How I Use Claude Cowork to Run My Operations (Full Setup)

Co-work is now available to everybody and I have been using it for the last several months and I've gotten pretty good with it. It's turned into my daily driver. But here's the thing. If you're

driver. But here's the thing. If you're

just getting started with it and if you haven't set it up the right way, you probably don't think it's all that special because it feels like a chatbot in the beginning. Connectors and

instructions and folder files, all that kind of stuff, it really turns into something special. And it's not that

something special. And it's not that advanced and you don't need to go into the terminal. So that's what we're going

the terminal. So that's what we're going to do in this video. By the end of this video, I will walk you through everything you need to know to get up and running with Co-work for the first

time, including all of my latest best practices, including things I just launched this week. So, set aside 30 minutes, download Co-work on the desktop

claw application, and let's get into it.

Now, welcome in everyone. So happy to have you here. If you're here, it's probably

you here. If you're here, it's probably because you're interested in trying. And

I don't blame you because it's really cool. So, let me just give you a lay of

cool. So, let me just give you a lay of the land here. On my screen, we're looking at the Claude Anthropic desktop app, which includes Claude chat, Claude

Co-work, and Claude Code. Many of you don't know that you can use Claude code in the desktop app. You can. And if

you're trying on the web right now, you won't see co-work or code. You'll only

see chat. Okay? Chat is available on the web. Chat is pretty cool, but it's

web. Chat is pretty cool, but it's really like a chatbot. It can write some stuff for you. It can even do some research for you, but it can't actually do stuff for you. That's where co-work

comes in. Co-work literally accesses

comes in. Co-work literally accesses your computer and does things for you.

Now, a simple example of this is just asking Coowork, are there any files on my desktop?

And we'll ask Coowork that. It'll look

at our desktop and it's just going to essentially say, "No, JJ, there are no files in your desktop." But the fact that it could actually view my desktop and tell me that is the unlock that

we're talking about in the video today.

Co-work because it can access your computer. As you can see here, it is

computer. As you can see here, it is able to do things on your behalf. It is

able to browse the internet for you. It

is able to organize files for you. It is

able to create files for you. And with

the use of connectors, it's able to access your data, bring it all into one workspace, centralize it, and use AI on

top of it to help you do more with AI.

So, when we get started for the first time, we're thinking about chat.

Honestly, I'm not even in chat much these days. Co-work, I'm in for all of

these days. Co-work, I'm in for all of my productivity, my daily stuff, my content creation, my sales, my marketing, my customer support, all of my project management. when I'm in co-work and then anytime I need to build

anything I'm in cloud code and when I mean build stuff dashboards internal tools apps SAS etc cloud code so let's get started with co-work and the first thing that you need to know how to do is

go into the bottom lefthand corner here go to settings and go to co-work within co-work we have a couple of different options here first we have the toggle

for dispatch dispatch allows you to from your device access your computer through co-work. Throughout this video, we're

co-work. Throughout this video, we're going to set up your co-work center, but that co-work center is only available on your device. It's only available

your device. It's only available locally. It's not available on the

locally. It's not available on the cloud. But with dispatch, and if you

cloud. But with dispatch, and if you toggle it on, then you'll be able to access your co-work and your computer from the Anthropic mobile app on your mobile device. So, that's definitely

mobile device. So, that's definitely staying toggled on.

After that, we need to go to right here, our global settings. Okay. Now, global

settings are included in every prompt that we sent to Claude. They're very

important. This is where we share with Claude a little bit more about us, who we are, the different types of projects that I'm working like the different terms that I apply and use in my

writing, any of my active projects, and some other things that we want to include. Right now, I have process, I

include. Right now, I have process, I have outputs, and how I want things to be outputed. I have email rules for how

be outputed. I have email rules for how I want to interact with my email. Now,

it's hard to generate it here. And if

you were to just do this right away, you'd probably have a bad output. So,

what I would do is I'd cancel this. I'd

go over to the new task and I'd say, "Help me generate global instructions for my co-work file. I want to map out more about who I am, my style, my

working preferences, who I work with, how I want you to help me with my email, my calendars, and any other things you think you should be included in my

global settings. Ask me.

global settings. Ask me.

So, I just used a voice to text tool there called Whisper Flow. It's able to put it in there quickly, and then I can go in there and start to work with Co-work to generate those instructions.

