Claude is Taking Over: Every New Feature Explained (Full Guide)

For the past 3 years, OpenAI has dominated, and it's felt like ChatGPT was just going to take over the world.

But now it feels like everyone's moving to Claude. Since the beginning of this

to Claude. Since the beginning of this year, Anthropic has been on a generational run, and over the past 2 months, Anthropic has shipped more useful features to Claude than most companies ship in 2 full years. And

these aren't just small updates. We're

talking about AI agents that can use your computer just like OpenClaude, full voice mode in the terminal, background agents that run while you sleep, a brand new Claude Co-work platform, and two

brand new models. And the craziest part is most people have no idea that half of these features exist. So, in this video, I'm going to break down every single new Claude feature, from Claude Code to

Claude Desktop, so that you can understand exactly what's available to you right now and how to effectively use it for your work. We're going to cover computer use, co-work, agent teams,

remote control, schedule tasks, and a lot more. And we're going to cover all

lot more. And we're going to cover all of the features that they've released since the beginning of this year. And if

you're a developer or entrepreneur or just someone trying to keep up with AI, this is the one video that you need to watch to understand why Claude might be the most important tool in the world to learn. And if you like videos like

learn. And if you like videos like these, please subscribe and hit that like button. It helps me out a ton.

like button. It helps me out a ton.

Let's not waste any more time. Let's

dive in. So, before we dive into all of the brand new updates to Claude and Claude Co-work, I want to talk about kind of the the three phases of AI I

think we've gone through since ChatGPT came out in late 2022. And I promise, majority of this video we're going to be talking about all the new Claude updates. I'm going to discuss all of it.

updates. I'm going to discuss all of it.

But I think it's important for people to realize kind of the different phases of LLMs that we've gone through, all the way starting back at the beginning. So,

the first era of LLMs started in late 2022 when ChatGPT ChatGPT was released.

ChatGPT was magic, right? It was this text generator that seemed to know all the answers. Yes, it sometimes

the answers. Yes, it sometimes hallucinated. Yes, it wasn't always

hallucinated. Yes, it wasn't always perfect, but it was a magic technology that could give you essays. It could

write your emails. It could do all these different things, and it took over the world. And so, this was largely

world. And so, this was largely text-to-text, right? You could ask

text-to-text, right? You could ask ChatGPT a question, and it could respond with text. It could also uh you could

with text. It could also uh you could also put images into ChatGPT, and then eventually, you know, late 2023, you could upload files into ChatGPT, and it could not only respond in text, but it

could also respond in code. Like, you

could ask questions about a specific line of code, or you could say like, "How would I create a you know, function or a code page that does X, Y, or Z?" And it could actually

spit out code in raw text. But, what

separated 2023 and most of 2024 from the next era, with the Vibe Coding era, is companies started realizing that you can actually render code directly. Like, you

could render it in the app that generated it. And the first company that

generated it. And the first company that did this was actually Claude. And what

they created was something called Claude Artifacts. And so, Claude Artifacts,

Artifacts. And so, Claude Artifacts, this is a good This is a good segue. All

Claude Artifacts are I'm just going to say, "Please create a React component that explains what Claude Artifacts

are." Claude Artifacts, simply, are

are." Claude Artifacts, simply, are where an AI model is going to generate code, right? In this early ChatGPT era,

code, right? In this early ChatGPT era, it could only just generate a code file.

It couldn't actually render the code file. But Claude Artifacts in May of

file. But Claude Artifacts in May of 2024, if my memory serves me correctly, so almost 2 years ago, it could generate code, and then it would render the code

right here. So, look at this. This is

right here. So, look at this. This is

front-end code, and it generated this, and boom, it immediately opened it up.

You can interact with it. And this was Claude Artifacts. And when this came

Claude Artifacts. And when this came out, this is when I dove in right away.

I dove in headfirst, and I wanted to learn all about this and how I could create apps with AI in the very early stages. And I very quickly went from

stages. And I very quickly went from Claude Artifacts to a tool like Cursor.

So, Cursor had Cursor Composer. Now they

just call it Agent, but you could literally go to Cursor, create a new project, and say, "Create a new app, front-end and back-end, use Firebase."

