Big Idea

The argument in one line.

The Claude desktop app has become a workspace for running many sessions at once, with one master session orchestrating the rest, so using it well now means managing sessions, context, and artifacts instead of single chats.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You use the Claude desktop app daily for business work and feel buried under dozens of chats in the sidebar.
You're a non-developer who avoids the Code tab because the name sounds technical, and want to know why that's a mistake.
You hit your Claude usage limits early and want to see what is eating your context and prompt cache.

SKIP IF…

You only use Claude through the API or a terminal and never open the desktop app.
You're on a free plan, since several features here need Pro or Max.

TL;DR

The full version, fast.

Anthropic folded code work into chat in the Claude desktop app and added a set of features for running many sessions at once. A session rail shows context use, nested sessions let one chat spawn child chats you can talk to, and a pinned master session can watch and steer the rest. Cloud-based Projects keep working when your computer sleeps. Side chats and split view let you ask questions or watch several chats without interrupting work. Artifacts now cover docs, slides, design, dashboards, and motion video, and can embed Claude itself. New slash commands and mods help you trim token use.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 01:02

01 · Intro: Claude 2.0

Overview of the 11 updates and who the presenter is.

01:02 – 03:30

02 · Session rail and context card

Environment, remote control, and context cards, and how memory files, skills, and MCP use your limits.

03:30 – 04:10

03 · Sections

Group chats into folder-like sections in the sidebar.

04:10 – 05:40

04 · Nested sessions

Sessions that start child sessions, versus one-way subagents.

05:40 – 07:20

05 · Master session

Pin one session to watch and steer all your other chats.

07:20 – 09:35

06 · Claude Code Projects

Cloud-run projects with shared memory, and when to pick them over nested sessions.

09:35 – 10:50

07 · Side chats

Saved side conversations that can read the main session.

10:50 – 12:15

08 · Split view

Show several chats on one screen at once.

12:15 – 14:40

09 · Artifacts

Docs, slides, design, dashboards, and Claude Motion, plus plan availability.

14:40 – 15:35

10 · AI-powered artifacts

Embed Claude inside shared artifacts so others can query them.

15:35 – 17:00

11 · Slash commands, mods, and shortcuts

Output style, doctor, skill doctor, advisor, insights, recap, rewind, mods like Ribbon, and key shortcuts.

Atomic Insights

Lines worth screenshotting.

Memory files, skills, and MCP connectors are pre-loaded into every new chat, so bloated ones eat your usage limits before you type a word.
Nested sessions are real chats you can click into and steer, while subagents are one-way orders that only return a result.
Pin one session, tell Claude it's your master session, and it can list which other chats are waiting on a decision from you.
A master session sees your 20 most recently active sessions by default, but you can ask it to look further back.
Claude Code Projects run on Anthropic's cloud and keep working while your device is off, but any files you upload go to Anthropic's servers.
Nested sessions run on your computer with all your local skills and files, so they stop when the machine sleeps.
Side chats can read the whole main session, so you can ask about progress without interrupting the work or spending main-chat tokens.
The old /btw popup was temporary, but side chats are now saved and you can open as many as you want.
Prompt cache goes cold after an hour, and the next message then rereads the whole conversation at normal cost.
Slash advisor lets a cheap model like Sonnet run most of the time and call a stronger model only on critical decisions.
Claude Motion makes animated video with code instead of a video model, so there is no per-generation fee and the output is easy to edit.
Turning on AI-powered artifacts lets a shared dashboard answer questions from anyone with the link, no Claude Code knowledge needed.
Pressing Escape twice triggers /rewind, which works as an undo button for a conversation.

Takeaway

Manage sessions, context, and artifacts deliberately

WHAT TO LEARN

The new desktop app rewards treating Claude as many coordinated sessions, so watch your context, pick the right session type, and use cheap side tools to save tokens.