Now the goal here is co-worker is going to interview me, get the information it needs, and then take that information and combine it into a global instructions that I can copy and then

paste into that global instructions area. So that's step one, getting set up

area. So that's step one, getting set up with your global instructions.

Now that we've set up our global instructions, we need to talk about this thing right here, work in a project. So

I have a lot of different projects to work from. And if I go over here to

work from. And if I go over here to choose a different folder, it pulls up the finder on my computer. Okay. Now,

here's what's you need to know. Co-work

because it's working on your computer is accessing files. Now, by choosing a

accessing files. Now, by choosing a project that it can work in, it will allow it to stay within just the folder and those files and not access

anything more unless it asks for your permission. This does a couple of things

permission. This does a couple of things and it's really important. By clicking

on this folder, you're telling co-work.

Now, if I were to chat here, it would go and look through all the files on that folder to see if they can help me answer the question that I may be asking. Now,

let me look into that folder real quick.

So, I'm going to go into my finder here.

I'm going to go into JJ and projects where I keep all my projects. And that's

arbitrary. It could be wherever you want to save it on your computer. Could be

your desktop, your downloads, whatever.

And then I'm going to go into co-workspace. And then I see right away

co-workspace. And then I see right away that I have a lot of things in here, right? I have my content creation. And

right? I have my content creation. And

within content creation, I have folders for each of the platforms that I create content on and so on. So if I wanted co-work to help me with an expost, I could give it access to this folder up

here. However,

here. However, that's a lot of information that it has to look through just to get to our expost. or I could just give it access

expost. or I could just give it access to the X folder and it could work with that too.

Now, the benefits of this by giving it access to just the X folder, that means that it's going to go into that X folder, read the skill that I have there, and just help me in that very

specific way. That's going to take less

specific way. That's going to take less tokens, meaning it's going to look through less things, and it's going to cost you less money because it's looking through less things. And it's gonna be

very specific with this output because instead of reading many other documents and maybe getting confused of like different directions along the way, it's very specifically being targeted to X.

Okay. So, whenever I create a new project in Co-work or want to work on a new project, it always starts with creating a new folder. So, in order to create a new folder, I'd go back to my

projects file right here and I'd simply go in here and I'd say new folder. And

then I say co-work ultimate setup. Okay. And I'm going to bring that

setup. Okay. And I'm going to bring that folder into projects. And I'm going to go into that folder. And I see I have nothing there. That's okay. We're going

nothing there. That's okay. We're going

to get a setup in this video. So now I'm going to go back in the cloud and I'm going to say, all right, let's go into that folder. So choose a different

that folder. So choose a different folder, projects, co-work ultimate setup. and we're just going to stay in

setup. and we're just going to stay in that folder. Okay. So now we are in that

that folder. Okay. So now we are in that folder and whatever task or anything we ask it to do will be done in that folder. So I'll say, "Hey Claude, are

folder. So I'll say, "Hey Claude, are there any files in this folder?"

All right. Now it's going to bring me there. It's going to ask it. And right

there. It's going to ask it. And right

away on the right hand side, you can see all of a sudden that we are in the folder for ultimate setup. We have these in the instructions here that are actually empty. We don't have any folder

actually empty. We don't have any folder instructions yet. We will come back to

instructions yet. We will come back to this. And then we see, hey, there's no

this. And then we see, hey, there's no files yet. Do you want me to get

files yet. Do you want me to get started? And that's where we will get

started? And that's where we will get started today. We'll see. Yeah, Claude,

started today. We'll see. Yeah, Claude,

I want you to help me set up this folder. I want to have a tople file that

folder. I want to have a tople file that shares more about who I am, what this folder is, or what this project is, and how we can make the most of it in terms of like my style and whatnot. So feel

free to interview me and then when you're done create a document that you could reference when working with me within this project or folder.

Now I keep coming back to projects.

Simply put whenever I create a new folder for me that's like a project right and so if we go back to the list of recents here we see that we have these individual chats. Now the problem

with these individual chats is this chat right here does not know what this chat was. So if I go to start a new chat,

was. So if I go to start a new chat, it's not going to have the context of what the previous chats were about.

