Getting the transcript
Reading the captions from YouTube. A video nobody has opened here before takes 10 to 30 seconds; this page fills in on its own.
Getting the transcript
Reading the captions from YouTube. A video nobody has opened here before takes 10 to 30 seconds; this page fills in on its own.

9x · @go9x
Where viewers went back to watch this video again, from YouTube's public Most replayed graph, lined up with what was said at that moment.
Most replayed moment #1
8:423.1x the video's typical replay level
saving us a bunch of time and allowing us to move away faster with our communication. Now, one bonus feature of Claude Co-work that a lot of people might not know about is its ability to control your browser. So, thanks to its Claude in Chrome plugin, you can actually have Co-work open up a Chrome
Said at 8:34
Most replayed moment #2
12:043.1x the video's typical replay level
copy also analyzed. And so we can see here of ads driving to the main sign-up. And then they also have a 1,000 Prompts lead magnet with a different set of ads pointing to that landing page. This whole thing would have taken me hours to
Said at 11:58
Most replayed moment #3
1:413.0x the video's typical replay level
Aaron and I during this live workshop. And as well in that folder, we have the whole text transcript from the recording. So, with that folder now selected here in Co-work, I can run a skill that I have prepared, which is called 9 x live get video highlights.
Said at 1:35
The graph counts replays. It does not show where viewers stopped watching.
Words
5,559
Runtime
28:03
Speaking pace
198wpm
Reading time
23min
198 words per minute, between the 181 median and the 201 75th percentile of 349 measured videos. That distribution comes from the 349-video hook study.
Opening (first 30 seconds)
Claude Co-work might be the most underrated AI tool right now. Most people are still using Claude like a chatbot, typing questions, getting answers, but Co-work is a completely different thing. It actually sits on your computer, reads your files, works across your apps, and does real tasks for you while you focus on something else. At 9 x we've trained over 10,000 business professionals on how to actually use this stuff, and I've personally tested every major AI agent out there, and Co-work is the one that keeps surprising me with what it can actually do. So, in
99 words, the words spoken in the first 30 seconds at 198 words per minute.
Free, no signup. See how the first 30 seconds hold attention, with rewrites.
What this transcript is
Every word below is the caption track YouTube publishes for this video, pulled from the video itself and reproduced unchanged. It is not Prepublish's writing, not a summary, and not a re-transcription: it is the video's own published captions. English captions, generated automatically by YouTube, in the video’s original language. Source: the video on YouTube. A channel that would rather this page did not exist can ask for its removal through the contact page, and it is removed.
Claude Co-work might be the most underrated AI tool right now. Most people are still using Claude like a chatbot, typing questions, getting answers, but Co-work is a completely different thing. It actually sits on your computer, reads your files, works across your apps, and does real tasks for you while you focus on something else. At 9 x we've trained over 10,000 business professionals on how to actually use this stuff, and I've personally tested every major AI agent out there, and Co-work is the one that keeps surprising me with what it can actually do.
So, in this video I'm going to show you eight real use cases, things you can set up today that'll save you hours every single week. No coding, no complex setup, just practical ways to put Co-work to work so you can stop doing the stuff you shouldn't be doing manually. So, let's get started with use case number one, and that's editing videos. And I was absolutely blown away when I discovered that Co-work could do this for me.
Um so, let's check it out. Here I am in the Claude desktop app. I've got the Co-work tab selected, and important to know that whenever you're starting a new task in Co-work, you should always be working in a dedicated folder on your computer. So, I'm going to be selecting here work in a project, and then I can select a folder. I'll scroll down and select choose a different folder, and the folder that I'm going to be working with here is this one, 9 x live agent skills for beginners, which contains the recording from one of our recent uh live AI workshops that we run each and every week for our community.
One which contains the screen share view from the webinar, and the other which is the speaker only view. So, there you can see Aaron and I during this live workshop. And as well in that folder, we have the whole text transcript from the recording. So, with that folder now selected here in Co-work, I can run a skill that I have prepared, which is called 9 x live get video highlights. So, that's all I need to type. I'm going to click let's go.
Now, Claude has got to work and given itself a to-do list. So, it's going to check what files are in that folder. It's going to read the full transcript, identify the top moments from the recording, and then actually cut five clips that I can share directly on social media, which includes adding our 9 x branding. And after about 5 minutes of working, Claude is all done. Here are your five LinkedIn ready clips from the agent skills for beginners workshop.