If you don't know what that is, don't worry, this will all make sense. And

make a full production app that I can use to make money. Right? You could type something in like that, and Cursor would go off, and it would actually attempt to write all of the code, and then you

would see all of your code files. Not

only was it just a front-end, right?

Like it is here with Claude Artifacts, it was a full production app. And so,

that's kind of the unlock that Cursor created. But at the end of the day,

created. But at the end of the day, Cursor was a tool for developers. Cursor

was mainly for devs. And from this spawned many companies. Lovable, Replit,

Bolt, Vibe Code, Emergent, all of these different companies emerged, and these are companies that tried to make Vibe Coding easier for people who had no coding experience or little coding

experience, for marketers who just wanted to create a quick landing page.

And up until recently, now on a lot of these platforms, you can create like full production application. And then, I think in the second half of the year, I

forget when it came out, a tool called Claude Code was released, and this took over the world by storm. Claude Code was a tool that you can use directly in your terminal. I realized that I could use

terminal. I realized that I could use Claude Code to create any app that I wanted to, and I was making internal tools for our company. I was creating landing pages, marketing material, and

even slide decks just of in a few prompts using Claude Code, and it would just store them directly on my computer.

And that's why so many people fell in love with Claude Code. And so, towards the end of 2025, people started to realize that you could just use Claude Code to do a ton of other tasks

autonomously and semi-autonomously. It

could do marketing tasks. It could

respond to emails. It could connect to like most of your other tools and just do tasks on your behalf. And then, very early on in 2026, we saw the birth of

OpenClaude. And so, OpenClaude was the

OpenClaude. And so, OpenClaude was the first AI agent, and the the AI agent harness that powers OpenClaude is called Pi. And it's somewhat similar to Claude

Pi. And it's somewhat similar to Claude Code, except OpenClaude was kind of finagled in such a way that it was meant for general-purpose agents, right? And

so, the This is the third era, right? We

we had this chat era from 2023 to mid-2024, and then we entered the era of Vibe Coding, and now we're in the era of general agents. We're trying to take all

general agents. We're trying to take all of this technology, which allows you to create any type of app, and we're trying to apply that to general agents. That's

what all these companies are doing. Like

OpenClaude, Manis, Perplexity Computer.

We have Co-work and Claude, and obviously OpenClaude. And this is

obviously OpenClaude. And this is becoming such an important race that OpenAI has decided to cut all of their side projects today. Like, an hour

before I'm making this video, OpenAI has fully scrapped the OpenAI Sora application, and they are focusing on creating one unified OpenAI super app,

right? That will combine ChatGPT. So,

right? That will combine ChatGPT. So,

it's going to be ChatGPT plus CodeX plus Atlas, which is their browser. And

they're going to combine all these products in one super app because they need to compete with the Anthropic super app, which is just Claude Desktop.

Because Claude Desktop, as we'll talk about in this video, has become a behemoth of an application. It is

actually insane how good this tool is becoming. And so, the beginning of the

becoming. And so, the beginning of the takeoff of Claude, their desktop app, was when they introduced Claude Co-work, which is Claude Code for the rest of

your work. They

your work. They created a platform that was not like chat, right? You can kind of see it,

chat, right? You can kind of see it, right? They'd created chat. They'd also

right? They'd created chat. They'd also

created Claude Code, and they're kind of going to apply it to Claude Co-work. And

so, Claude Co-work is kind of the midway point between Claude Code and chat. And

let's dig into this a little bit. So,

right? We have chat, right? Where you

could chat with Claude, and then you had code where you can basically have this AI agent that could do anything it wants on your computer. And if you run Claude

Code, right? When I run Claude Code, I

Code, right? When I run Claude Code, I run it like this. I run Claude {dash} And don't worry if you don't understand this. This is not super important. This

this. This is not super important. This

is just illustrating my point here.

Dangerously {dash} skip {dash} permissions. And so, when you run Claude

permissions. And so, when you run Claude like this, it bypasses all permissions.

So, it'll just do things, right? I can

say like, "Completely reorganize my desktop folder. I want you to literally

desktop folder. I want you to literally reorganize everything, upload it to the cloud, I want you to try and create a video using the VO API." Like, my point is is you can get it to do anything that

your like imagination wants you to do.