Open the session rail's context card to see how much memory files, skills, and connectors cost, because they load into every chat and spend your limits first.
Use nested sessions for local work that needs your files, and Claude Code Projects for long jobs that should keep running with your computer off.
Pin a master session and ask it which chats need your decision, so you stop hopping between ten or twelve conversations.
Ask quick questions in a side chat instead of the main thread, since it reads the whole session without interrupting work or burning main-chat tokens.
Start split view with two chats side by side and add more only when it feels easy, since ten panes overload your own attention.
Use artifacts for docs, slides, design, and dashboards, and turn on AI-powered artifacts so teammates can ask a shared dashboard questions.
Run slash doctor and slash skill doctor to find skills and settings that waste working memory, then switch off unused ones.
Use slash advisor so a cheaper model handles routine work and a stronger one steps in only for critical decisions.
Add a mod that shows weekly usage and whether your prompt cache is warm, so you avoid messages that reread a cold conversation.
Learn Ctrl+Alt+Space, double Escape, Ctrl+D, and Ctrl+/ to open Claude anywhere, rewind, dictate, and list every shortcut.

Glossary

Terms worth knowing.

Session rail: A panel on the right of the Claude Code window with cards for the environment, context usage, connectors, and outputs of the current chat.
Nested session: A chat started by another chat that sits under it in the sidebar, reports back when done, and can be opened and steered directly.
Master session: A pinned session you designate to look across your other chats, surface pending decisions, start new sessions, and organize the sidebar.
Prompt cache: Short-term memory Claude keeps for a chat so it doesn't reread the whole conversation each message. It goes cold after about an hour.
Artifact: A shared canvas, such as a doc, slide deck, design, or dashboard, that you and Claude edit together and can publish as a link.
Mod: A small add-on, made by describing it to Claude, that changes how the Claude Code app looks or works.

Resources

Things they pointed at.

08:00toolRibbon mod ↗

00:40linkClaude 2.0 Guide PDF ↗

08:50productRoboNuggets community ↗

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

Cloud has just been completely revamped by Antropic. And they've added several new features, shared some official tips, and even some hidden functionalities that make the desktop app a completely different experience. And of that, one of their lead engineers called it Cloud 2 .0.

You can now have a master session that runs the rest of your Cloud chats for you. add and build mods that literally change how Cloud works. And they've even shipped practically full replacements for software you might be using, like Google Docs or even some of your video editing tools, like the animations in this video, for example, which were made with the new Cloud.

So today I'll show you the 11 most important updates that they shipped. so that you too can get more done with Cloud 2 .0 and fully take advantage of everything that it can do now. And if you're new, my name's Jay.

I spent over a decade working with brands you probably know, have been in AI since my master's in data science, and now I'm leading our AI business in one of the largest AI communities globally. So when you open the new Cloud desktop app, you'll now see that there's only chat and code. And that's because as of last month, code work has been folded into chat.

Now, if you want to make full use of Cloud, you have to use Cloud Code because Cloud Code is clearly where Entropic is shipping the best and the most updates because they know that it's their star product. And if you're new, don't be intimidated by the word code, by the way, because that's really just a misnomer. Anyone can use cloud code because you just prompt and direct cloud in plain English anyway.

So you don't have to become a developer to use cloud code. So just go and switch to that tab because that's also where we'll start with the latest updates, which is around the interface. Now, the big one that they shipped in the interface is what they call the session rail.

It's this panel on the right side of your window. And if you don't see it, just click show session rail at the top right of the UI and it pops open for you. Now, the session rail, in essence, is just a set of cards that tell you what is going on in this particular session.

The environment card shows which folder this chat is working in and whether it's running on your computer or in the cloud. Inside it, a very useful switch is this remote control button. Toggle it on and you can keep going with this.

exact session from your phone. So you can start something at your desk and check on it while you're away. Now, here's a pro tip because in settings under Cloud Code, you can turn on connect new sessions to remote control.

So what that does is that every chat is now going to be ready for your phone and you don't have to manually toggle this every time. You can also turn on keep computer awake while Cloud works so that your computer doesn't go to sleep halfway through a task. Then at the bottom of that, there's the context card.