That's where projects comes in. So if I were to create a new project, and if I were to create this project from the folder that we had just created, which

is the co-work ultimate setup, then we would start to share memory between those chats. Let me show you to help me

those chats. Let me show you to help me manage my day-to-day as a business professional as a that's fine too. All right. So now

what we have is we have a project created and within that project we're still in that co-work folder. Okay. Now

over here is that chat that I'm working on and so it's asking me some questions.

All right. Let me show you the lay of the land of the projects and then I'll come back to tying in that chat back into this project and show you the reason why it's important. When we're in a project here, we can ask questions

just like we did before. When we do ask a question, there would be an agent that shows up here and starts working. Now,

you can ask multiple questions and then you have multiple agents showing up below. And then you can see from the

below. And then you can see from the project view your agents working for you using your computer browsing the web doing stuff for you and manage all your

agents from here. That turns you into an AI orchestrator. That's really cool.

AI orchestrator. That's really cool.

Other than that we have project instructions here. Project instructions

instructions here. Project instructions we can treat very similar to global instructions.

Hey Claude, help me set up project instructions for the co-work ultimate setup. These project instructions should

setup. These project instructions should focus on helping me as a business professional navigating my day-to-day with my working preference, my styles,

my communication preferences, etc. So, generate a clean and compact instructions that I can copy and paste into this folder here. So, it's going to go off now and create the instructions

to help me. So, I can just copy and paste there. Generally AI is better at

paste there. Generally AI is better at prompting or giving instructions than I am. So I just a ask AI to write these

am. So I just a ask AI to write these instructions for me. Next we have scheduled allows you to run reoccurring things on a schedule. So you have you

can run it on manually. You can run it hourly, daily, weekends, weekdays. And

this could be anything along the lines of check my inbox and then let me know if I have anything important or review my calendar, check my inbox, check my Slack at the beginning of every day and

then give me a report and maybe a suggestions of how I should approach my day based on priority. All of that stuff you can do and you would just write in the prompt there and Claude can do that

for you on a daily basis. The reason why it's even better in projects than it is over here as a general schedule task is because it will take all the information

of this project, the files within it, the project instructions, your recent chats, and use that as context to give a better result for the scheduled task

because it is now within a lot of files that are helpful that can help it get a better result.

If you were to set up a scheduled task over here, this is like very isolated.

This is very much, hey, I'm just going to go off and do this thing that you want me to do, but I don't have much of other information. So, it's just like

other information. So, it's just like you just randomly hired a contractor to go do something. Whereas, if you're in a project and you go in and set a scheduled task, this is like an employee

that knows more of this project that can go off and do things for you. So,

outside of that, we have the folder and the files within this folder. Now, right

now, this is still empty. All right, I do want to finish out this instruction over here to create our first file. But

before I do that, let me finish my thought here.

This is a folder on your computer. And

at any time, you can click on it and it's going to show the folder on your computer, right? You can also go in

computer, right? You can also go in there and easily add files into that folder. I like to add a lot of files

folder. I like to add a lot of files that add relevant context. For example,

I recently co-hosted or was a lead instructor at a workshop that we led with Anthropic, Anthropic Claw. And for

that workshop, I had to help with the event logistics, etc. And Anthropic sent us a lot of information about a good way to host the event, what the curriculum should be, everything. So, I created a

folder on my computer. I took all the documentation that Anthropic provided and put it in that folder. And then I said, "Hey Claude, look through all this documentation, understand it, and write

a tople summary of your findings." And

then going forward, I had this project on my computer for Anthropic Workshop.

And whenever I had any kind of questions, I could just go here and it would be able to reference all these files on this computer and be able to help me with that context. That's the

benefit of having all those folders and files for that project. It's getting

that information. Say you want an Excel sheet of analytics data, you want a report that you know for your board, you want a presentation, whatever. You could

put it in that folder. Claude will read it and use it to help you respond. Okay.

So, let's get back to over here. We want

to create a project level instructions or a top level document for this folder.

It's going to be the first file we put in there. And generally, it's just going

in there. And generally, it's just going to be like almost like a read me or something that Claude should have right away whenever it works with this project to understand more about what this project is, the players involved, all

that kind of stuff. So, Claude is now asking me a series of questions. What is

the primary work purpose of this workspace? I'm going to have this as my

workspace? I'm going to have this as my main workspace.