We also have actually a markdown file that Claude has prepared for us giving an overview of each clip, uh the summary of what's included, as well as the full transcript uh from the workshop. But even more impressive, we here we have the five video files. Um let me open up, for instance, clip four here, skills are just text. In one click, I can open this up in QuickTime Player. So, here we have one of the recommended highlights from Claude from uh Aaron from Gum Loops' live webinar on agent skills.
Let's take a look. Yeah, let me show you what a skill looks like. This is one from an AI >> There's Aaron up here in the top right corner. >> I know, and I'm going to say this throughout, everyone on the internet is like, "Oh my god, I built this skill. I'm so amazing." They're just text. It's just text, right? That's all you have to remember for today. They're text that the agent reads. >> So, this was the the highlight of Aaron making the point about agent skills just being text.
So, what you might have noticed there is the skill in Claude Co-work, I've basically instructed it, depending on what is being shown on the screen and the transcript at the time, it should cut between showing the screen share view. So, this is the screen share view where we have the speaker top right corner. This is what we get from our Zoom recording, but we also get this speaker only view. And when it changes at the start here, Aaron's talking about skills are just text, so it's important that we see the text on the screen.
And then as Aaron continues making a point, we see it just cuts to the speaker only view, which is going to perform much stronger on social media. Here you can see an additional point that I made. And even more impressive, what we also have in the skill, let me show you the end of this video. and your style guidelines of your organization. And here we have the 9 x branding outro appended on to the end of all of our social media clips.
Now, if we take a look at that working folder that we started with, instead of just having just the recording uh videos that we started with, we now see Claude Co-work has added all of those five social media clips, which are different highlights from that webinar. Here you can see me explaining an analogy about how skills work, all of which have the 9 x outro at the end. Now, for us, this is easily saving us 1 to 2 hours every week of doing this video editing manually, which we used to do in our video editing tool, uploading the recordings, finding the exact moments, and then cutting everything together, and including our branding.
Now, Claude takes care of the whole thing. Now, obviously not everybody is running live webinars every week, but I can think of a ton of additional use cases for this. Maybe you have a YouTube channel and you want to cut up highlights from your YouTube videos, or if you're doing user interviews or collecting testimonials via video calls, you could give Claude the whole hour-long interview with a user and get it to extract the key highlights, and maybe post those as testimonials on your website.
Now, if you remember, all I needed to kick off this workflow here in Co-work was by calling a predefined skill that I have set up, which is get video highlights. And all of the eight use cases that I'm sharing today have a predefined skill set up. And if you check the link in the description below, you'll be able to download all of them and test them out for yourself. All right, onto our next use case, and that's creating marketing emails directly in your CRM tool.
So, here I am in Claude Co-work, and I already have a prompt written out here about an email that I want to create and add into our email marketing tool. So, it says, "Can you please draft an email that we can send to our students who have just completed our AI Builder training program? Congratulate them on completing the course and encourage them to download their certificate and add it to their LinkedIn as a certification." Now, what we'd expect if we were to send this prompt off to regular Claude chat, is we would get back a very well-written email, but just in text form, and we'd actually need to go and implement that email.
Now, watch what happens when I add this dedicated skill that I've created called create marketing email. I'm going to send this one off to Claude Co-work. And we can see that Claude has picked up that skill, and it's getting to work. And we can see it's started with its to-do list, and more importantly, it's connected to our email marketing tool Customer.io. And now it's going to create that email directly in our tool.
And after literally just a few seconds, we get the confirmation here. The email is ready for review in the design studio. Let's see how it goes. We have a link to open it up in Customer.io, our email marketing tool. Let me open this one. And here we have it inside Customer.io, our full email is ready. As I said, this was literally just a matter of seconds. It's got the name of the email template here, AI Builder course completion congratulations.
We have the subject line, you did it, your AI Builder certificate is ready. And now we have our full drafted email following the exact style and using our exact 9 x email template um with, of course, our footer, the font that we use, the right button, everything is there. What I really like about this approach is obviously most people know now that uh AI tools are very good at generating HTML, so we could have just had uh Claude create the whole HTML email, but then that's very, very hard to edit.