Now think of context as Cloud's working memory for this chat. The card just shows you how full it is. what's taking up the space because with the newest models of cloud they have a 1 million token context window which to put that into perspective that's roughly the length of four harry potter books put side by side and what's great with this view is you can see apart from your messages which are your prompts you can view your memory files your skills and your mcp connectors and how much they are already taking up space in your chat now that's important because those actually use up your usage limits on your very first prompt because they get pre -loaded on every chat that you start so with view you can see if they're becoming too long and that you might need to clean them up and you can get started with that cleanup with this prompt on the screen you'll also see your connectors here which are the apps that you've plugged into cloud and an outputs card for every file that this chat has made and by the way i've put every prompt resource and shortcut that you will see in this video into this free pdf guide

And you can just get that, send it to your cloud, and ask it to either set things up for you or teach you about the new features that you may not be using yet. So you can just grab that in the description below if you need it. Now, to keep all of your chats and sessions tidy, you can now sort them into sections here in the left sidebar, kind of like folders.

If you click the three dots, the ellipses next to any chat, then choose move to new section, then you can create that section and give it a name. Then after that, you can just drag any session into that folder. So instead of scrolling through 50 chats to find the one that you need.

You can have one section per project, per client, or per part of your business. And something related to that is this awesome feature called nested sessions. Now a session is just one of the chats in your sidebar, right?

But now a session can start its own sessions. And what that will do is have those sub sessions sit right underneath it. And when each one of these sub sessions finish, they report back to the main session that started them.

And to try out this new feature, you can just ask Cloud to do a few tasks at the same time and also ask it to start nested sessions for each one. So let's say you're launching a new marketing campaign. Then you can just have one main session start a sub session for your sales page, another sub session for your emails, and another one to create your ads.

Now, it's important to differentiate this versus subagents. And if you're new, subagents are basically helpers that a session can send off to do one job. And they're still useful, but basically how a subagent works is they're more like a one -way order.

The main session can send it a prompt and you get a result back. Sure, you can peek at its progress, but you can't really jump in and talk to that subagent the same way that you can with nested sessions. Because with nested sessions, each one of these is a real chat in your sidebar that you can click into it, see everything.

it did and keep working with it directly so let's say you want to steer the session that is creating your ads more directly then you can just jump into that chat and do that more directly and on top of this here's a tip that most people don't know yet because in cloud code one session can actually see your other sessions and so what this enables is the concept of a master session so that if you want to stop jumping between 10 or 12 chats in your sidebar you can just pick one session You can even pin it to the top so it's always visible and just tell Claude that this is going to be your master session.

Now, once this is set, there are now a few ways that you can use this. For example, you can ask your master session for the critical decisions that your other chats are waiting on. What it will now do is to go and look through all of those chats, pick up the ones that are actually waiting on you for a decision.

summarize them for you from within just that master session so in a way this pin chat can now be sort of your personal assistant that orchestrates cloud and all the other chats that you have previously started you can even let it start new sessions for you and control them from just this one master session and if you want to tidy up your workspace in your cloud app you can also ask it to group your chats into sections For example, I just asked mine to make a new section for all of the chats related to this video, and it just did it without me clicking anything.

Just to note, this feature, by default, it can see your 20 most recently active sessions, but you can, of course, just ask Claude to look further back if you wish. Now, a closely related update is Cloud Code Projects. It's the same idea where you have one main session that is running a whole team of chats for you, but there are a few big differences.

The biggest one is that projects run in the cloud, so they keep working even when you close your device. Nested sessions, on the other hand, run on your computer. So obviously, if it's not awake, then those sessions would not run.

However, nested sessions do have access to all of your skills, files, and tools in your device. which is really useful if you're doing tasks that need your business context. Cloud Code projects can of course use your skills and files too, but you need to add them to the project first.

Now, you can, of course, upload those files manually, but the best way to do this is to just ask Cloud to get all of those skills and context that is necessary for that project so that it can get them from your computer while it's still awake. Do note that natively, since Cloud Code projects run on Anthropix Cloud, that also means that any business file that you upload is going to go into Anthropix servers.

The other difference is that projects also have shared. custom memory so you can set the rule once for the entire project and every chat session that you start with in this project would adopt that rule. In Projects, you also get a nice overview page that shows what's done, what's still working, and what's waiting on for you.

So a good rule of thumb is if the work needs your computer or if you prefer to work locally, then just use Nested Sessions. If it's a big ongoing project or job that you want running even while you're away, then go and explore Cloud Code Projects. Now, an important bit is that right now, Projects is still being rolled out.