You can just get Claude Code to do things. And this is good for developers,

things. And this is good for developers, but it's also quite dangerous for normal users, right? Giving an AI model full

users, right? Giving an AI model full access to your computer, especially when it can put your data out on the internet and uh send your data to the wrong person, it's quite dangerous. So, they

created Co-work. And so, Claude Code controls your computer, right? It works

in directories, which are just files on your computer. And the chat just kind of

your computer. And the chat just kind of works on the app, right? Chat doesn't

really touch your computer. It just

works in the app itself. Claude Co-work

creates its own little environment, right? It creates a little sandbox on

right? It creates a little sandbox on your computer. You can think of it just

your computer. You can think of it just like an isolated uh little folder on your computer, and it only touches that folder. So, if we were to go to Claude

folder. So, if we were to go to Claude and we go to co-work and we wanted to create a new project, so we could create a new project, we can choose an existing folder and we could say we're going to

put this in the Riley Brown folder.

We're going to call this marketing.

We're going to create a marketing folder. So, we're going to open this up

folder. So, we're going to open this up and we're going to create this. And so,

this lives within this folder on my computer. It's not actually going to do

computer. It's not actually going to do anything outside of that folder. It's

only going to operate within that folder. You can give context to other

folder. You can give context to other folders on your computer by like adding a certain file. Like you can add files, but at the end of the day it'll stay

isolated, so it's much safer. But Claude

co-work makes it really easy to like connect new things. So, I can hit this plus sign, go to connectors and I could hit manage connectors. I could go to Figma for example and we could connect

Figma. And now I could say agree and

Figma. And now I could say agree and allow access and I could very easily connect Figma to my co-work. And so now it has all of these different tools.

Like it can read, it can generate a diagram on Figma, it can read my files on Figma and we have always allow for all of them, right? And you can allow

certain tools needs approval. Let's just

say always allow and you can literally give it full access to Figma. And so,

more of these tools, right? Like Figma,

like Gmail, I can connect my Gmail. All

right, I just gave it access to my email. If we go to Gmail, you can see

email. If we go to Gmail, you can see that it can get Gmail profile, it can list my drafts, it can basically do anything, right? We can just approve all

anything, right? We can just approve all of it and you can give it full access to your email. You can do this for Canva,

your email. You can do this for Canva, you can do this for Excalidraw. Ooh, I

didn't even actually know you could do this. And so, there you go. In about 2

this. And so, there you go. In about 2 minutes I connected basically everything that I've can connect to that is useful for my business. I can go back to

co-work and I can say, "Please look at my Riley content database, see the

videos I'm working on and my voice and I want you to write a and are we working on a Yeah, I can work in the marketing

folder and I say, "I want you to write an intro for a video that explains

Claude co-work and how it's different than Claude code and we're going to do a demo of each." And so, Claude co-work

has a little bit more than chat, right?

The chat interface is just the chat.

Here it shows the different progress.

Here it shows what folder we're working in. It shows what context, right? It's

in. It shows what context, right? It's

going to track the tools and references and files used in this specific task and then it's going to put the task here on the left side. And so, this was Claude co-work and you can see here it really

quickly went through, found the folder where I make all of my video scripts and now it's trying to understand my voice.

Now it's going to grab a couple more scripts to nail my voice and also see what's currently in progress and then it will do the task. And then here we can

see that Claude co-work responded and we have an intro that sounds exactly like me. Claude just dropped something that

me. Claude just dropped something that changes everything. Most people know

changes everything. Most people know about Claude code. It's the most powerful coding agent in the world. I've

made dozens of videos about it and I use it every single day. But now there's Claude co-work and most people have no idea what it is, how it's different from Claude code or why it even exists. This

is exactly how I would do my intros. I

could say, "Create a PDF of this, please." And here, look, it's using a

please." And here, look, it's using a skill. And so, we're going to get into

skill. And so, we're going to get into skills in just a second. I'll explain

what skills are. Skills are basically a markdown file that explains how to do a certain task. And so now a lot of people

certain task. And so now a lot of people are saying that most software is going to be replaced by skills because anything that you might want to do on some external software can actually just

be done by an agent with a special set of instructions, which are called skills, directly in a tool like co-work.

