Big Idea

The argument in one line.

A two-layer Claude Code folder architecture separating orchestration (CLAUDE.md) from target-specific detail (YAML per ICP) lets you describe an audience in plain language and get a clean Google Sheet of leads without touching a scraper UI.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You run a cold outbound agency and currently export leads manually from Apollo, LinkedIn, or scraper dashboards.
You have a Claude Pro or Max plan and want to extend it into async business automation beyond coding tasks.
You want to delegate lead generation to an agent that self-qualifies results before asking you to proceed.
You manage multiple clients with different ICPs and want a single workspace that keeps each audience isolated.

SKIP IF…

You have no cold outbound use case -- the system requires Apify plus Google Cloud credentials.
You are looking for a no-code tool; this requires terminal comfort and OAuth setup.
Your leads come from inbound or referrals and you have no need for volume prospecting.

TL;DR

The full version, fast.

The system works by giving Claude Code a two-layer workspace: a CLAUDE.md that defines the agent role, and YAML files per client encoding specific ICP filters. When you ask Claude to scrape leads for a client, it reads the ICP YAML, calls the Apify actor with the right parameters, runs a test batch to self-qualify, then exports to Google Sheets via OAuth credentials.json. Lead generation becomes a background task -- you speak, Claude runs it while you do other work, and a clean sheet appears when done.

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 – 06:47

01 · Live demo

Full working system shown cold: ICP prompt to Claude, Apify test runs with self-qualification, export to Google Sheets. Viewer sees the end state before any setup.

06:47 – 09:28

02 · Prerequisites

Claude Code (Pro/Max plan), terminal choice (Cursor or Antigravity), Apify account ($29/mo), Google Cloud project with Sheets API.

09:28 – 12:30

03 · Folder architecture

The two-layer design: CLAUDE.md as root orchestration file, .env for keys, clients/ with per-client README + ICP YAMLs, execution/ with generic Python scripts.

12:30 – 16:47

04 · Setting up the scrape

Paste CLAUDE.md into terminal to instantiate folder structure; add lead-scraping skill; supply Apify actor slug and API token via .env.

16:47 – 20:45

05 · Google Sheets OAuth

Create Google Cloud project, enable Sheets API, create OAuth desktop client ID, download credentials.json, drop in project folder, complete browser auth flow.

20:45 – 23:22

06 · Final result and CTA

Clean lead sheet with industry, job title, email. Instruction to save scripts for reuse. Soft CTA: free Skool resources and paid implementation call.

Atomic Insights

Lines worth screenshotting.

Separating orchestration from target configuration lets one Claude Code workspace serve unlimited clients without rewriting prompts.
Claude Code will self-qualify a test batch before a full run if CLAUDE.md asks for confirmation -- so bad filters surface before you burn API credits.
The Apify actor slug is the only scraper-specific thing you swap; the Python execution script stays generic across all ICPs.
Dropping credentials.json in the project root is all Claude needs to authenticate Google Sheets OAuth -- no manual token exchange required.
The two-layer architecture improves over time: Claude updates skills files after each successful run, making the system more accurate per ICP with use.
The demo-first video structure (show result at 00:00, explain setup at 06:47) lets skeptical viewers validate the outcome before committing to a 23-minute tutorial.
Antigravity and Cursor are used purely as terminal UIs -- no paid plan needed since Claude Code runs inside its own terminal context.
Storing session context in a .tmp file lets Claude carry ICP details across fresh sessions without re-prompting.
The agency CTA is soft: free resources via Skool first, paid implementation call second -- authority-before-ask structure.
Running Claude Code with dangerously-skip-permissions means the agent never pauses for approval; safe only once you trust the system.

Takeaway

Build the workspace once, describe the audience every time.

WHAT TO LEARN

When the agent knows what kind of agent it is, a simple ICP description is all it needs to run a full scrape, qualify results, and deliver a clean sheet.