Here, everything is completely customizable because we're creating it in the tool that we actually use. And as you all know, probably AI is not going to nail it always the first time, so this really is the best of both worlds. I can get Claude to quickly spin up uh an email just with a simple prompt, but then I have it here to make any final adjustments, maybe some wording adjustments, or if Claude has gone completely in the wrong direction, I can of course go back into Co-work and ask it to make changes to this template.
But really, what this unlocks for us, and I think something that applies to many, many companies, is previously there were only a couple of people in our team that used Customer.io, but many people in our team need to send out email communication to our students, to our customers, and there would always be a bottleneck to the one or two people that knew the tool. Now, we've set up this create email marketing skill. This is something that our whole team has access to.
So, now when anyone in our team needs to send out a branded marketing email, they have the full ability to do it themselves. They can simply prompt Claude Co-work. It will create everything in Customer.io. They can make their changes and send it out. This is saving us a bunch of time and allowing us to move away faster with our communication. Now, one bonus feature of Claude Co-work that a lot of people might not know about is its ability to control your browser.
So, thanks to its Claude in Chrome plugin, you can actually have Co-work open up a Chrome window and do work directly in the browser, which opens up our next use case, and that's scraping websites. And to demonstrate this, I want to show you a skill that I built for scraping data from the Meta Ads Library, so that you can analyze any competitor's ad strategy in a matter of minutes. So, all I need to provide this Meta Ads Analyst skill is the link from a certain advertiser in the Meta Ads Library.
So, for those of you that don't know, any company running Facebook ads, Instagram ads, all of their ads are made public that you can see here in the Meta Ads Library. So, in this case, I've got looking at some ads from an AI newsletter. And all I'm going to do is I'm going to copy the link uh of this page and send that one off to Claude. And what Claude's going to do, it's actually going to open a browser, visit this page, and then analyze all of these ads.
And we can see our skill's kicked off, and the first thing it says, "Let me first get a browser tab and identify the advertiser, then kick off the full pipeline." And now Claude is asking for my permission, is it able to use its Claude in Chrome connector and visit the Facebook website? I'm going to say allow for this website. And now it's mentioned that the page is loaded, so let me shift over to my Chrome window. And here we can see Claude is in progress.
Claude started debugging this browser. You can see there's this button stop Claude if we ever want to take over, but right now Claude is in control of my Google Chrome and is right now looking at all of the data on this page. So it successfully was able to determine that this is the Superhuman AI newsletter with 51 active ads. Let me set up the pipeline and start extracting. So what it's going to do now, it's going to go and download every single image and video ad from the Meta Ads Library, analyze every single landing page.
So a lot of advertisers, they run ads pointing at different landing pages. What this Co-work session is going to do now in my browser, it's actually going to go and visit every single landing page from those ads. So it gets a full idea of this advertiser's funnel and marketing strategy. All right, now I've jumped forward. I'm going to like Claude Co-work was really working hard here. This whole process took about 10 minutes.
It was diligently looking at every single ad, extracting it, and now we have our results. So here's what I found across the Superhuman AI's newsletter, 51 active Meta Ads. The strategy in one sentence, a pure top-of-funnel subscriber acquisition machine. Every ad drives to one of two email sign-up pages, no paid offers anywhere. And what's even more impressive was we have this full report HTML that Co-work has created for us.
Let me open this one up. I'm actually going to open it in my browser. And then now we have a full overview of this advertiser's strategy, single-funnel newsletter growth machine. The key takeaways, the copy angle, so what type of messaging are they using in their ads. But probably the most impressive, we have the funnel analysis. So divided up by the different landing pages, so we have the main newsletter sign-up. And for this landing page, every single ad that they are running pointing to this landing page with the copy also analyzed.
And so we can see here of ads driving to the main sign-up. And then they also have a 1,000 Prompts lead magnet with a different set of ads pointing to that landing page. This whole thing would have taken me hours to create, going through and checking out all of these different ads. And this is something that you'd probably be paying a marketing ads specialist a bunch of money to create for you. What's even better is you can actually go ahead and analyze a bunch of different competitors, and then have Co-work use that as input for coming up with your own marketing strategy.
All right, onto our next use case, and that's generating branded assets. And I'll give you an example for us here at 9X. So every week we host a free live AI workshop covering a different topic. And for every event we host, we need to consistently create assets that match a certain template. So if we have a look, for every event we need to have the uh cover image for our Luma event, the thumbnail for the live stream on YouTube, and also a scaled-down version that we use in our email copy.