And so, instead of going to a PDF creator, we're just telling an AI model to create a PDF. And here we can actually see that it's like, "Here's your PDF." We can hit preview and here

your PDF." We can hit preview and here we can open this up. And if we go to finder and we go to that marketing folder that we created, let's go inside the marketing folder, we have the PDF

inside of it. So, we have this folder that we're working out of, which is this marketing folder and it created this PDF. All right, so that is Claude

PDF. All right, so that is Claude co-work. And so, what I want to show you

co-work. And so, what I want to show you real quick is I want to just show you all of the announcements by Claude and all of the new features that they've released in the past few months. So,

I've divided them into these two folders. So, we're going to kind of go

folders. So, we're going to kind of go through these rapid fire. So, obviously

they announced co-work. I showed that to you earlier and so then they released that the Slack MCP is available on all plans. Here they show the example of

plans. Here they show the example of someone who said, "I was at a conference for 3 days." Or they were out out of office. "Please summarize everything

office. "Please summarize everything that happened in Slack while I was gone and draft a message to send to the general chat." And so, here it's

general chat." And so, here it's drafting up a message and you can see here in co-work it drafts up a message and then it gives them this like UI

right here. It's and it allows them to

right here. It's and it allows them to view that message and you can say everything looks good and you can hit send. They also made it much easier to

send. They also made it much easier to connect Gmail to Claude co-work. They

also upgraded the Claude Opus 4.6 model and they made it much better and it can sustain agentic tasks for much longer.

So, these models are thinking for longer and longer whether you're using it for co-work or you're using it for Claude code. And then Claude released a new

code. And then Claude released a new version of Opus and this version is 2.5 times faster, but it's four times more expensive. And so, usually these are

expensive. And so, usually these are really well-funded teams and now they're using a model or a version of Claude Opus 4.6 that's 2.5 times faster, but

it's four times as much. And I can't begin to describe to you how expensive this is. I've talked to developers at

this is. I've talked to developers at companies who spend they spend over a thousand dollars a day on tokens and that's them coding by themselves and their employers think it's more than

worth it because they're getting so much more done. Later they announced that

more done. Later they announced that Claude is on PowerPoint. So, it you can actually directly integrate Claude into PowerPoint and it can generate full-on

PowerPoint slides, which is a really cool use case for enterprises mostly.

They also released Claude co-work on Windows. This was on February 10th. And

Windows. This was on February 10th. And

then they announced that the Claude code feature on the desktop app got dangerously skip permission. So,

remember when I ran it in the terminal with dangerously skip permissions, you can enable it by clicking on ask and then you turn it on this like act mode and you can go enable for this session

and you will see this like yellow bar at the top. It's kind of ugly and it says

the top. It's kind of ugly and it says high risk Claude code can take actions without asking. And remember, it goes in

without asking. And remember, it goes in order, right? Safest to middle safe,

order, right? Safest to middle safe, Claude code is the least safe because Claude code has the most freedom especially when you use it with this dangerously skip permission. It can just go off and edit any file on your

computer and oftentimes I'll have it just fully install programs and I'll say, "Search find an app that we can use for this and just install it to my computer." And it will happily do it.

computer." And it will happily do it.

And so, that's what's fun about Claude code is it has full control and they're making it a lot easier to enable this mode on the desktop app. Then Claude

went through a series of announcements saying that they're going to basically allow for free plan users to get access to more of their fun features. So, I

believe that their free plan is pretty decent. I'm pretty sure you get more on

decent. I'm pretty sure you get more on ChatGPT especially because they're running ads now and Anthropic is just blitzing the the like the productivity

use cases and it's working and they're making a really generous free plan. Then

they announced a little feature in Claude code or this is Claude co-work, you could create instructions. So,

folder instructions. So, remember when we created that folder in co-work, you can actually create a file within that which gives specific instructions for that specific folder and Claude will

make sure to follow it like a little instructions manual. Then they announced

instructions manual. Then they announced that Claude code on the desktop app has access to plugins. But there's a ton of plugins that people have created that

Anthropic has created and plugins you can think of as a set of instructions files that gives Claude access to certain things. So, if I use Claude

certain things. So, if I use Claude code, I can give it access to our internal database for our company as well as our maybe another internal app

or Slack and those set of files or instructions can be a plugin that allow my team members to use as well. So, if I have an agent setup that's really good, I can immediately share that setup with

my team so that their agent can do something similar. And then on February

something similar. And then on February 17th Claude Sonnet 4.6, our most capable Sonnet model yet, was released and this is just a faster and cheaper AI model.