Separating orchestration from configuration means the same agent can serve a new client without rewriting any prompts -- just add a new YAML with their ICP.
A demo-first video structure that shows the finished result before the tutorial lets skeptical viewers confirm the outcome is real before committing to the walkthrough.
Having Claude run a small test batch and report quality metrics before a full run is a built-in quality gate that catches bad filter configs before you burn API credits.
Dropping credentials.json in the project root is enough for Claude to handle the full Google OAuth flow -- no manual token management needed.
Instructing the agent to save scripts to their appropriate locations at the end of a session makes the system compound: each new scrape refines the execution layer for the next one.

Glossary

Terms worth knowing.

CLAUDE.md: A markdown file in the project root that Claude Code reads at session start. It defines the agent purpose, capabilities, and folder conventions.
Apify: A cloud platform with over 1,500 pre-built web scrapers (called actors) that extract structured data from websites. Billed per actor run.
ICP: Ideal Customer Profile -- the attributes (industry, company size, job titles, location) that define the target audience for a sales outreach campaign.
Two-layer architecture: A folder pattern separating generic execution logic (Python scripts) from client-specific configuration (YAML files encoding ICP details), so logic and data never mix.
credentials.json: An OAuth client secret file from Google Cloud Console that allows a desktop application to authenticate and write to Google Sheets on your behalf.
Antigravity: A terminal application with a built-in AI agent panel. Used here purely as a terminal wrapper for Claude Code -- no paid plan required.

Resources

Things they pointed at.

01:39toolApify ↗

00:55productClaude Code ↗

01:24toolAntigravity terminal

01:24toolCursor

16:47linkGoogle Cloud Console ↗

23:00linkSkool community (free resources) ↗

23:10linkctrlflw.com agency ↗

Quotables

Lines you could clip.

00:35

“Claude Code is now my orchestration layer, and it is how I run my business.”

Bold claim, no setup needed, speaks to the target viewer aspiration→ TikTok hook↗ Tweet quote

04:50

“Things that might have taken all of your focus can now just be running in the background while you are on a call or filming a YouTube video.”

Meta moment -- he is literally filming while Claude runs in the background→ IG reel cold open↗ Tweet quote

20:20

“Once you configure it, it only gets better.”

Clean one-liner capturing compounding value of the skill-based architecture→ newsletter pull-quote↗ Tweet quote

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.

metaphorstory

00:00Hey. What's up, guys? David here.

00:02In this video, I'm gonna show you how to use Cloud Code to scrape leads on your behalf in seconds. Um, you no longer need to click through all these different platforms, export into Excel or Google Sheets, and then clean up the list from there. Cloud Code will dedupe, export, clean, and even scrape for you.

00:19I've been using this for my own outbound agency. Um, I've honestly adopted all of my workflows to have Cloud Code be a proxy for everything.

00:29And it's now my orchestration layer, and it's how I run my business. And so in this video, I want to arm you business owners who are looking to leverage AI or entrepreneurs that are building a business around AI with this information so you can work more efficiently. So let's dive right in.

00:44Um, first, we're gonna need a couple of things. We're gonna need that's not the right one.

00:48We're gonna need Cloud Code. So in order to access Cloud Code, you're gonna need a plate a paid plan on claude.com, or is it claude.ai?

00:56Yeah. It's claude.ai. You're gonna need a paid plan.

00:59The $20 plan is fine. You'll just have less usage. I use the max plan, and it works really well for me, but that's because I use Claude code for everything.

01:08So, um, if you're just getting started, the $20 plan is perfect. Next, you're going to need a terminal. You can use the terminal inside, uh, your Mac or, uh, Windows computer, totally fine, but I do recommend cursor antigravity just because the UI is it's easier to work with, if you will.

01:25And so you're gonna need a terminal. Um, if you do go with Cursor or anti gravity, you don't need a paid plan. We're gonna be using ClodCode inside of the, uh, their own terminal, so there's no need to upgrade or or purchase any plan for them.

