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.

Jeff Su · @JeffSu
This video has no Most replayed graph yet: YouTube shows one only once a video has enough views. These are the moments viewers replayed most in Jeff Su's most watched videos.
Most replayed moment at 3:13
3.7x that video's typical replay level
will change and wait for confirmation. Think of these like training wheels on a bicycle, right? You can remove them as you get more experience with Co-work. Next, under settings, capabilities, enable both memory features. I leave location metadata off, like that's going to help with data privacy.
Said at 3:07
Most replayed moment at 11:35
4.1x that video's typical replay level
more than creativity and you have source materials to check against, use Notebook LM. There are a few more specialist AI tools I use but didn't make this list because I don't use them every day. But to quickly go through them, Gamma for presentations, 11 Labs for voice cloning, Zapier and N for automation,
Said at 11:28
Most replayed moment at 7:56
4.5x that video's typical replay level
OpenAI, GBT 5 is extremely good at critiquing and improving its own instructions, which is a theme we'll see again later on. Pro tip, add this meta prompt to your texture spanner app so you can quickly access it like this. So you basically have OpenAI's prompt optimizer tool available for free. By
Said at 7:49
The graph counts replays. It does not show where viewers stopped watching.
Words
2,800
Runtime
14:37
Speaking pace
192wpm
Reading time
12min
192 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)
This is my inbox folder and throughout the day it fills up with files I want to keep. Reports, spreadsheets, screenshots, receipts, etc. But instead of organizing all of this myself, my AI automatically reads each file, gives it a descriptive name, and moves it to the right folder. It's been doing this every day without fail for the past 5 months. Now, if you think that sounds too good to be true, there is a catch. You need to spend around 15 minutes upfront creating a rulebook to teach your AI how to organize files
96 words, the words spoken in the first 30 seconds at 192 words per minute.
Free, no signup. See how the first 30 seconds hold attention, with rewrites.
Sentence shape
| Measure | This transcript |
|---|---|
| Sentences | 167 |
| Average words per sentence | 16.8 |
| Longest sentence | 44 words |
| Questions asked | 6 |
| Sentences containing a number | 13 |
Most used terms
Filler phrases
27 in total: like 13 · right? 5 · basically 3 · you know 2 · I mean 1 · actually 1 · literally 1 · sort of 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.
Run the check on the words above: where attention is likely to drop, with a rewrite for each weak line. The free check shows the scores and the one issue costing the most.
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.
This is my inbox folder and throughout the day it fills up with files I want to keep. Reports, spreadsheets, screenshots, receipts, etc. But instead of organizing all of this myself, my AI automatically reads each file, gives it a descriptive name, and moves it to the right folder. It's been doing this every day without fail for the past 5 months. Now, if you think that sounds too good to be true, there is a catch. You need to spend around 15 minutes upfront creating a rulebook to teach your AI how to organize files according to your preferences.
And the best part, the framework I'm about to share works across all the major AI tools like Taj, Claw, Gemini, and across all the major storage platforms. So, this setup's going to work no matter what tool you use. Let's get started. Diving right into a situation we've all found ourselves in. Say your doctor sends you your latest health report over email. After clicking download, the PDF lands in your downloads folder.
At this moment, you know, you should give it a useful name like 2035 annual health report. And you know, you should organize it under your health folder, resources, annual health reports, so that you can find it easily next time. But that's a lot of work. So maybe you just throw it under the health parent folder, telling yourself you'll do it later, right? Or you just leave it in downloads. So the problem isn't that we don't know we should organize our files.
We all have a rough idea in our heads of how we want to do it. The problem is having to do that tedious work every time a new file comes in. So, wouldn't it be great if we can just set the rules and have AI do the grunt work of renaming and filing for us? On that note, let's cover the three essential things you need to include in your rule book. First, an organization framework that tells the AI where a file belongs and why it belongs there.
Second, naming conventions so the AI keeps the file names consistent and easy for you to find later. And third, a filing workflow that tells the AI when to ask for clarification. Starting with the organization framework. Longtime viewers know I'm a huge fan of Thiago Forte's PAR framework that organizes your files by projects, areas, resources, and archive. I won't waste your time nerding out about that today, so I'll link to Thiago's video below.
My version is the area first parap framework which I think fits more naturally with how we look for files. You basically start with an ongoing part of your life or work. For example, health. And within the health area folder, we have a resources subfolder that holds material I'll keep referring to like a folder of annual health reports or a standalone insurance claim template for reimbursements. We also have a project subfolder that holds active work with a clear endpoint like researching and planning for lasic eye surgery and an archive subfolder that holds completed projects or outdated records I want to keep like health insurance coverage from previous years.