Which task do you want to do? I'm going

to do writing content. I'm going to I'm going to do all of the above. Okay. What

does working well together look like? I

wanted to think with me and then I would want it to do stuff there. Right? I'm

not looking for just like a quick doer.

I want someone that was really helping me think through stuff the way I would think at a deeper level to make sure that the approach that we set off on is the right approach. Right? So now it's going to work through here and look at

all the information I gave it and create that doc for me. And when I mean create the doc, it's literally going to go create a file on my computer and save it to that folder. Pretty cool. Now I also

want to take this because we're working with this chat for this project and move it into my project to the co-work ultimate setup. So now if I go back to

ultimate setup. So now if I go back to projects co-work ultimate setup I can see that I have a chat here for creating the files in the folder and then I have the other chat for the work

instructions. So if I click in the work

instructions. So if I click in the work instructions I can see that it gave me this work instructions here and so I'm going to copy this.

Go back to ultimate setup. go to my instructions and paste these instructions in. Okay, now this project

instructions in. Okay, now this project is going to have much better instructions of who I am, what this project is about, and just increase the

quality of every chat that I ask it through this project. Also, look at this. We now have a file within our

this. We now have a file within our folder that Co-work created for us. If I

click on it, look at this about this workspace. This is JJ's main folder. Who

workspace. This is JJ's main folder. Who

is JJ? I run the Cloud Code community. I

run the 10X build a channel. My name is JJ Angler. I have a podcast called This

JJ Angler. I have a podcast called This Week in AI that I'm bringing back shortly. And I'm working with 10X in a

shortly. And I'm working with 10X in a multitude of ways. And so here are the people I work with. Here's my ICP. All

that kind of stuff. Really good stuff.

So we're cruising here. Okay. Now we're

ready to introduce you to connectors.

So, we got co-work set up global instructions. We've created our first

instructions. We've created our first project instructions. We've created a

project instructions. We've created a folder and files for that project. We're

up and running. We got a couple of big things that we need to talk about still.

Connectors, skills, plugins, but right now we're going to talk about connectors. Connectors are available

connectors. Connectors are available right here. And right now, you could see

right here. And right now, you could see that I have these connectors available.

If I wanted to add more, I can go to add connector and I can see all these other connectors. So, for example, let's look

connectors. So, for example, let's look at linear and see what options it might give me. Say if I wanted to approve

give me. Say if I wanted to approve that, I could approve that. But right

now, I'm going to pass and I'm going to go back to here, go back into my connectors. I'm going to go manage connectors and I'm going to go into each of the connectors I have right

now. Okay, so customize connectors. And

now. Okay, so customize connectors. And

I click into it.

So for the connectors that I have installed, they allow me to do a couple things. If I click on Gmail, for

things. If I click on Gmail, for example, this is my Gmail connect. This

Gmail connector allows me to do the various things. It can read only. It can

various things. It can read only. It can

get my profile, get my drafts, my labels, my emails, etc. And specifically, I am allowing permissions saying always allow. So anytime you want to use these tools, just allow it. I'm

going to let you do that. If I wanted to oneoff, just say, hey, you know what?

This is not allowed. I can go block it or I can say if it needs approval and you want to ask me every time or human in the loop that will work too. After

that we have a different tool. It's the

right tool. This is creating an email draft. Again I'm allowing it to create

draft. Again I'm allowing it to create an email draft. So what connectors do allows you to take in external data not only to bring it in but for a co-work to

do actions with that external platform.

Gmail means co-work can read my emails and then create drafts for me to review and then go in to Gmail and send. Okay.

Now for Google Calendar, it can find meeting times. It can find event

meeting times. It can find event details. It can read my events. It can

details. It can read my events. It can

suggest meetings. It can place meetings.

All that kind of stuff. And these two work together. So, I could say, "Hey

work together. So, I could say, "Hey co-work, go check my calendar. See what

time I'm meeting with Alex and then email him and see if there's another time that we can meet on Friday instead." Co-work will literally go

instead." Co-work will literally go check my calendar, check Alex's calendar. Write an email that is a good

calendar. Write an email that is a good time to meet with Alex next. And then

tell me, hey, it's ready for you to review in your draft. And you will do it. That's the benefit of co-worker.

it. That's the benefit of co-worker.

