:05:34you know, it could see like, you know,

1:05:36your buttons are the same color as the

1:05:37background. like first fix, make the

1:05:39buttons like brighter. So like stuff

1:05:41like that, it can give Claude like the

1:05:43ability to actually like scrape websites

1:05:45properly and across across the default

1:05:47web search, fire crawl, and amplifier.

1:05:48There's like pretty much nothing that

1:05:49Claude like can't do for me across the

1:05:52internet.

1:05:52>> And this is just mitigating the back and

1:05:54forth of just bottlenecks. It's just

1:05:55getting you more and more and more

1:05:56context.

1:05:57>> Yeah. Yep. And it's less it's less um

1:05:59human in the loop steps, right? Like if

1:06:00you didn't have ampify or fire

1:06:02connected, it used ampify before when we

1:06:05did the scrape my Instagram account

1:06:06example. Damn.

1:06:07>> So, like if I didn't have appy

1:06:09connected, it probably would have said

1:06:10like I can't connect to Instagram. Like,

1:06:12can you send me the videos or something?

1:06:14>> Context for people out there listening

1:06:15too, like you don't have to say

1:06:16anything. This is embedded in the brain

1:06:18on the LLM platform.

1:06:20>> Yeah. Yeah.

1:06:21>> Damn.

1:06:21>> You can think of like you can think of

1:06:23our um really capable employee that

1:06:26we've just hired. You can think of him

1:06:27as he's got his desk, right?

1:06:29>> Yeah. and he's got next to him, he's got

1:06:32a bookshelf full of skills which we're

1:06:33going to get to, but he's also like got

1:06:35the tools like laying around like he

1:06:37might have like uh, you know, he's got

1:06:38his computer. He can see like he's

1:06:41almost got this up, right? If we go

1:06:43applications,

1:06:45you can imagine that our ultra capable

1:06:46employee has this screen up, right? So,

1:06:49anytime we give him a task, he doesn't

1:06:50you don't need to tell him go use

1:06:51Ampify. It's like if you tell him, "Yo,

1:06:53scrape this brand's Instagram page and

1:06:56give me the top best performing videos."

1:06:58>> He's going to just pick the right.

1:06:59>> Yeah. He's just going to look at his

1:07:00apps. Okay, cool. Like, I should

1:07:01probably use Appify. And then he's just

1:07:03going to go use it. So, you don't need

1:07:04to actually specify it.

1:07:06The next one is Composeio. Now, this is

1:07:08like some absolute source I'm going to

1:07:10put you on, right? Some quick backstory

1:07:12about how I found out about this

1:07:14wonderful tool. I had everything like

1:07:16baked into Claude. I had all my

1:07:18connectors built into Claude like I

1:07:19showed you before and Claude went down

1:07:22for an hour the other week. Like the

1:07:23servers were down and I had like so much

1:07:25work to do. I'm like, "Oh my god, I I

1:07:27and I and I can't work without Claude

1:07:28now." So I had to open up like Catgt's

1:07:32codeex to like work in there. But and

1:07:34and the way that my folders set up, you

1:07:36know, we went back to the demo folder

1:07:38before. I've got like this OS folder

1:07:40stands for operating system where I work

1:07:42out of which I'm going to show you guys

1:07:43after this. So the way that that's built

1:07:47is I own all the context files. I own

1:07:49all the skills. They're not embedded in

1:07:51claude. I can open that folder up in

1:07:52codeex and keep working, right? That

1:07:54that part was fine. The issue that I

1:07:57found was I had like 50 different tools

1:07:59connected to claude. And then I started

1:08:01to get to Codex to work and it like

1:08:04didn't have any of my tools. So I

1:08:06started to go through and add in like

1:08:07all these tools and and I was like there

1:08:09must be a better way. That's how I came

1:08:11across Composio. So what Composeio does

1:08:13is we can open this up. You can see

1:08:15connect apps. They have way more than

1:08:18than the Claude desktop app. They have

1:08:20thousands of tools that you can connect.

1:08:23Same way with Claude, like I showed you.