It's different than Opus 4.6. Opus 4.6

is still more expensive and it I think it is better at most tasks, but Sonnet 4.6 is a lot more affordable and it's very good. And it also has a 1 million

very good. And it also has a 1 million token context window. That's in beta and I think we're going to talk about that more in just a second. And then yeah, Figma released their MCP and you can

actually connect Claude much easier to Figma. And so, a lot of designers use

Figma. And so, a lot of designers use Figma, right? They create these design

Figma, right? They create these design boards and then they need to convert them into code. The Figma MCP server and you could literally just go to Claude code and just be like set up uh the Figma MCP or like we did earlier, you

could go to co-work. I like to use it with Claude Code because again, it has access to all your files and if you're coding, you're usually going to be using a directory on your computer and that's where like your code files are located.

So, you could just say, "Please update this app to look like this Figma board."

So, a month ago today exactly, Claude co-work released some more updates specifically regarding co-work plugins.

The main one being the ability to of of admins. So, if you're an admin at a

admins. So, if you're an admin at a company or a manager, you can create plugins, which are basically sets of skills, right? Agents, connectors,

skills, right? Agents, connectors, skills, and commands. You can actually create these plugins and then share them with other members of your team. So, if

you come up with a really good workflow or a really good agent, you can basically duplicate that agent pretty easily and give access to someone else on your team. And then they've also added a ton of other companies and

plugins from other companies. every

single company on Earth is slowly going to throw plugins on Claude because this is just how people are going to find software soon. Like, people are just

software soon. Like, people are just going to want to use software through Claude or other agent platforms. Next up, Claude co-work released scheduled tasks. So, on co-work, you can schedule

tasks. So, on co-work, you can schedule tasks. I can kind of show you this. I

tasks. I can kind of show you this. I

believe that if you're in co-work, I think you can go {slash} schedule. Yes.

Create a task every morning to summarize my email for important tasks and draft

the important ones to sponsorship opportunity.

Yep, and here it goes. I'll set up daily morning email summary for you. First,

let me check what tools and contexts are available to make this work well. So,

you can schedule things really easily in Claude co-work, which is pretty cool.

And then Claude announced Claude for Excel and Claude for PowerPoint and they sync together seamlessly, so you can edit sheets, you can edit PowerPoints, and can you update the model and build

my deck for tomorrow's meeting with my PM? So, it's a lot easier to use

PM? So, it's a lot easier to use multiple extensions at the same time, which is very cool. This one is actually one of my favorites and one that's really easy to just implement in your

daily workflow and it just kind of happens for you if you ever ask for a So, here you can just ask for [music] a diagram.

So, not only are is it generating charts, but these charts are actually clickable. And it's really fast. I

clickable. And it's really fast. I

actually don't really know the technology behind it.

But look, it has just like really good visual reasoning and it's starting to remind me a little bit of Excalidraw and it's just able to like really diagram

really interesting things.

Here, um before I design anything, I want to understand the light. Can you

build a tool that shows the shadows move across the lot throughout the day?

This one's cool if I remember correctly.

[music] Yeah.

I mean, that's just really cool. It's

like a little mini app that you can like create diagrams and stuff with.