That's the benefit of connector. taking

external data and bringing it back into co-work so AI can work on top of that data in a unified manner to do more for you. Cool. I also want to shout out

you. Cool. I also want to shout out desktop right here with Chrome. This

essentially means that if you enable it that co-work can log into Google Chrome under your profile and access Google Chrome, can tab around Google Chrome,

can browse around Google Chrome and just generally use your browser. Cool. So

that's connectors. They're really

powerful and you can use a ton of them.

And if there aren't any listed, you can actually add your own custom connector as well. Generally, it's a suite of API

as well. Generally, it's a suite of API endpoints, connections to other platforms that include tools that allow you to read data and to write or to do things with that data. Cool. Let's move

on to skills.

So what are skills and how are they different from plugins? A skill is almost like a standard operating procedure. Something that you do over

procedure. Something that you do over and over again that you can give to AI to do for you. An example of this is every week I write a newsletter. It's

called the Ultrathink newsletter. It's

an AI newsletter. Best in the business.

I'll put the links in the description.

Definitely check it out. For this

newsletter, I use a skill that does a series of steps to help me get to a first draft quicker. It doesn't fully do everything for me because I don't want

it to, but it helps me get to that first draft quicker. So, instead of taking

draft quicker. So, instead of taking four hours to do my research and my prep, it takes me 10 minutes. And then

I'll go in there and finalize and really add my human touch to it. But the

process is it's a skill called Ultraink.

And whenever I go to invoke it, it has a series of instructions that I've told it to do that runs. First, it asks me what I want to write about. And then a after it's done asking me what I want to write about, it does market research of any

kind of current topics that are happening in a AI that week. And then

it'll actually write the draft and say, "Okay, JJ, it's ready for your review."

And then that's when I get in there and review it. It also will have a

review it. It also will have a self-evaluation check. And this is

self-evaluation check. And this is something that I built into the skill.

The self-evaluation check is a checklist of like 10 things that each draft really needs. And that way once it finishes the

needs. And that way once it finishes the draft, the evaluation step will run. And

if it passes the evaluation step, it will send it to me. If it doesn't pass the evaluation step, it will rewrite the article until it does. That's a

self-improving loop that is very important and helps you get to a better result faster. But that also means you

result faster. But that also means you need to define what a good result is.

Okay, one of the ways I like to define what a good result is by showing it examples of good results. Now, let me show you how I can do this using one of my connectors. So, if I go into my

my connectors. So, if I go into my connectors here, I see that I have my Gmail connector, right? So, I'm going to

ask Claude to create a skill that helps me write my emails in my authentic tone.

And in order to do this, I want you to review the last 50 emails that I sent and then to analyze those for my writing patterns,

my tone, and to use that to create instructions for this skill to create a successful email on my behalf. So this way, whenever I ask you to write an email,

you're using this tone. Also, I want this skill to also help me to triage my email inbox. So, if I want ever want to

email inbox. So, if I want ever want to invoke this skill, it's going to help look through my email inbox and it's going to filter out some of the spam.

It's going to call attention to the things that might be important and it's never going to delete anything. Although

it could archive things, but it's very important that whenever it does archive, it responds back letting me know what it archived so I could see everything that it's doing and I have that transparency.

So, this skill does two things. It helps

me draft better emails in my tone and it helps me work through my inbox in a more efficient manner. Claude, please go and

efficient manner. Claude, please go and create that for me now. And creating a skill is as simple as that. And Claude,

I could see now that I have this agent running for me. These other two agents have finished up. If I want to go into that agent, I can click on it within that. It's going to work and it's going

that. It's going to work and it's going to might ask me some questions and then it's going to build the skill for me and then it will become available whenever I do a backslash. I'll see all my skills

right here. And then I can go in and

right here. And then I can go in and access that skill. For example, I've already created the skill. It's called

the email voice writer. And I could just tag that in and say, "Help me write an email for Alex." And it's going to say X, Y, and Z. And then it will go and

draft that email for me in my tone.

Right? So, it's a way for Claude and AI to do repeatable things, very high quality in a consistent way. If I go

back to my folder structure here, you'll see that within my original co-workspace, I have folders for each of my platforms.