It's the exact same idea for workrelated areas. For example, in this product marketing area folder, product marketing resources might contain a brand kit holding my logos and brand guidelines. The projects folder contains projects I'm actively working on, and the archive folder holds completed or inactive material. Moving on, naming conventions make files easier to find. So, as a rule of thumb, use a consistent pattern based on the details you'll look for later.
For example, if I have a standalone document, I keep the name short and descriptive, like email subject line best practices, so I know what's inside without opening it. For project folders, I put the year and quarter first, like 30 Q3 trip to Seattle, so I know when the project started and what it's about. And for records of a specific event or transaction, like receipts and invoices, I include the full date, as in help scout invoice 20309, right?
So I can identify the particular record I need. Finally, the filing workflow basically tells the AI what to do when it runs into something the rulebook doesn't cover and how to save our preferences so it knows what to do next time. On that note, let's jump into a real example. First, create an empty folder somewhere on your computer. I'll call mine JS root folder and inside it, create 00 inbox 01 reference, personal finances, business finances, health, and travel.
Next, open the starter file management rule book I've linked below. Go to file, download, markdown.mmd, and move that file into 01 reference. And this gives us a plain text rulebook the AI can read and update. Then give your AI access to the root folder you created earlier. And I'll name my project file management. And although I'm using chatbt work in this example, you can use the same rulebook approach in claude codeex and Gemini spark.
And just to be clear, this will not work in a normal web chat. you need to use the desktop versions of Kashibb, Claude, and Gemini since they need direct access to the files on your computer. Now, I'll move these four files I know I want to keep from downloads into the inbox folder. Then, back in Chad PT work, I'll say, "Read my rulebook in the 01 reference folder and help me organize the files in the inbox folder." And we're just going to fast forward a little bit here.
First, the AI gives us the bottom line up front, followed by a breakdown of its proposed changes. each file's original name, proposed new name and proposed location. For example, I picked travel for the itinerary, and health for the visit transcript, and nothing has been moved yet. But check this out. The AI is asking me to clarify whether these two purchases are personal or business because it can't tell from the receipts alone.
And so I reply, "My personal credit card ends in 888 and my business card ends in 9090. Go ahead with all four files and remember the card rule." And we'll fast forward again. And just like that, all four files are renamed and moved to their respective folders. And now if I open the rule book and look under the organization framework section, you can see the AI saved our card rule. So it knows how to process receipts correctly next time without asking me.
Although that was a very simple example, it's the exact same process as your folders grow. For example, within our health folder, we can add the resources, projects, and archive subfolders we covered earlier with annual health reports and doctor visit notes inside resources. The AI can use a rulebook, those folder names, and each files contents to work out where it belongs. Pro tip, you can even ask the AI to analyze the files in your folders and recommend new subfolders to create and group your files in.
For example, not to jump ahead, but this is my actual business records folder in Google Drive. And it was Claude that first created these four subfolders. And second, grouped and moved my files into these four subfolders. Put simply, the rulebook gives the AI the organizing principles. And your folder structure and saved preferences tell the AI how to apply them to your files. Now, I've been using this exact system for my YouTube production workflow in Dropbox with my editor.
And luckily for me, Dropbox agreed to sponsor this video. For context, I used to manually rename and upload files into separate folders for my video editor. But now with Dropbox connected to Chachi PT work, I can just give it access to Jeffos, the folder on my computer containing my rulebook. Then dump all my video files into a inbox folder in Dropbox and say, "Organize the files in my Dropbox inbox for my editor using my handoff rules.
Show me the plan first. And let's fast forward. Here we can see that Chacht proposes new names and locations for the files. And after reviewing the plan, I can just say approved. And after fast forwarding again here, we see the files are renamed and moved from the inbox into that video's Dropbox folder. And now that Dropbox in Chachib has organized my files using my handoff rules, I can ask it for an email draft with the folder link, paste into Gmail, and send it to my editor.
He opens one Dropbox folder and everything's ready to edit. To do the same with your team, click the link below to try Dropbox in Chachi PT. Thank you Dropbox for sponsoring this video. All right, now let's cover a more realistic example where our files are stored on a cloud platform like Google Drive and where we already have some sort of file management system in place. First, go ahead and install the Google Drive for desktop app and sign in.
This makes your drive folder available in Finder on the Mac or File Explorer on Windows. And the purpose is so that we can give AI access to them, right? And any changes you make here will sync back to your Google Drive. You don't have to use Google Drive, by the way. This works for the Dropbox desktop app, the One Drive desktop app, and even iCloud Drive. And yes, I know you can connect to the Google Drive plugin or connector, but to oversimplify, connecting or rather sharing the Google Drive folder on desktop directly makes it much easier for AI to process those files.