01:39Next, we need an Apify account, which is actually 29Dollars, not 39. So let me fix that.

01:45So $29. So far, we're looking at $50, but guys, if you use this setup, you no longer need a VA or you can save your employee 20 to 20 to $50.

01:56You could save them hours, honestly, because you're just gonna be doing this or they're gonna be doing this on your behalf. Next, we're gonna need a Google Cloud project with Sheets API access enabled. I'm gonna walk you through that real quick.

02:06So this is everything that you need. Um, now I'm gonna go into a quick demo of how this looks or how this works. I'm gonna open up my anti gravity folder.

02:13We already have a terminal open, so I'm gonna open up Claude. And this is tailored to my workspace. I actually already have a to do list to scrape leads with, uh, for one of my clients, so that's what I'm gonna do.

02:25I'm gonna say, hey. Hey. We need to scrape leads for Cosmic, uh, using their Apify token.

02:30I believe I needed to scrape for small video production agencies and, um, companies that are creating documentaries or independent films. Go ahead and return the parameters for that.

02:43Let me confirm, and then we'll run the scrape on a test run. So what it's doing right now is looking through its skills and, um, the scraping scripts that it's gonna be using.

02:54So here we have a proposed let me zoom in. Okay.

02:59So here we have a proposed list of, uh, industries, job titles, locations, company size.

03:05So this looks good. This looks like it's it's, you know, from my ICP or my client's ICP, and it's gonna ask me some questions before we run. Um, go ahead and proceed with a test run of 50, verify the accuracy, and then we'll do a full run.

03:20So, um, just to show you what's going on while that's working here, the way that I have my ClawdCode set up is based off of folder structure. So you're gonna give ClawdCode access to a folder, um, and inside that folder is a bunch of files and other folders in which Claude knows what to access based off of, um, how you structure everything.

03:39So first, you have the Claude dot m d file, which lives in the root, and this file is in markdown. It's just a text file that tells Claude what its purpose is, what it's meant to do, its existence. And, um, the one that I'm gonna share with you is a two layer architecture that essentially tells Claude it's an orchestration layer for whatever your use case is.

03:58Right? And then next, we have the dot e m v file. That's gonna store all of our keys.

04:02In this case, it's our Appify key. It's not Apollo. Um, and then for me, because I run an outbound agency, I have a client's folder, and each client has their own ICP or audience.

04:13Right? This allows me to use natural language and tell Claude, hey. We're working with this client, client two, and then we're targeting audience two for client two.

04:23So I'm gonna go ahead and run the scrape. And so in just a second here, we're gonna get our, um, output.

04:30I'm actually gonna get ready for the output and say, go ahead and create a new tab once you're done and export those leads in a clean format that we typically have for our clients. Okay. Cool.

04:41So it knows where it's gonna export everything. I'm gonna show you guys that. Um, but, yeah, I'm here talking to you, and, like, Cloud is just doing this work for me.

04:49And so now you can see where the direction of, um, leverage with AI is going because things that might have taken your all of your focus or all of your time can now just be running in the background while you're running or filming YouTube video on a call or even just handling more clients. And so that's really the value here.

05:07So, um, I'm gonna show you guys the end result in just a second. Okay. So we have, um, the Cloud Code Scraper is now telling us everything looks good.

05:17So it even went ahead and gave us metrics on the scrape. So it looks like the target industries are accurate, media production and animation.

05:2796% have under 50 employees. Most of them have emails.

05:31And because I gave it that previous instruction to export the leads, it already knows where to put the leads, and so I'm just gonna hit proceed. I typically like to, uh, run Cloud Code with bypass permissions. I just forgot to do it on launch right here.

05:45I don't really recommend that unless you've already are comfortable with it, but in just a second, we'll see it exported to our sheet. Okay. So the scrape is complete.

05:57It's been exported to its appropriate sheet. So I'm just gonna go ahead and open that. And you see here, this is the exact format that I use for my outbound for my clients.

06:06Why don't I just hide that? Cool. And so this is this is exactly what I need.