So, for example, in X, I have a skill for X. In LinkedIn,

for X. In LinkedIn, I have a skill for LinkedIn. And the

reason being is the way that I write on both of those platforms is different. My

tone, my voice, all that kind of stuff is different. So, I was able to feed it

is different. So, I was able to feed it my previous expost, my previous LinkedIn post and say, "Hey, look at the way I've been writing. Look at my best performers

been writing. Look at my best performers and create a skill to help me write in LinkedIn in a more consistent way." Now,

whenever I need to write a LinkedIn post, I can just go in and do a back slash LinkedIn or something like that and say, "Help me write a post about X, Y, and Z." Then, it's going to use my writing style to help draft that post

for me. Right? It's not just content

for me. Right? It's not just content creation. It could be a skill to create

creation. It could be a skill to create a presentation deck. It could be a skill to as a thinking partner almost like a mentor, a CEO, coach, anything like that. Skills can be used in many

that. Skills can be used in many different ways, and it's usually like one-off tasks. They're really important.

one-off tasks. They're really important.

And whenever I do something like more than once, I'm usually creating a skill for it.

So, let's talk about plugins now. What

is the difference between a skill and a plug-in? Plugins are like a suite of

plug-in? Plugins are like a suite of things like a workflow almost like an employee like marketing. This plug-in

will help you with many aspects of marketing. It could help you draft a

marketing. It could help you draft a blog post, do a multi- channelannel campaign, review content around your brand voice. It has many skills within

brand voice. It has many skills within it. So, it's almost like you created a

it. So, it's almost like you created a marketing team worth of skills and bundled it all together to create one marketing team plugin. Also, it's almost tax season right now. Turboax has a

plugin to help you do your taxes. If you

want to do designing, finance, product management, this is a suite of tools and skills that you can use. Look at all these skills in here to help you do

more. So, it's like a bundle rather

more. So, it's like a bundle rather skills are like just a one-off thing.

Plugins are a bundle. And right away, I can also see that plugins can access connectors. So look at this plugin can

connectors. So look at this plugin can access Slack linear asana Monday everything that a PM might want to manage can access all of those and it could do all the basic PM stuff and I

can invoke that anytime to help me do this in a more consistent way. Also

let's say I download this plugin. What

if I wanted to make updates to the plugin? I could do that as well. I could

plugin? I could do that as well. I could

update the plugin to be more personalized to my liking. Plugins are

really powerful. Whether you have something out of the box like this or if you want to build your own personal plugins like an executive strategist, a deal analysis, you can do that as well.

And you can go over here, you can upload a plugin, you can add a plugin, you can share a plugin with your team. I

recently just created a video showing you how to create or share plugins and skills with your team. I'll post it maybe like right here. Shows you three different ways that you can do that.

Imagine if all everyone on your team could build skills and plugins and then share them together and really you work so much better and faster. It takes a little bit to get there, but if you can,

it's awesome. So, plugins are really

it's awesome. So, plugins are really awesome. And when you pair that with

awesome. And when you pair that with working with plugins and skills in your projects that have access to all your files, your preferences, man, you can start to work really fast.

All right, last thing I want to call out before we get you out of here.

How do you stay within your rate limits?

So, let me just share a little bit about what I'm finding success with and just share it back to you. So, you can subscribe to co-work with the $20 plan, the $100 plan, the $200 plan. I'm

personally on the $100 plan a month. And

I hit co-work a lot and also claude code a lot all day. Sometimes I hit limits, but it's pretty good, right? The $20 a month plan, you're probably going to hit a bit more limits. But let me talk to

you about some ways that you can avoid hitting lines, right? So, first off is your model selector. Opus, it's like your Ferrari. You don't need to drive

your Ferrari. You don't need to drive the Ferrari to pick up milk and eggs, right? So, unless you're doing like

right? So, unless you're doing like advanced research or analytics or analyzing data or like really thinking hard, you don't need to be on Opus. Opus

is really smart. You don't need it for a lot of things. Sonnet 4.6 is my daily driver. That's the default model that

driver. That's the default model that I'm on most of the time. It's really

smart, but it's way cheaper than Opus. I

think it's like 10 times cheaper than Opus. It's fast and still does a great

Opus. It's fast and still does a great job, especially for help me write an email. I don't need a Ferrari to help me