All right, back in ChachiBT work with the same folder mounted, I'm just going to click the plus icon, select my Google Drive folder, and give it access to the root my drive folder. And although I'm using chatbt work in this example, the workflow is the same in Claude. In fact, I'd say it's even more intuitive in Claude since you can just mount multiple folders to the same chat. Pro tip, since we're just asking the AI to look at the folder and file names to understand the structure, we do not need to make the entire drive available offline for this exercise.
The AI can still read the basic metadata. Now, before moving anything, I want the AI to give me two snapshots. First, the current state showing how my files are organized right now. and second, a proposed future state, showing how the files could be organized using our rulebook. And comparing the two makes it easier to see what needs to move and where. And to do that, I'm going to use this prompt, which I'll also link down below, to get both snapshots as readable folder trees without changing anything.
You'll see what I mean. And let's just fast forward this part. All right. So, even though I've done this many times, it's still pretty incredible to see AI map out my entire drive hierarchy. Like my top level business folder, Sushi Labs, has zero standalone files sitting directly inside it, but it has 805 files total. And I can see all the subfolders underneath as well. Next, the AI flags the main structural problems such as my business finance records living in two different locations and a quick share folder sitting at the root level, which goes against my own framework.
Now, I want to be very clear. There is no right or wrong way to organize your files. Even though I follow my area first parap framework religiously. I share a lot of files and free templates with you guys. So I want to keep the quick share folder in the root level for convenience. Right? So I'd write that as an exception in my actual rule book. Back in chasht work under the proposed future state section we see the AI recommends reorganizing our files in this way.
For example, under the parent sushi labs folder, the AI recommends five subfolders. And even though it only shows the number of files for now, we can ask it to specify what files go where after we've answered the questions below. Here, the AI is asking clarifying questions such as whether or not my accounting records should move to business finances or remain in business operations. Right? And although I'm skipping these in the interest of time, when you do this, actually answer these questions so the AI learns your preferences.
Obviously, your folders and suggestions will look different from mine, but the process stays the same. Give the AI access to your file management platform. Ask it to map your current state. Then ask it to propose a future state using the starter rule book. Once you're happy with the proposal, ask the AI to walk you through which files and folders to move and where, step by step. Start with area folders you use frequently and move existing folders together where possible.
Ignore file name changes for now so it doesn't become a huge renaming project. To be clear, you could let AI make the changes, but I recommend doing this first pass yourself with the AI guiding you so you get familiar with where everything lives. And unlike the web version, the Google Drive desktop app makes the process super easy since we can literally just select and drag files and folders into their new locations.
Pro tip, if you want to start from scratch, create a legacy folder at the root level and move all your unsorted files and folders into it. So, it's like starting with an empty root folder. That way, you can build your new folder structure from scratch and gradually bring over what you need. By the way, the free starter rule book I've linked down below is more than enough to get you started. But if you want the optimized file management system I use in my own workspace, you might want to check out the Co-work Academy course link down below.
All right. Once you've organized your existing files, let's create a schedule task to help keep them that way. Staying with our Google Drive for desktop example, create a 00 inbox folder at the root of my drive. And pro tip, add this folder to favorites in Finder or Quick Access on Windows so it's easy to get to. Next, drop a few files you would realistically want to keep into the inbox folder. And with that folder mounted, ask the AI to organize those files using your rulebook and give it feedback until the files are named and filed the way you want.
I've covered schedule task multiple times in other videos, so I won't waste your time here. But basically, once you're happy with the results, stay in that same chat and ask the AI in plain English to create a daily schedule task that checks your inbox and proposes changes for you to review and approve. Pro tip, make sure the schedule task runs in the project folder containing your file management rule book and has access to your Google Drive folder so that it can read your rules and find the files it needs to organize.
Here's a report from my scheduled run earlier today. And notice there are no follow-up questions since the AI knows my preferences so well at this point. So, I just reply approved and all files are renamed and organized. After you set this up, you'll probably give a bit more feedback in the first few days as the AI runs into new situations and learns your preferences. But after two weeks, the reports should be around 95% accurate with minimal input from you.
By the way, if you want to get 1% better at AI every week, sign up for my free newsletter to receive one insanely actionable AI tip you can apply in under 2 minutes. Link down below. As a quick recap, your day-to-day workflow should look something like this. You drop the files you want to keep into your inbox folder and the schedule task gives you a daily report with proposed changes. You review it, give feedback and approve.
Then the AI renames and files those items for you. And when you ask it to remember a preference, it updates the rule book so you don't have to repeat that feedback next time. If you enjoyed this, you might want to check out my CHAPC work tutorial next. See you on the next video. In the meantime, have a great one.
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: paste a draft and see where it stands before you record it.
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.