And obviously the fact that we run these each and every week, you can see just how many of these assets we've had to manually create until now, thanks to Claude Co-work. All right, so over in Claude Co-work, all I need to do is give the name of our next uh 9X Live Workshop, say who from the 9X team is going to be the host. So in this case, I'm going to be hosting the next session on Claude Co-work for marketing. And then I'm going to run the 9X Live Generate Assets skill that uh we have set up, and let's see how that one goes.
So there we go, it's going to start running its skill. It has its to-do list that it needs to generate the YouTube thumbnail, the email image, and the Luma cover. And then it's going to check all three images have correctly been generated. And there we go, just a moment later we have those three images ready to go. Let's take a look what it's come up with. So here we have Claude Co-work for marketing with me as the host.
We have the uh YouTube thumbnail, the email image, which is hard to see, it's just a downscaled version. And of course, we have the Luma cover. And this is in a fraction of the time it would have taken me to actually create these three assets myself. And what's super powerful about this skill, the assets it generates is going to be dynamic depending on the length of the title and how many hosts there are. So here, let's say we're going to create another set of event assets for our live workshop Claude Code for Beginners.
And this time there's going to be two hosts, myself and Alex. And I'll send this one off. All right, and again moments later we've got our three assets. Let's take a look first at the YouTube thumbnail, and now you'll notice it's not just my image there, but also my co-founder Alex is appearing here as a host. Um and the font size is basically dynamic, maximizing the amount of space um in the different image aspect ratios.
Very, very cool. Let's take it one step further. In many cases in our 9X Live Workshops, we also bring in external guests. So imagine here I'm going to be running a Build an AI Agent Live with us session, and I'm hosting it with Jacob Bank, who's the founder of Relay. All I'm going to do is I'm going to add a folder here as a project. And inside that folder, I have two images. One is Jacob's image, that's the other guest that we're going to be having and should be appearing on the event assets.
And relay.app company logo. So I'm going to open this one up. Always allow and then send this one off. So now Co-work's telling us all three images are ready and looking great. Let's see how it goes when we have a guest speaker. And there we go, Build an AI Agent Live with us using that dynamic image that I added in the folder, and of course putting the name and company logo here for the Luma cover. Now, we are obviously using these for our live events, but this is a use case that can be applied across so many use cases, wherever you have templated assets.
Whether that's email headers, blog article images, social media posts, ads, whatever you're using, if it has a consistent structure, you can create a skill like this one to teach Claude Co-work to create this on repeat. And if you want to see exactly how this skill is set up, check the link in the description below where you'll be able to download all of the skills that I'm showing today in this video. Next use case, migrating content.
And one of the ways that we use this at 9X is in our workflow for publishing our AI and automation tutorial. So we draft all of our AI tutorials here in Notion, including screenshots. So for instance, this one which is an tutorial for building an AI meeting agent has the connected YouTube video and all of the steps involved in setting this one up. And once we're happy with the tutorial here in Notion, we have to publish it over to our free community.
But one major problem in our workflow, if I were to just grab this whole Notion page and select everything and copy it, and then drop that into our community tool which is Circle, we can see the text properly gets copied over, but it's missing all of the important images, the screenshots here in the tutorial. And as well, the YouTube video isn't embedded correctly at all. And this just costs our team a bunch of time when it comes to publishing these free tutorials, needing to go through, find the right spot, then upload the image again, and also re-embed any videos that we have.
And now we've handed this work completely off to Claude Co-work. All I need to do is drop in the link to a tutorial that's ready to publish, and then we have of course created our very own skill for this, Publish Notion Tutorial to Circle. And let's send this one off. So what this skill is going to do, it's going to download all of the data from that Notion page, including all of the images, and then prepare it in a way that it can be imported into Circle.
And then we connect directly with the Circle API to create the post. And now we get the confirmation our post has been created. I can just simply open it up here and take a look. And there we have it, our perfectly formatted post. We can already see the YouTube video was embedded correctly. And yeah, perfect. Here we have every single image um from that Notion page downloaded and uploaded, inserted in the exact right place um of our post.