email. I don't need a Ferrari to help me write an email. And then finally, we have Ioku right here. This is really fast and it's still quite smart. It's as

smart as 4.5 sonnet it used to be. So,

like not that far off and it's crazy cheap. So if you're running like really

cheap. So if you're running like really worried about usage limits, I would tell you to like whenever you have thinking stuff that you need done, go into sonnet. But other like transactional

sonnet. But other like transactional stuff, just say in Hayoku, you could turn off extended thinking for those that like you don't need so much. Like

extended thinking is getting more context, taking more memory, all that kind of stuff, more tokens. If you have a big conversation, maybe this could help. If you need really need to do some

help. If you need really need to do some deep thinking, it will help, but don't use it by default.

Also, you see here that I've spun up a conversation for each new thing. Each

new task is a conversation. That's the

right way to go. You don't want to just keep going into the same conversation here and having to do a new thing because what happens is the whole conversation is shared every prompt.

That means there's a lot of information being shared that one isn't applicable.

That kind of blurs what is what you want Claude to know about the task that you're asking. and three takes up more

you're asking. and three takes up more tokens because you're sending more information which means more to process.

It's more expensive and really you only need the last thing. You don't need all this other stuff. So you should be creating a new chat for every kind of

new task you have that will make it so you're only sending the information that Claude needs to know when it needs to have it. Reducing the token size which

have it. Reducing the token size which reduces your cost and it goes faster because there's less to process. So you

get better quality outputs quicker and cheaper. But that's just up to you to

cheaper. But that's just up to you to manage that better. Like back in the day with chatbt, we used to have that one thread of I use this thread to write all of my marketing copy and it was just like you wrote it in the same thread for

3 months or a year. Don't do that. That

is not good practice. You do not want to do that here. This is being replaced with files. If you need something, if

with files. If you need something, if you have that thread that's really important, create an MD file about it summarizing the highle summaries and then Claude could reference it when it needs to reference it, but it doesn't

need to bring in a year's worth of conversation for every single prompt that you have with it. Make sense? So,

this is how you get set up in co-work.

It is awesome. Just don't get stuck at the beginning where you spin up co-work and you don't do any setup because it will feel like a chatbot and it will probably be more expensive and the

quality won't be as good. It takes a little bit of setup. It does 30 minutes, an hour, but that 30 minutes, that hour it's teaching you how to use it. You're

having AI write all this stuff for you.

You're feeding it the information it needs. Don't ever forget or just to take

needs. Don't ever forget or just to take a moment to say, "Hey, Claude, look through my folder. What's going on here?

Create a summary of my folder. Hey

Claude, for like bigger folders, create a workspace map that details all the files I have in this folder and when you should maybe reference those files so if I ever need it to do a deep dive, it

knows where to look already. Think about

it like that. Think about indexing these files. I have this big folder and within

files. I have this big folder and within this big folder I have subfolders and within those folders I might have even subfolders. But each of those allows you

subfolders. But each of those allows you to go to the next level and be more specific with the task at hand. be more

specific about the project you're working on. Being able to be more

working on. Being able to be more specific in that way allows you to get better results and the cost is less. So,

take your time, get set up, create your folders, create your projects, create your skills, your plugins, add your connectors, your global instructions, turn on dispatch, get connected on the

mobile app. You will be 10xing your

mobile app. You will be 10xing your daily productivity before you know it.

If you enjoy this video, please like it.

Please subscribe. I do a video like this every single week showing non-developers how to do more with a I love it. I do

this stuff all the time. I'm doing a lot of trainings right now, not only with Anthropic, with a company I work for, 10X. We're doing a lot of enterprise

10X. We're doing a lot of enterprise training, helping executive leadership teams and just general like leadership and employees get upskilled with these latest skill sets. It's an incredible

time and it's even a better time to be a non-developer that has rich domain expertise that you could take that domain expertise, train AI on it to help

you do more with it at a very deep level. It's financials, taxes, whatever,

level. It's financials, taxes, whatever, you know, and you don't even need to write any code for it because Claude writes it all for you now. So take that deep domain experience where whatever it

is, put it into a system, have AI help you expand it even more. You're going to find great results. So subscribe. I'll

see you next week for our next video.

Thanks for tuning in. As always, I'm JJ and that's it for now. Peace.