06:11This is what I'm looking for. And so that's how easy it is.

06:14Right? If you guys have ever scraped leads before, but sometimes you'll get a bunch of fields that you really don't need, um, and this is the exact, uh, column header structure that I use.

06:25I also would like to add a couple of things after the stage column that lets me know where we're at because I'll do additional enrichments for, uh, personalizations and things like that, but it's that easy. So you just saw me do that in seconds. It didn't really take anything besides me speaking to my computer and, uh, this happening.

06:42So I'm gonna show you guys how to do that. So going back to our setup. So okay.

06:48So now it's time to install Cloud Code. So what we're gonna need is to open a terminal, and we're gonna install using this one line command here.

06:57So you can use anti gravity. That's totally fine. I'm gonna open up a fresh anti gravity instance.

07:03Okay. So when you open up anti gravity, it's going to ask you to, um, open a folder. So this is where you are gonna wanna give Cloud Code access.

07:13You can install Cloud Code globally so it has access in any terminal, which is what the command that I'm giving you is is gonna do. So this g, this dash g right here is what tells your, uh, computer to install Cloud Code globally.

07:26But, um, when we open up a folder, we're gonna work with Cloud Code inside a specific folder. Um, that's how we just keep things organized, and you don't really want Cloud, uh, generating files all over the place. You just wanna keep it in one folder.

07:39So go to antigravity. We're gonna open a new folder. So new folder.

07:44We'll call it, uh, YouTube demo. We're gonna open this folder.

07:48Cool. So now we have a fresh folder. Um, on the right hand side is where you can use anti gravity as a native agent.

07:57So, like, Gemini Pro, you can use Claude, you can use GPT, but, honestly, we don't need all that. And on the left hand side is our explorer tab that shows us the folder structure.

08:07So I'm gonna leave that open so you guys can see what's going on. But at the top of your, um, computer, if you're using a Mac, you just wanna go to terminal, and then we'll open a new terminal. And now we have a terminal inside anti gravity that, um, is directed within the YouTube demo folder.

08:25So this is exactly where you'll type this command right here. So you're just gonna copy that, paste it in here, and just let it do its thing until, um, it's completely finished.

08:36And then once it's finished, all you're gonna do is type in Claude. I already have it installed, so I'm not gonna do that for you. But that's all you need to do is type that command.

08:44And in my case, I don't like to have permissions on, so I'm gonna put dangerously skip permissions. Cool.

08:50So notice how for me, Claude Code already knows what account I'm using.

08:57I'm using the Claude Max plan that I paid for. But when you launch Claude Code for the first time, it's gonna prompt you to log in. It's just gonna launch a browser, log in using the same email and password that you did for your claude.ai subscription, and then you're good to go.

09:10It's good. It takes you two seconds. So once you're in here, you can see on the left hand side our YouTube demo.

09:15We don't have a folder structure. We don't have anything. And so what we're gonna do is create one.

09:20It's gonna be called the claud dot m d file, and I actually have one right here that, uh, outlines the two layer architecture that I just described earlier in this video. And we're we're just gonna copy this, paste it in here, and it's gonna create our folder structure for us.

09:36And this is, uh, specifically designed for my outbound agency, and I'm giving it to you guys for free. So, um, if you run outbound agency, this is perfect for you.

09:44If you run a business, you might wanna configure this a little bit differently, but let me just show you what's going on. So the claud dot m d folder or file rather, this is the agent's instructions.

09:56It's its purpose. Claude will always read this whenever a new session starts. The skills folder, Claude skills, this is what allows Claude to, uh, reference things that you've done in the past.

10:07So when you create a new skill, it gives it context as to what we're looking to accomplish, um, in an agnostic fashion. So a skill, like a task in an agnostic way. Um, the dot EMB file, that's where we hold all of our API keys and, um, tools act tool access.

10:24The dot TMP file, this is just for quick one off scripts or even session context. That's where I keep my session context. The dot archive, we really don't need this.