And then on top of that, after they released the visual elements in Claude desktop app and their web app, they released that all the models are by

default 1 million context window. So,

this means that you can pack way more into Claude Code. You can give it access to just massive repos. Like, you can give full coding projects to Claude Code. You can give it access to like

Code. You can give it access to like years of your notes and these models context windows are just getting bigger and bigger. And the accuracy with more

and bigger. And the accuracy with more and more context is just getting significantly better across many domains, which I think is really important for a lot of things like law

because law you have to analyze so many different documents and it's it's really intense. And then this is the one that

intense. And then this is the one that where it starts to get really fun. So,

Claude, they are shipping a new feature in Claude co-work as a research preview and it's called dispatch. It's one

persistent conversation with Claude that runs on your computer and you can message it from your phone and come back to finish work. This is a huge trend in

AI right now. It turns out that this whole general agent movement that has been started with open claw and many people are running open claw on a Mac mini and they're messaging it from

Telegram. There seems to be a deep

Telegram. There seems to be a deep desire for being able to just use your phone to message your computer to get it to do interesting things and this is

just basically their version of it. So,

OpenAI decided to basically acquire the guy who created open claw and then Claude Code just said, "Hold my beer.

We're going to replicate all of its features." So, now there's this there's

features." So, now there's this there's this battle between open claw and Claude Code or Anthropic on who can create this like better general agent and it seems

like being able to message a computer that does things is like a really important feature to people and so that's their version is is dispatch. You

can message it. But that wasn't enough.

Being able to message it from your Claude app wasn't enough and then they released channels. So, we just released

released channels. So, we just released Claude Code channels, which allows you to control your Claude Code session through select MCPs starting with Telegram and Discord. This means you can

message your Claude Code session with Telegram. And so, again, this is the

Telegram. And so, again, this is the theme of people wanting to have the power of a computer while being on their phone and messaging that computer. I'm

not quite sure why this is so important to people, but it is. And so, this is a trend that I expect to continue to go on. And then on March 20th, they created

on. And then on March 20th, they created the ability to work in a project. We did

that earlier with the Claude co-work demo. It's where you basically start

demo. It's where you basically start from a specific folder and all of your files get created within that project.

So, you can create many different Claude co-work sessions that all work out of the same project and they'll kind of share context over time. The the context will build up. You'll create a bunch of

files and and artifacts that live within that project. And then yesterday, today

that project. And then yesterday, today we're releasing a feature that allows Claude to control your computer mouse and keyboard and screen, giving it the

ability to use any app. Now, this is insane. This is end game for AI agents.

insane. This is end game for AI agents.

Claude Code allows you to like control commands on your computer. It can edit files directly. It's kind of working in

files directly. It's kind of working in the background. The next iteration of

the background. The next iteration of this or like the next level of automation for these computer use agents is the ability to just directly use your computer, move your mouse around, type,

and you're seeing this a lot like with Comet from Perplexity. ChatGPT has an agent mode where it can spin up these little browsers and control these little browsers. And now Claude, who's been

browsers. And now Claude, who's been working on kind of the computer use models for a while since like Claude 3.5, you could use these computer use agents. It turns out they're getting

agents. It turns out they're getting better. They're still not perfect, but I

better. They're still not perfect, but I think by the end of the year, AI might be better and faster at using the computer than the average human, which is scary. And if an agent can use a

is scary. And if an agent can use a computer, that means it can basically do anything and it will literally be a co-worker. And this is why a lot of

co-worker. And this is why a lot of companies who are like going through layoffs are having the people record their workflows on their screen. Like,

they're saying like, "Hey, can you show me this workflow process and do a screen recording?" because they're going to use

recording?" because they're going to use that information to kind of train these computer use agents in doing their job and then they're going to let them go.

And it's becoming a lot It's becoming quite controversial, but it's happening.

Okay, so we talked about all of the latest announcements on Claude co-work, right? We have just kind of the if we

right? We have just kind of the if we think about it for a second, the overall trend is like more connections, more ways to talk with the Claude desktop app, more channels that you could talk

to it. Like, you can use different apps,

to it. Like, you can use different apps, much easier to connect with all of your existing tools. The existing tools that

existing tools. The existing tools that you use can kind of work in parallel, right? You can create a PowerPoint and a

right? You can create a PowerPoint and a spreadsheet. And like as you update the

spreadsheet. And like as you update the spreadsheet, you can have it update the PowerPoint and everything is just becoming more interconnected, right? It

is becoming a little bit more like a co-worker every single day. So, that's

kind of the overall trend for like the desktop app. Here, we we still need to

desktop app. Here, we we still need to go through the announcements for Claude Code. Some of these might be a little