1:08:24It's one click login. You click connect,

1:08:26sign in. They've got everything under

1:08:28the sun here. And then

1:08:29>> is this like an MCP for MCPS?

1:08:31>> Yeah, it's like an M. Yeah,

1:08:33>> it's like an MCP. It's like a bridge

1:08:34basically.

1:08:35>> Yeah. Yeah. So I will connect all my

1:08:38tools in. And another thing, another

1:08:40thing is in the Claude desktop app, you

1:08:41can connect Gmail, right? And log in and

1:08:44you've got Gmail connected. But if

1:08:45you've got three Gmail accounts, you

1:08:46can't connect all three in Claude. Pain

1:08:49in the ass. And which most people have

1:08:51like multiple email accounts. So in

1:08:52here, you can see I've got three active

1:08:54Gmail connections. That was another

1:08:55thing that I liked about it.

1:08:56>> So in short, you're basically saying

1:08:57that connect all of your MCPs to

1:09:00Composeio and then if you're in aio

1:09:02>> and then you and then Composio is like

1:09:04that single MCP with all the MCPS

1:09:06underneath it.

1:09:06>> Damn. So you can see here in in Claude

1:09:09in connectors Composeio. This one won't

1:09:11show up as a browse connector. You'll

1:09:12have to click add custom and then add

1:09:13the URL which I'll show you how to do

1:09:15later.

1:09:17>> And

1:09:18yeah, we can see I've just got

1:09:20Composeio. So I could like theoretically

1:09:22just have that. I don't need the others.

1:09:24You can even connect the other ones I'm

1:09:25talking about ampify and firecrawl. You

1:09:27can connect them through composio.

1:09:29So

1:09:30>> damn

1:09:32>> composio that means like for my setup

1:09:35trying to like be you know robust and

1:09:37protect myself against like platforms

1:09:39going

1:09:39>> you need to go to your bunker in

1:09:41Argentina cuz the world is crumbling you

1:09:43can just you can just hit up composio

1:09:44and just bring all your [ __ ]

1:09:46>> the MD files the skills of composio

1:09:48>> [ __ ] so when I in future now if I want

1:09:51to work in codeex I don't have to

1:09:52connect all these tools again I just

1:09:53have composio and then if I want to try

1:09:55out let's just say like complexity

1:09:57computers is another one I haven't

1:09:58really used it much just another car,

1:10:00another different flavor of the same

1:10:01thing.

1:10:02>> But if I decided I wanted to use that, I

1:10:04could just I'll just compose it to there

1:10:06one time. Yeah. And then it's

1:10:07everything's hooked in. So for everybody

1:10:09out there listening, I want to really

1:10:11really stop and think like where my

1:10:14brain used to go and this is the problem

1:10:15that I used to have with my team. It's

1:10:17like, okay, I've never been on codeex.

1:10:20So if Claude went down, I'd have to go

1:10:22elsewhere. Like I don't even know if

1:10:24it's as simple as just clicking the

1:10:25connector to get Composeio, but you

1:10:27could literally just go into chat GBT

1:10:29and just say, "How do I connect

1:10:31Composio?" And I think everyone just

1:10:33needs to understand that all the

1:10:35answers,

1:10:36>> yeah,

1:10:36>> all these in between the lines answers

1:10:38are literally in the LMS. You just tell

1:10:40them this is what I need done and they

1:10:41can do it for you.

1:10:42>> That's how I learned all this.

1:10:43>> Yeah.

1:10:44>> Like the only like all this stuff I'm

1:10:45sharing with you today, I just it's

1:10:47pretty much all learned from asking

1:10:48Claude.

1:10:48>> Yeah. People need to go through the

1:10:50exercise of like if you're not a

1:10:52technical person and you'd have to bug

1:10:54somebody on your team to do something

1:10:56small for you,

1:10:57>> force yourself to not bug them. And what

1:11:00may take you or them a minute, two

1:11:02minutes,

1:11:03>> yeah,

1:11:03>> it'll take you 5 minutes, 7 minutes, 10

1:11:05minutes if you're not that technical,

1:11:06but you will get to the answer. So, it's

1:11:08more just like

1:11:10>> figuring this [ __ ] out by yourself is

1:11:11huge.