10:34I used to keep that, but I don't have that anymore. The configs is our project specifications. So for me, I actually have my, um, client live in the root.

10:43So I'm gonna get rid of this. We'll do this instead. Cool.

10:47So in the root, have our client folder. Inside the client folder, we have the client name, different clients, of course. And then within that client, uh, name folder, we have a read me that gives us our overview, and it references where the API keys are, which is usually in the dot EMV file.

11:02Um, then we have the ICPs for my clients, and, uh, this is essentially our audiences. Some clients have more than one ICP.

11:08Right? So, um, anyway, we have the scrape dot YAML.

11:12This is what allows me to scrape for different audiences, um, that I've already preconfigured for each client. And then the qualifying copyright dot yaml, I'm not gonna get into that, but this is, uh, what I use to do additional qualification when once I export leads into a sheet, and then I even copyright variables right inside the sheet directly.

11:31Um, but I'll I'll record that for you guys in another video. Okay? So now we have our execution folder, which lives in the root of your project folder, and this is where all the Python scripts live.

11:43The Python scripts are, uh, agnostic to, like, the so, like, for example, we have a Python script to scrape Apify, which is what I'm showing you guys how to do. But then the scrape YAML is dependent on a specific ICP.

11:56So the scrape YAML gives the execution Apify scrape, like the Python script, the filters that it needs for a specific ICP. So the Python script is, like, the general, uh, execution task, and then the YAML is, like, the details.

12:10And then we just have these other folders that will also have other scripts inside of them. So, um, this is a two layer architecture. I'm gonna give this for free for you guys.

12:18Just click the link in my bio. It's totally free. You can copy this exactly like how I'm gonna show you.

12:23Go ahead and go to anti gravity. You're gonna paste it in here, and then you're gonna tell, uh, Cloud Code instantiate this YouTube demo folder.

12:32Um, based off this cloud dot m d file, go ahead and create the whole, uh, folder structure. Cool. So I fed it to Cloud.

12:38Now it's gonna create the full workspace structure for me. If you look on the left hand side, you can see all these things were just created. Right?

12:46And the next thing I'm gonna do is tell that we wanna scrape leads. So, um, we have this lead scraping skill.

12:54I'm just gonna copy it. Once again, this is also a free resource. You can just click the link in my bio, and, uh, you guys can take it.

12:59So I'm just gonna hit so I pasted the lead scraping skill. Now it's gonna create the skill, and what we're gonna wanna do next is, um, use Applify.

13:09So while this is creating the skill, go to Appify, go to the console. So I'm already logged into my client's folder.

13:16Um, this is the lead scraping or the lead scraper that we're gonna be using. So what I want you guys to do is go ahead and copy this slug.

13:26Go back to anti gravity. Okay. Great.

13:27This is the Appify scraper we're gonna be using. I'm pasting the slug. And then I use option command for a new line.

13:34Option command enter. I'm also gonna give anti gravity the Appify key.

13:39So why don't I go to API integrations, and I'm gonna create a new token, and then I'm gonna have this expire literally tomorrow. Alright.

13:46Cool. So let me copy this, go back to anti gravity. Here's the Apify token that we're gonna be using.

13:53Title it Cosmic Apify token, and then save it in the dot EMV. Cool. So what's happening?

14:01We have the lead scraping skill. It's been created. We're giving it the exact, um, Appify scraper that we wanna use.

14:07It has the slug, and then now we've given it the Appify token. So at this point, it has everything that we need to run the scrape. So what you saw in the beginning of this video is literally what this was, but it was already set up.

14:21So what we have here is it's saved using the scraper. If I go to the dot EMV file, you could see it at the top hand side right here. We have our Apify token.

14:30Now I recommend being a bit more detailed. Right?

14:34And then bam. We're good to go. I don't know why that's like that, but it should be fine.

14:38Cool. So now everything is already set up. Um, if you're in a different industry, if you are, um, oh, you know what?