So while Nested Sessions work on any plan, Projects is still on beta. And right now, they're only on Pro and Max. And if you don't see projects in your sidebar, just check under more.

which is where you will usually find them. And if it's not there either, you may need to join the waitlist with the link that you see on the screen, which is also in the PDF guide. And by the way, if you want to learn how to build and sell AI systems that businesses actually pay for, then that's pretty much all we do over at the RoboNuggets community, where not only do you get access to the Cloud Living Masterclass, which we update every week and takes you from zero to mastery with the latest on AI, but you also get access to our Agents as a Service course, which walks you through how to actually get paid for all these AI skills that you are learning.

You also get to be part of a genuinely great community of AI builders. In fact, you can see just some of the recent wins our members are getting from the program right here.

So if you want to start earning from AI, then check that just in the pinned comment below. Now back to the video. Now let's say Claude is in the middle of a long task and a question pops up into your head.

Like you want to know, for example, how long still until this build finishes, or you just want to remember what this build is even about. If you ask it in the main chat, that is not advisable for a lot of reasons because you interrupt the work. you also spend more tokens now that is what side chats are for to open one you just click the ellipses at the top right or even type slash btw once you do that a new chat window would pop up next to your main one and this side chat the main benefit of it is that it can read your whole main session so you can ask it to explain what cloud has done so far or to check the works status Now, this is a great upgrade because before, slash BTW or slash by the way, would just pop up a temporary chat that didn't really get saved anywhere.

But now you can start as many side chats as you want, and each one is kept so you can go back to it whenever you need to. Now, if you're always clicking back and forth between chats on the left sidebar, Split View puts several of them on one screen at the same time. To do it, just hold Alt or Option on Mac and click a chat on your sidebar.

You can also right click the chat, choose open in split view, and it will show several panes in your cloud app so that you can manage multiple sessions at once. And if you want to really squeeze in as many as you can, you can actually zoom out by pressing control and minus or command and minus on a Mac. And when I did this, I could fit up to 10 different chats on my screen at once.

But of course, be careful because having that many things on your screen fills up your own brain's context window. So this one's definitely not for beginners. You can probably start with two side by side.

And just add more once that feels easy. Now let's talk about what you can make inside Cloud. Because Entropic now has its own take on Google Docs and Google Slides, plus tools for design, dashboards, and videos.

Each one of these makes what they call an artifact. Think of it like a canvas where you and your Cloud agent are working on the same design or dashboard. And you can even leave comments for Cloud to fix it.

And just like Google Docs, you can publish any of these as a live link, or you can share them with your team or a client. Docs, of course, work like Google Docs, where you and Claude write the same document together. Slides makes presentations, and you can also download them as PowerPoint or even present straight from Claude.

Design is sort of Claude's answer to Canva, where you can create social posts, websites, and anything visual, really. And very clearly, Entropic is moving Claude Design in here for good, because the old standalone Claude Design site, the one at claude .ai slash design, is actually closing by the 14th of December. In fact, if you go to settings, you can even upload your own design systems right here now in Claude Code.

Or you can do what most people do and just have a skill that brings in your design system. Then a new artifact that they just added is dashboards. So dashboards can pull live data from your connected apps.

So every time that you open the link to this dashboard artifact, the numbers are going to be up to date. And finally, Cloud Motion makes animated videos with code instead of a video model, which makes it much cheaper because you're not paying per generation. And it also makes these videos more malleable.

I actually did a full tutorial on Cloud Motion, so I'll just link that as part of this video if you want to learn more. Now, not every plan gets every one of these, by the way. So here's a quick table of which plans get what.

As a quick headline, if you're on Pro or Max, right now, the only thing you don't have access to is Cloud Motion. But for sure, it'll be added real soon because of how popular it is. Now to take artifacts to the next level, you can actually now use Claude to create what Entropic is calling AI -powered artifacts.

To turn this on, just go to Settings, then under Capabilities, you can switch on AI -powered artifacts right in this toggle. With this on, the artifacts that Claude is building for you can have Claude... inside them.

So if you want LLM capabilities from within the artifacts that you make, then this is definitely a feature that you want to try. And because you can share artifacts with your teammates, this is where it gets really useful because let's say you have a dashboard that you and your team are using all the time. You can actually ask Claude to add a chat window or a search bar right inside it where anyone can just type a question and Claude answers it using the data on the dashboard.