Code. Some of these might be a little bit more technical than the desktop app ones, but I I think there's something here that you could learn. So, let's go through these really quick. So, here we

see Claude Code now supports session sharing. This is really interesting

sharing. This is really interesting because one thing that we actually do when we're hiring people, so for hiring a developer, we will give them a task or like we'll do a work trial. And if at

the end of the work trial, we'll see like what they ended up with, but we'll also see their session history or the conversation history with Claude Code as they build whatever it is that they're

building. And you can actually learn a

building. And you can actually learn a lot about someone by seeing the way that they interact with like a super smart AI that can build stuff. You really learn a lot about their problem-solving. I think

this is going to be a very big thing with Claude co-work, right? As we're

starting to use these really powerful general agents, I think employers are going to start asking for your history to see how you talk to the agents and I think it says a

lot about your problem-solving. And so,

yeah, you can very easily just share You can actually just make them public. You

can make your conversation history with Claude Code public, and then you can share them with people. You can share the link, and this is really, really cool. A similar feature that they

cool. A similar feature that they released is {slash} insights. And so, if you type in {slash} insights, it will actually generate kind of like this really long summary of all the projects

you've worked on. It'll show you exactly how many lines that you've submitted.

It'll show you your habits and what tools that you commonly use. And so,

this is Claude Insights. So, I think this one's pretty cool. This one was actually insane. So, Claude Code now

actually insane. So, Claude Code now supports agent teams, and so it's in reach research preview. So, instead of a single agent working through a task sequentially, basically a lead agent

will delegate to multiple teammates that will work in parallel to research, debug, and build while coordinating with each other. Again, this is another trend

each other. Again, this is another trend is like multi-agents. Right now, Claude Co-work is very like singular threaded, and I think the next iteration of these

general agents, after the the kind of the single tasks get better, will be a team of agents for general tasks. We're

already seeing many really good developers are creating these like multi-agent frameworks, and they're operating five to 10 AI agents at the same time to build their applications.

And so, I think this is going to become a massive trend. The smartest people at companies who are really good at AI are just going to be using many agents at once. Claude Code also released rewind.

once. Claude Code also released rewind.

So, you can actually just type {slash} rewind, and you can rewind a conversation by hitting {slash} rewind or hitting escape twice. Claude can

summarize the part of the conversation that was rewound. So, you can use this to explore different paths and take the learnings back in time, which is very, very interesting. Okay, this one went

very interesting. Okay, this one went insanely viral. So, there are many

insanely viral. So, there are many startups here in San Francisco that do security. So, like as you commit your

security. So, like as you commit your code to GitHub, there's like an automatic security review, and like Code Rabbit, and there's a bunch of other ones as well. Well, Claude Code built it

directly into their app, and so now it has security built in. So, when you're using Claude Code, you can go up there and press security scan, start a scan,

and it will actually scan your code base for vulnerabilities. And this is

for vulnerabilities. And this is actually good for vibe coding. If you

create an app and you want to make sure that like there's no security issues, you can just run one of these security scans. This might not be perfect now. I

scans. This might not be perfect now. I

think a year from now this will just be perfect. Like this will be more

perfect. Like this will be more effective than having like a team of humans analyze for vulnerabilities. Like

I just If you just look at the general trend, there's no way that this won't just be perfect, especially for like not super complex code bases, but for medium-size code base or any app that I

might want to create, you know, for my followers. Like I genuinely believe

followers. Like I genuinely believe you're just be able to press one button.

It'll scan it for security, and that'll be that. And then uh this was in late

be that. And then uh this was in late February, but they announced that Claude Code in the terminal, you could connect it via a remote control. So, here's

Claude Code If you type in {slash} remote control, it gives you this link, and now you can actually control this session of Claude Code in your terminal with your phone, which is pretty cool.

Okay, so this was a really important one. So, they rolled out auto memory.

one. So, they rolled out auto memory.