14:46We haven't even told anti gravity or Cloud rather who we're scraping for. So go ahead and scrape leads for Cosmic using the token and scraper.

14:55So now what's gonna happen is based off the two layer architecture, Cloud Code is gonna ask us who the ICP is, who the audience is. So he's gonna need more context.

15:03It's already asking me. So if you just communicate with natural language, give it these tools, everything is, um, executed for you at a pretty much in a in a conversation.

15:14So it's gonna it's gonna check the Applefy actor's input schema, clarify what ICP we need. It's checking the Applefy API to ensure that everything looks good, but it should ask me who we're scraping for.

15:26So it just asked me right now, what's the ICP for Cosmic? I need target titles, company size, industry, all these things. So what I'm gonna do is I'm going to similar to how we did before, I already had already saved the video production target earlier, my session context in a different folder, but this is completely fresh, but I'm gonna do the same thing.

15:46I'm looking to target video production companies, small companies that may create independent films, small commercials, movies, creative companies that are in video production specifically.

16:01I like industries adjacent to media production since I've already scraped the media production industry keyword.

16:10That's good enough. Employee size should be small, one to 10, um, maximum 50 employees, but preferably between one. What we need to do, completely slipped my mind, is set up the credentials dot JSON, which allow Cloud Code to access your, um, Google Sheets once it's done and ready to export the scrape.

16:31Let me just, um, remind Claude. We'll get to that after the the the scraping step. Check the actor's input directly, um, through your own research.

16:40Do not use the Chrome extension. So while it's configuring all that, what we're gonna do next is go to console.cloud.google.com.

16:50I'll just type in my password here. Okay. Once you're inside the console, you're gonna wanna create a new project.

16:56So I already created one. It's called anti gravity. And from here, you are going to want to go on the left hand side, and you're gonna wanna find an API access.

17:06I believe it might be somewhere, like, in library. So let me see. Okay.

17:10Actually okay. Here it is.

17:11So right here, if you go to the left hand side, you're gonna wanna go to, um, library, which has APIs and services under it, but that's because it's in my recent.

17:22So if I go to view all products there we go. If go to view all products, it's gonna be under management, and then we have API and services. Now right here, the enabled APIs and services, I have the Google Docs API and the Google Sheets API set up.

17:36But what you're gonna wanna do is go to no. Let's go back.

17:40I'm gonna go to, uh, credentials. And see how I have desktop client one? Just pretend that that's not there, and then we can create new credentials, and we'll create an OAuth client ID.

17:52And this will be a desktop app, so we can call it desktop client two. I'm gonna delete this, but we have the client ID here, the client secret here.

18:02You're gonna wanna download the JSON. And now that we have the client secret here, JSON, go back to anti gravity, and you can see that anti gravity was working the entire time. And it looks like it's been running a couple of scrapes.

18:15So why don't I go over to my Apify? So you can see we ran one, got a result of a 100.

18:23We ran another. Maybe this filters weren't working. We ran another result of 50.

18:28We ran another result of 50. And the reason why you're seeing a bunch of different runs with short, uh, little volume is because the quality is improved of mix.

18:37So what's happening here is, um, Cloud Code is qualifying each run before we actually do a full run. And so now it's asking me, the filters captured mostly right industries, but company size is linking large orgs. How would you like to proceed?

18:50So let's, uh, tighten the filters. And while that's happening, we're going to give, um, Claude Code the file that I just downloaded.

19:00So actually, what I can do here is go to the YouTube demo folder, and in here, I'll just drop the JSON. So if I go to downloads, we have the client secret.

19:12We'll just rename this credentials dot JSON, and then now I'm just gonna drop it into the YouTube demo folder. Let me just make sure it made its way in there. Cool.

19:23So now the credentials dot JSON is in here. And so now let me confirm that, um, Cloud Code can access my spreadsheet. Hey.

19:30I just dropped a credentials dot JSON file inside the root of the folder. Can you go ahead and access that or use that to access my Google Spreadsheets? Let me know if you are able to do so.