1:11:12>> It's just like, yeah, massive unlock to

1:11:13get in the habit of just asking Claude

1:11:15any stuff. It's like, how do I do this?

1:11:16Like just just ask what it will tell

1:11:19you.

1:11:19>> 100%.

1:11:20>> And if you start following the steps and

1:11:21get stuck and you just screenshot this

1:11:23error, paste it and call it. Yo, hit

1:11:25this. Like what do I do? Like

1:11:26>> that is a what you just said right there

1:11:28is a bar on its own. That's a cheat code

1:11:30on its own. It's just doing that. It's

1:11:31just screenshot. This is where I'm at.

1:11:33Can't get past it. What do I do?

1:11:34>> Let's keep going. So these Chrome Dev

1:11:36Tools.

1:11:37>> So the next one's Chrome Dev Tools. So

1:11:39this is a an MCP built by Google Chrome.

1:11:42Maybe we can connect that through

1:11:43Composer. I don't know. I haven't tried.

1:11:44But essentially what this does is it

1:11:46allows Claude to spin up its own browser

1:11:49if that makes sense. So like I use this

1:11:50a lot if you're building like websites

1:11:52or apps and Claude like wants to check

1:11:54its own work cuz you know you might say

1:11:56like build a website you know these

1:11:58colors make it look nice the brand

1:12:00guidelines it might build it and send it

1:12:02to you and it's just like wrong. It's

1:12:04not at all like what you wanted. Chrome

1:12:06Dev Tools is like one I use that allows

1:12:08Claude to like be able to check its own

1:12:09work in a sense. Like it can spin up a

1:12:11browser and look at the website it's

1:12:12built and go, "Oh no, that's not good.

1:12:14I'll keep working on that."

1:12:15>> As opposed to like in platform or a PDF

1:12:17file. It's like an actual browser.

1:12:18>> Yeah. It's like an actual browser. Yeah.

1:12:20And like another thing that I use it for

1:12:21is I don't know another example the

1:12:23other day I saw on X someone was putting

1:12:25together like a register for like

1:12:26founders in Australia like almost like a

1:12:28a phone book of of founders in Oz. And I

1:12:32clicked I was like this is cool. I'll

1:12:33sign up. And then it was like a, you

1:12:35know, bunch of questions and it was

1:12:37going to take me like probably 20

1:12:38minutes to sit there and answer them

1:12:39all. So I just like got the link, pasted

1:12:42it into court. I'm like, yo, can you

1:12:44apply to this website for me? And I just

1:12:45kept doing other work. And then because

1:12:47it has all the context about me, it knew

1:12:48exactly how to answer all the questions

1:12:50perfectly and in my voice and but like

1:12:53without Chrome DevTools, it couldn't

1:12:54have had a browser to like go and

1:12:55actually enter it in. Little bit

1:12:57different to firecore. Firecol is like

1:12:58more scraping, but this is actually like

1:12:59it can take actions on a browser. So,

1:13:02Chrome DevTools is a great one to get

1:13:04set up. These are pretty much all free,

1:13:05by the way. Another one is Playriter.

1:13:08This is one I use. Same vibe as Chrome

1:13:10DevTools. It's just another browser for

1:13:11Claude to use. Chrome DevTools spins up

1:13:13a fresh browser. So, it doesn't have any

1:13:14of your sites logged in, right? So, if

1:13:16you ask Claude like go send 100 LinkedIn

1:13:18connections to this list and it's just

1:13:21got Chrome DevTools. It'll pop up the

1:13:22browser and it'll be like LinkedIn login

1:13:24and it'll just get stuck.

1:13:25>> Whereas, and you might log in on that

1:13:28browser. Yeah. But then in a week's time

1:13:30if you go to do the task it spins up a

1:13:32new browser again it doesn't have the

1:13:33login saved. So playrider just allows it

1:13:35to control your current browser

1:13:38>> and that like hypothetically like that

1:13:40could automate like one password with

1:13:42filling in the passwords advanced fields

1:13:43for like your address where you want to

1:13:45ship it to stuff like that

1:13:46>> on on one password real quick. That's

1:13:47another one I've connected one password

1:13:49to claude. So

1:13:50>> risky business bro.