And so, Claude now remembers what it learns across sessions, so your project context, debugging platform patterns, preferred approaches, etc. So, one of the big reasons that Open Claude became

so popular is it has like good memory built into it. So, like you can message it, you can start new sessions, and and it'll actually remember something from 7 days ago, and as you use it, it gets

better. Or at least in theory, as you

better. Or at least in theory, as you use it, it gets better. Well, Claude's

kind of rolling out some similar features with an auto memory feature, so that as you use Claude Code, so Claude Code now remembers what it learns across many sessions. And so, now these coding

many sessions. And so, now these coding agents that are becoming general agents are getting better memory, which is something that I think many people want, especially for general tasks. This one's

pretty fun. You can actually use voice mode directly in the terminal version of Claude Code. So, if you see here, you

Claude Code. So, if you see here, you can go {slash} voice, and now you can hold space to speak, and then you can speak and say refactor the authentication middleware, and it will

send that in. So, you can use your voice here directly from the terminal. You can

create scheduled tasks. And so, you can schedule a daily code review. So, every

day on a schedule, you can get it to review your code. And this is a fun final one. This is the last feature

final one. This is the last feature announcement that we'll talk about. You

can in Claude Code, you can type {slash} btw. And what this will allow you to do

btw. And what this will allow you to do is let's say your agent is working for a really long time. Often times, tasks can take up to 20 minutes if it's doing something really long. Well, now if you want to chat with the agent while it's

working, you can just type in {slash} btw, and you can just ask it a little question. Like what does this retry

question. Like what does this retry logic do? And then it'll just respond to

logic do? And then it'll just respond to you kind of on the side. Like it's like, "By the way." And then you can ask it another question, and it's kind of off to the side. The fun way to interact with Claude while it's working for a

really long time, and it doesn't affect the main conversation. And so, that's it. In summary, this is what we talked

it. In summary, this is what we talked about. By the way, I took the the

about. By the way, I took the the subtitles or the SRT file from the video that I'm recording now, which is like the subtitle file. I didn't even know Claude Code or Claude desktop app could take it, and I said, "Please create a

diagram based on what we've talked about so far." Like this is just the straight

so far." Like this is just the straight subtitles file. It tried to use Figma,

subtitles file. It tried to use Figma, and I was like, "No, no, no, no. Use the

built-in one." Right? Because of these visual diagram updates, I wanted to test this one. Said, "Use your built-in

this one. Said, "Use your built-in diagram view." And it created this.

diagram view." And it created this.

Right? You can Oh, you can click on it.

Wow, that's pretty crazy. You can click on it, and it will immediately do a follow-up question. That's pretty cool.

follow-up question. That's pretty cool.

I didn't know that. But yeah, this is kind of an overview of what we've talked about. A lot of Co-work updates, a lot

about. A lot of Co-work updates, a lot of Claude Code updates, some integration updates. All of it is kind of becoming

updates. All of it is kind of becoming one general agent tool, and Claude didn't do what OpenAI did, right? OpenAI

has CodeX, right? This is their coding tool that they've created. They also

created ChatGPT. They also created Sora.

They also created Atlas, which is their browser. Claude or Anthropic decided not

browser. Claude or Anthropic decided not to do that. They created one super tool, and many people are switching to Claude.

It is the most popular piece of software in the world right now. We thought Open Claude took over the world, fastest growing open-source project ever created, but Claude decided not to

acquire them. They decided to stay

acquire them. They decided to stay focused, and it's working right now. I

hope this video was useful. We talked

about a lot. Claude is very important tool to learn, but there's going to be many more tools in the future. I'm going

to be making a lot more videos on Claude. I still like to use Open Claude

Claude. I still like to use Open Claude because it's more autonomous. One thing

that I didn't really talk about is the fact that Open Claude runs on one specific computer, and you generally don't turn it off. You can also run it in a virtual computer all the time, and

that's where I create my agent team. I

want my AI agents to be more proactive and feel like a person and have a different personality. So, I'm going to

different personality. So, I'm going to continue to talk about Open Claude in the future, and I'm going to talk about my Open Claude team a lot coming up. So,

I'm really excited for that. So, make

sure to follow along if that interests you, and I'll see you here for the next video. Peace.

video. Peace.