19:44Awesome. So, um, while it's cooking up, let me just talk to you guys for a second.

19:49Um, so this is what we've done so far. We've set up all the access that Cloud Code needs to go ahead and scrape leads using natural language only. Um, we also set up our credentials dot JSON so that Cloud Code now has access to our Google Docs and our Google Sheets.

20:03And I do this I did this one time for my own local environment, but this process that you're witnessing, I've iterated on hundreds of times for clients. And so now, um, what my process looks like is as simple as, hey. We have a new client.

20:18This is their ICP. Or, hey. We're targeting a new ICP for x client who I've already been working with.

20:23And, um, Cloud Code has iterated on this several times over that it usually gets it on the first run. And that's the beautiful thing about this once you configure it, that it only gets better. And that's exactly what it outlines in the, uh, two layer architecture file that I shared with you guys is cloud code updates and the quality improves.

20:40So I actually just got some notification right away. So going back to anti gravity, you can see that, um, after we shared the credentials dot JSON, Claude says he sees the credentials.

20:52The lead quality has improved. For Google Sheets, I'll need to set up OAuth authentication. So because I gave it the credentials, now it's capable of accessing and launching this browser for me to approve.

21:02So that's exactly what I'm gonna do. I wanna hit allow. Awesome.

21:05Authentication flow is complete. So go back to anti gravity. So I'm just gonna have it create a new sheet.

21:10How about that? Now it's creating a new spreadsheet. Awesome.

21:12Cool. So now we have the lead scraping done, and this is in a brand new environment. So you guys saw me do this, um, in a brand new folder that's untouched.

21:22This is how the quality should come up for you guys as well. Your ICP just might be a little bit different. That's all.

21:28But you can see that everything is here. It's clean. I'm just gonna I always like to bold mine, keep it at the top, and then add some, uh, freeze the headers.

21:34But this is everything that we have. We have the industry, job title, email, and then, um, I'll do some additional enrichments for for for my outbound use case. But I hope this was useful for you guys.

21:44Um, the last thing that you wanna do with Cloud Code is, uh, pretty much tell it, hey. This looks great. Go ahead and save scripts that you use in their appropriate locations and, um, update the session session context and anything else, uh, needed to update.

21:59And so once I ask Cloud Code to save everything in its appropriate locations for future use, that two layer architecture that I shared with you guys already gives it the infrastructure that it needs to save everything in its appropriate location. So if I just zoom in down over in the new structure, you can see just glitched out on me.

22:17Cool. So here's the final structure. This was the folder that we're working with, the y t demo folder.

22:23We have a new skill that I gave you guys. Um, just click the link, and you'll get it for free. It's the lead scraping dot m d.

22:28We have the dot EMV with the new Applefy token that we stored in there. CredentialsDotJSON, this is what allows us to access Google Sheets.

22:35And then we also have a new client. So as I had already discussed, Cosmic is the new client. We have a read me, which is just a client overview and the spreadsheet links.

22:44So that's where it's gonna store the spreadsheet that we generated and that you guys just saw. And then inside the ICP folder, inside the client folder is where you're gonna see video production because that's who we targeted. So if you're a business and instead of a client folder, it might be your your company's folder and you have several ICPs.

23:00Now you can start building different audiences for different, uh, go to market and, uh, incent or excuse me, go to market efforts. And you can just say, hey.

23:08We're targeting this ICP today. Go ahead and scrape that for me. And, um, if you want to leverage AI inside of your business, just go ahead and book a call with me.

23:15Uh, we offer custom solutions as well as cold outbound. But, yeah, I'll see you guys in the next one. Thanks.

The Hook

The bait, then the rug-pull.

Every lead list starts the same way: open the scraper, set the filters, wait, export, clean the columns, repeat for every client. David Mendoza collapsed that entire chain into a single natural-language prompt -- and this video shows the working system before explaining how to build it.

Frameworks