So in this way, you just made this artifact AI powered or LLM powered, which means that your teammates, especially if you're sharing it with a wider audience who may not necessarily know how to use Cloud Code, they can just go to that dashboard and ask it questions to get the data that they want without having to learn Cloud Code.

They just go to that link that you shared to them and that dashboard app itself will be LLM enabled. Now, Cloud Code also has a bunch of new slash commands that are worth knowing. A slash command is basically a shortcut that you type, and Entropic actually published new ones and even updated some of the more useful ones from the past.

And here are some of them. Slash output style changes how basically Claude talks to you. So if its responses feel too long or too technical, you can actually switch it to a shorter, plainer style, like this Eli5 style that I've shared in another video before, which just makes Claude speak in a non -technical way.

And of course, you can always ask Claude to make your own unique output style as well. Slash Doctor is also really useful. It runs a checkup on your setup, looking at your settings, your Cloud .MD, and offers to fix any problems it finds.

And relatedly, Slash Skill Doctor shows what each of your skills costs in Cloud's working memory and how often you actually use them. So you can switch off the ones that you never use so that you can save on some tokens. Slash Advisor is also a useful but underutilized command because it lets Claude call in a stronger model like Opus or even Fable, but only when a big decision comes up.

So you can run a cheaper model like Sonnet, for example, most of the time, and it will automatically just use a higher model if it deems that decision to be critical. So that way you won't have to burn through your limits on the big models. And lastly, three more quick ones.

Slash Insights gives you a report on how you use Claude code and features to try. Slash recap gives you a one line summary of where a chat is up to so that you can remember what it is that you're talking about with Claude. And slash rewind takes you back to an earlier point in the chat, which is sort of like Claude's undo button.

And the next one is mods, which is one of my favorite new features in the Claude desktop app. A mod is basically a small add -on that changes how the Cloud Code app looks or works. And you can make one just by describing it to Cloud.

The first one that I use every day is this mod I made called Ribbon. It's a slim bar above my chat box that shows what's left of my usage for the week. And it also shows me if my prompt cache for that chat is warm or cold, which really helps cut down my token burn.

Now, if you don't know what the prompt cache is, it's basically shorter memory that Cloud keeps during your chat so that it doesn't have to reread your entire conversation every time that you... send it a message because after an hour that prompt cache goes cold and then the next message makes cloud reread your entire conversation at the normal cost which can really eat into your limits so seeing that at a glance through this mod actually saves you tokens in the long term this one's free and open source by the way so i'll leave the link down below for that if you need it and the second one that i really like is this one called savvy progress And apart from putting a progress bar above your chat box so you can see how close Claude is to finishing long tasks, for bigger jobs where a team of helpers or sub -agents is also doing the work, it actually shows you a live panel of every one of them here at the right, complete with little pixel Claude crabs that you can even customize.

Now, I did a full video on Claude mods, so check that out if you want to go deeper in this new feature. And finally, a feature that is underrated in my view is learning a few keyboard shortcuts that actually save me a lot of time. So for example, if you press control alt and space, you can open Claude from anywhere on your computer.

So if you're halfway to an email and want Claude's help, you just press those three keys and ask your question. And that opens a new chat in your app. If you press escape twice, that actually triggers slash rewind so that you can go back to an earlier prompt if you want to undo any conversation after that certain prompt.

And if you press control and D, you can talk instead of type because that is the shortcut to enable voice transcription in Claude. And to see every shortcut there is, there's actually quite a lot. You can just press control and slash or command and slash on a Mac, and that will show you the full list.

Keyboard shortcuts may be a small thing, but if you use Cloud a lot, then I think learning them is definitely worth your while. So there you go. Those are the 11 biggest updates to the new Cloud desktop app.

And as usual, thanks for watching till the end. I do more of these sessions where the whole intention is to let you use cloud better for your business. So if that's something that you're into, then feel free to subscribe down below.

That's it for this one. I'll see you all next time. Cheers.

The Hook

The bait, then the rug-pull.

A lead engineer reportedly called it Claude 2.0. Jay walks through the 11 updates that change how the desktop app works, from a master session that runs your other chats to mods and full replacements for docs and design tools.

CTA Breakdown