This saves us a bunch of time, and this sort of manual content migration, this happens in so many different workflows where you're working between different tools. And this is the type of work that drive you and your team absolutely crazy. So really think about what part of your work can you actually hand off to Co-work and have it really do the mundane part of moving data, reformatting it, pushing it to another tool um so that you can focus on being creative and putting out more content as opposed to simply just copy-pasting content between tools.
All right, next use case, creating formatted reports. And this is sort of a two-in-one use case cuz it involves both data analysis and then actually taking that data and creating a well-structured and well-formatted report. And how we use it at 9X, one very common use case is whenever we run AI workshops for companies. After we run a workshop, we typically send out a feedback survey to all the respondents so we can really understand the impact that the workshop had and the improvements that are potentially made for the company, which we can share with our stakeholders.
So here you can see an example survey which I have anonymized and redacted, but these are real responses coming from a company where we ran a workshop. Now, if we were to just send this CSV to either the founders or maybe the learning and development team of the company where we ran the workshop, they'd probably say great, thanks, but no one would actually take a look at this. And this is definitely not a good format for seeing what is actually going on.
So over in Claude Work, we have a feedback report skill that we've created, and all I need to do is give it the name of the workshop and the company we ran it for. For this example, I'm using Redacted Inc. to anonymize what's actually going on. So, we can see here that Claude Work actually crunches the data from that CSV, but even more impressive is this document feedback report that it created. Let's open this one up in Google Drive, and here we have feedback report AI for legal and finance workshop, which we ran in March of 2026 for Redacted Inc.
So, we have an executive summary with the high-level stats, the overall rating from the participants of the workshop, the AI impact, the intent to use AI, all of those numbers crunched and dynamically added into this well-formatted document. And this document is something that people will actually read. What did the participants like most? What are areas for improvement? What use cases are demanded and next steps. And this is something that we can send directly to the client.
Obviously, we'll take a look at this, and the beauty of the fact that this is a document which is then editable in either Microsoft Word or Google Docs is we can have AI come up with a very, very good first version, check that everything makes sense, obviously check that these numbers that no hallucination has gone on, and then very quickly send this out to the company. All right, onto our next use case, and that's triaging emails.
I receive a bunch of trash email, so I've set up Claude Work with a skill that it can actually take a look at emails I recently received, filter out all the noise, and just alert me to the ones that I actually need to act on. So, over in Claude Work, I have this email triage skill, and the first thing this does is connect directly to my Gmail and pull in my recent emails. And once it's read my recent emails from Gmail, I've actually set up an Airtable base where whenever it determines that there's an email that requires my attention, it's going to save it in Airtable, and that way it knows if it runs this again that it's already alerted me about it.
And then finally, if there are any emails that I need to be notified about, it's going to alert me on Slack. So, we can see here it is finished, email triage complete, scanned 28 emails, found one new actionable email. And here we can see that Slack message nicely formatted. I've obviously had to blur a lot of this for confidential reasons, but here in one click, I can just open in Gmail. We see here 28 emails check, one needs your attention.
I can just open this one up, and then directly reply here underneath. And of course, I could even take this further and have it directly draft an email reply based on previous knowledge that I give it. Now, you're probably thinking quite rightly that it's always a bit pointless needing to go into Claude Work and always trigger the email triage skill, and you're absolutely right, and that's why I've set this up as a scheduled task.
So, we can actually here under scheduled create a new scheduled task. You can see I've got mine set up the email alerter. So, the instructions are quite simple, it's to use my email triage skill, and this runs every single hour. All right, onto our final use case, and that's creating agent skills. So, as you've seen, all those seven use cases we just covered involved skills in Claude Work, also referred to as agent skills.
And actually, Claude Work comes with a built-in skill to create your own skills for you and your company. So, I know that might sound a little bit meta and a little bit confusing, but let me show you what I mean. Here I have a task that I've completed with Claude Work. I went back and forth with Claude Work on a task, and the task actually involved taking one of our AI tutorials that we draft in Notion, and I asked Claude Work to publish this for me inside our community platform, which is hosted on Circle.
And actually, in order for Claude Work to publish this post in our Circle community, it took quite a little bit of back and forth, me explaining some things, Claude Work tried to interact with a bunch of APIs. We can see some connectors here. So, I want to make sure that all of this back and forth that I've just had with Claude Work in order to get to the outcome that I wanted, which was basically taking our Notion content and publishing it to Circle, so that we can do this now on repeat.