1:13:51>> Yeah

1:13:51>> that's a risky one.

1:13:52>> Yeah it is.

1:13:53>> I don't recommend that for people out

1:13:54there. That's that's like the last

1:13:55that's the last level.

1:13:56>> Yeah. What happens though when it's on

1:13:57my computer and it's like getting a

1:14:00password, it pops up with the Touch ID

1:14:02and I have to like approve it with my

1:14:03finger. So

1:14:05>> I stand corrected. I like that.

1:14:06>> Yeah. Yeah. That that's kind of cool. I

1:14:08feel like Iron Man or something.

1:14:09>> Someone might come for you and drop off

1:14:10your finger. [laughter] That's a

1:14:11different story. Higfield's the chosen

1:14:12one.

1:14:12>> Higsfield's the last one. So Higfield is

1:14:14just like,

1:14:15>> dude, they're they're dominating.

1:14:17They're crushing the AI content space.

1:14:19Like they like

1:14:21>> their playbook is just like I don't

1:14:22know. It needs to be studied. like they

1:14:24they've just done an incredible job with

1:14:26like influencer marketing and like just

1:14:28they've made themselves at the name as

1:14:30the top place to go for like AI content

1:14:33like images and video.

1:14:34>> Y

1:14:35>> so you know

1:14:36>> and that at its simplest form guys is

1:14:38you just prompt it and it creates images

1:14:39and videos for you like at its simplest

1:14:41form.

1:14:42>> So that's a great tool in itself. Like I

1:14:43I have some buddies that like live in

1:14:45there and make some like insane AI

1:14:48content like UGC videos and static ads

1:14:50and stuff but by default Claude does not

1:14:53have a way to generate images. So like

1:14:55Chad GBT does they've got an image model

1:14:57whereas Claude doesn't. So this is

1:14:59especially one that's useful for Claude.

1:15:01I mean even CHBT doesn't have video

1:15:03generation whereas Higsfield is

1:15:04basically like the composio but for

1:15:07image models. So it basically pulls in

1:15:10Google's top models, CHBT's top models,

1:15:12like all the Chinese ones that are

1:15:13getting cracked as like it pulls them

1:15:15all in and allows you to just access

1:15:16them through one MCP with Claude. So to

1:15:18show you the Higsfield MCP in action,

1:15:20this is just a demo. I don't I don't

1:15:21work with Bloom at all. Shout out Greg.

1:15:23He's

1:15:23>> shout out Uncle Greg. Great episode. I

1:15:24don't know if it's going to be it should

1:15:26probably be the episode before this.

1:15:27Absolute legend.

1:15:28>> Yeah, I had the pleasure of meeting him.

1:15:29He's a great dude.

1:15:30>> Great.

1:15:31>> So Bloom Creing Gummies. I just gave it

1:15:33a product image and it's like yo scrape

1:15:35competitors ads running in the meta ads

1:15:37library and generate 20 static image ads

1:15:39using the Higsfield MCP and then it used

1:15:41firecrol to scrape the meta ads library

1:15:44and then it's connected to meta as well.

1:15:46This is like an official connector that

1:15:48Met have released so you can get your

1:15:49agents to like do ads. Another great one

1:15:50I recommend. But the Higsfield one,

1:15:52right? You can see it's gone through and

1:15:55it started like generating static ads

1:15:57with Higsfield. And like these are these

1:15:59are pretty good. I mean some of them

1:16:03it's always going to be the case with

1:16:04ads like it's you have to generate like

1:16:06a hundred and there'll be like maybe

1:16:07five to 10 really good ones. So you can

1:16:09see here it's like it's done a bunch of

1:16:10different static ads and like the as a

1:16:13as an ads background. I would run like

1:16:16that's a that's a that's a cool one. I

1:16:18mean the the gummies don't look perfect,

1:16:20but we could just have added an image of

1:16:22the actual gummies and then it would

1:16:23have nailed it perfectly. The like

1:16:26that's a great ad. The product's a

1:16:27little squished. That's a that's a

1:16:29[ __ ] phenomenal ad. That's a great ad

1:16:32too. Creatine for the girls. Like you

1:16:33can see here, it's it can actually like

1:16:34use this MCP to like start creating

1:16:36static ads. And um this is just off a

1:16:39really basic prompt, but if I was trying

1:16:42to like automate as much of the ads

1:16:44process as possible, I'd build a bunch

1:16:46of skills around like static ads and

1:16:47like headline writing and copywriting

1:16:49and stuff, which we'll dive on soon. But

1:16:51that's just to show you like the

1:16:52Higsfield MCP and like some some use

1:16:55cases.

1:16:55>> Love it. Guys, this episode is sponsored

1:16:57by Ketone IQ. I often sit down in this

1:17:00chair for over three hours and stay

1:17:03super dialed with ketones. It's a

1:17:05completely different category of fuel

1:17:07for your brain. Clean, [music]

1:17:09sustained, and no crash. I take a shot

1:17:11before recording, before deepwork, and

1:17:14any session that I need to be sharp.

1:17:15Head to ketone.com/open

1:17:17residency for 30% off your first

1:17:20subscription and a free gift with your

1:17:22second shipment. That's [music]

1:17:23ke.com/openresidency.

1:17:27Guys, this is something that I use every

1:17:29single day. Non-negotiable. Give it a

1:17:32try.

1:17:33This episode is brought to you by

1:17:35Momentous. Most Americans today are

1:17:37walking around deficient in omega-3,

1:17:39vitamin D, and magnesium. Three of the

1:17:41most foundational things in energy,

1:17:43recovery, and sleep. And even if you're

1:17:45already taking supplements, the industry

1:17:47is so loosely regulated that you really

1:17:49don't know what you're taking. I know

1:17:50the team behind Momentous personally,

1:17:52and that's why I trust it. Every

1:17:54product, every batch is third-party

1:17:56tested [music] for NSF, certified for

1:17:58sport. So, what's on the label is

1:18:00exactly what's in the bottle. It's why

1:18:02over 200 pro sports teams trust them,

1:18:04and why [music] my whole stack is

1:18:06theirs. Omega-3, vitamin D, magnesium,

1:18:08and creatine every single day. If you're

1:18:10putting real effort into your health,

1:18:12you want to know that it's actually

1:18:13working. Go to live momentous.com. Use

1:18:15code openresidency for up to 35% off

1:18:18your first order.

1:18:20All right, guys. All right, so we gave

1:18:21you guys those six core tools that you

1:18:23need and then obviously some MCPs that

1:18:25we both like individually and as a

1:18:27company. Let's transition into skills.

1:18:30That's for the newbies out there. Just

1:18:33to retouch on it, it's basically SOPs

1:18:35for for AI. Yeah.

1:18:37>> Why don't you just dive into

1:18:39the playbook on how to utilize skills?

1:18:41>> All right, let's do it. So, I would say

1:18:44skills is probably like the one of the

1:18:46most important things to master with all

1:18:47these agents. So where we're at now is

1:18:49we've got our like really capable new

1:18:51employee and we've, you know, given him

1:18:53all the context. He knows everything

1:18:54about us in the business. Now we've

1:18:55given him tools so we can do work. He's

1:18:57already now going to be great. You can

1:18:59now start giving him work. He's going to

1:19:00do an amazing job. But we want to be

1:19:02able to train him on our processes and

1:19:05the way that we do things. So an example

1:19:08of skills, they're literally just you

1:19:10can think of them as SOPs for AI. The

1:19:12same way if you had a VA that was

1:19:14helping you with like customer support

1:19:16for example, you would probably have a a

1:19:17Google doc built out where it's like

1:19:19customer support triage workflow. Step

1:19:22one, log into monday.com. Step two,

1:19:25navigate to this section in Monday. Step

1:19:27three, find tickets. Do you know, you

1:19:30would outline it really really like like

1:19:32simply to a VA like so it's literally

1:19:34foolproof so they couldn't screw it up.