I want to turn this into a skill. So, baked inside of Claude Work, we actually have a skill creator skill. You can see here create new skills, modify and improve existing skills, and measure skill performance. This is a way that we can take our processes and teach Claude how to do them. So, in this conversation where I've just had this back and forth of publishing a Notion tutorial to our Circle community, I can paste the following prompt.
Can you please use your skill creator skill to create a skill called publish Notion tutorial to Circle, which takes as the input the link to a Notion page that's ready to publish, and then follows all the processes we just completed here in one go. And this is my big recommendation when creating skills for the first time. Skills are something that should be iterated on. So, simply start a session with Claude Work, go through the process once.
Once you're happy with the result, tell Claude Work to create a skill so that it knows how to do it on demand. And this is something that you can then share with the rest of your team. So, here we go, I'm going to send this one off, and we can see that it's running its skill creator skill. All right, now after a few minutes of work, in Claude Work has finished, and you can see it's created the skill here, and in one click, I can save this skill to my Claude Work account.
So, I'm going to click save skill, and now you can see skill saved. And to test that my skill has been successfully added to my account, I can simply open up all of the skills by typing forward slash and then writing out the name of the skill. And here we see publish Notion tutorial to Circle, and now I can use this skill whenever I need to publish one of our AI and automation tutorials to our community, and I don't need to have this back and forth with Claude Work explaining the process every time.
To see all of the skills that you've created here in customize, we can then click on skills. And if we take a look at that publish Notion tutorial to Circle skill, we can see the skill MD file. So, this is the full explanation of the workflow that Claude Work should do every time this skill is called. And on top of the skill markdown file, we have some additional reference files, and in this case, Claude decided that we need to create some scripts.
So, it wrote a full Python script that will take that Notion data, format it, and prepare it in a way that can be imported into Circle. What I always recommend is really thoroughly test your skills. Try this out a few times, and then if you're not happy with it, you can also ask Claude to edit it. So, here in the top right, you can just say edit with Claude and tell it what it's currently not doing right, and it will actually fix that skill.
And once you're ready to share this with your team, you have two options. You can download this skill and then send it to them on Slack or via email, or you can go ahead and create an organizational plugin to share a bunch of skills together. Now, if you want to test out any of these use cases for yourself and also get them working for you and your team, all of the skills that I've created, we're giving away for free inside the 9 x community.
You can join for free, just check the link in the description below. However, one thing I want to point out, all of the skills that I'm sharing are very specific to the way we work at 9 x and the tools we use. So, when you want to go ahead and import them, what I recommend, just upload the skill into your Claude Work chat, and then ask it to tell you what the skill exactly does and which tools are required, and then it's going to explain exactly how the skill is set up, and then you can modify it to exactly your use case.
Now, Claude Work has made using AI agents incredibly simple for business professionals, but there's an even more powerful version, Claude Code. And despite what you might think, it's not just for developers. So, click this video right here, where I show you eight use cases that any professional can get value from. Things like building an AI agent that prepares your entire newsletter from a single command, a finance agent that can process 50 invoices in a matter of minutes, or deploying full websites straight from Notion.
If you want to see how far you can really take this, I'll see you over there.
The words are the caption track's own and nothing is reworded or re-transcribed. Paragraph breaks are placed between sentences so the text reads as prose.
Free tools for your own script. No signup, no login.
Paste your draft and see where viewers are likely to drop off, with a rewrite for each weak line.
Paste the first 30 seconds of your own draft for a hook score and rewrites.
Check your draft against YouTube's advertiser-friendly guidelines before you record it.
Read this channel's public videos and transcripts, and download a writing brief for it.
Sentence shape
| Measure | This transcript |
|---|---|
| Sentences | 271 |
| Average words per sentence | 20.5 |
| Longest sentence | 69 words |
| Questions asked | 5 |
| Sentences containing a number | 25 |
Most used terms
Filler phrases
67 in total: actually 29 · uh 13 · like 8 · um 8 · basically 3 · literally 2 · sort of 2 · I mean 1 · right? 1.
A literal whole-word count of the same phrase list the Prepublish browser extension uses, so a phrase inside another word is not counted and a phrase used in its ordinary sense still is. It is a count and not a judgement.