1:19:36We're literally just taking exactly that

1:19:38but giving it to an agent. And instead

1:19:39of a Google doc, it's just a markdown

1:19:41file. And to illustrate like the problem

1:19:44that skills solve and how useful skills

1:19:46can be, I'm just going to give a simple

1:19:48example. So, let's just say you need a

1:19:50proposal written and you've got your

1:19:53tools and and and context all set up and

1:19:54you start a new session and you're like,

1:19:56"All right, create a proposal for this

1:19:58new client. Like, here's the price,

1:19:59here's the offer, blah blah blah blah.

1:20:01Send it off." it will give you back like

1:20:03a PDF and like very rare it's going to

1:20:06oneshot it exactly how you want. You're

1:20:07probably going to have to go back and

1:20:08say like I don't like how you put the

1:20:09price at the top should be at the bottom

1:20:11instead. Like the logo looks a bit funky

1:20:13like put it in the top left. Like change

1:20:15the colors of the text from blue to

1:20:17black.

1:20:17>> This is the V1 world of chat. Just back

1:20:19and forth, back and forth, back and

1:20:20forth.

1:20:20>> Yeah, this is still like well no this is

1:20:21using agents still. Like you've still

1:20:23got agents like I mean this applies to

1:20:24chat too but but I'm talking about like

1:20:26an agent here. Like let's say you're in

1:20:28code or co-work or getting it to make

1:20:30this proposal. You have to just explain

1:20:32your preferences, right? And eventually,

1:20:34maybe after 15, 20 minutes of back and

1:20:36forth, you're going to it's going to

1:20:37deliver this proposal that you're happy

1:20:39with, right? Great. You had the proposal

1:20:41written, but what happens in a week's

1:20:43time when you want to create another

1:20:44proposal for another client, and you

1:20:48start a new session, and you're like,

1:20:50"All right, create a proposal for this

1:20:51client." Now, it's it's not going to

1:20:53remember any of those preferences. I

1:20:55mean, you you could have it save it to

1:20:57that memory.mmd like we did earlier, but

1:20:59it's it's not the right place for it.

1:21:02You're you're you'd be you'd be making a

1:21:04deacto skill. The the best place to have

1:21:06that is in a skill.

1:21:07>> So, if you if you're starting a fresh

1:21:10chat and said to create the proposal, it

1:21:12wouldn't like remember any of those

1:21:13preferences and you're just going to

1:21:14like not you're going have to go through

1:21:16that whole process of explaining

1:21:17everything over again, which is just a

1:21:18pain in the ass. Whereas with skills,

1:21:21you would tell Claude, you know, save

1:21:23this proposal as like a skill. Save this

1:21:26process that we've just been through as

1:21:27a skill so I can use it again in the

1:21:29future. And Claude would build a

1:21:31proposal skill and it would save

1:21:33everything that that had just done. It

1:21:35would save like all your corrections. It

1:21:37would save like it would probably save

1:21:39the end result as a PDF into the skill

1:21:41as well as like a reference for it to

1:21:43look at. And then it packages that up

1:21:45nicely into into this like skill folder

1:21:48and saves it. So now anytime in the

1:21:50future if you ask it to create you a

1:21:52proposal, it's just going to like whip

1:21:54up this SOP that it's written for

1:21:55itself. Literally a markdown file of

1:21:57step-by-step guide and then it will just

1:21:59follow it to a tea. And then you can

1:22:01basically just set like train your AI

1:22:04agent on exactly how you like things

1:22:06done and exact processes.

1:22:08>> So duplicatable. And I think anybody

1:22:10that's advanced, I mean, how many skills

1:22:12do you have? I would imagine an insane

1:22:14amount. Yeah, we have a lot.

1:22:16>> Yeah. Yeah, I've got a lot.

1:22:18>> As far as this goes, like a very very

1:22:19easy example and we're going to dive