Modern Creator
Why Modern Creator?
Creator Magic · YouTube

Claude Code Built a Web Scraper to Replace a $249/Month SaaS

A Claude Code session replaces a $249-a-month scraping subscription, but the fix that actually works is a tactic change, not better code.

Posted
1 weeks ago
Duration
Format
Demo
educational
Views
85.5K
896 likes
Big Idea

The argument in one line.

Claude Code can build and redeploy a working scraper in minutes, but the real unlock that turns a 33% success rate into over 90% is swapping data-center IPs for pay-as-you-go residential proxies.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…
  • You pay for a scraping or data API service and wonder if a few Claude Code sessions could replace it for pennies.
  • You run AI agents that need live web data (pricing, reviews, sentiment) with no clean API available.
  • You're curious what residential proxies actually do and why data-center IPs get blocked.
SKIP IF…
  • You need enterprise-scale scraping with guaranteed uptime and a support contract, not a one-person project.
  • You're not comfortable reading alongside an AI coding session as it writes and deploys code.
TL;DR

The full version, fast.

A creator builds a custom web scraper, Drone, with Claude Code instead of paying $249 a month for a scraping API. The first version runs from a VPS data-center IP and gets blocked about two-thirds of the time, since sites like ChatGPT and Perplexity specifically flag hosting-provider IP ranges. Swapping to pay-as-you-go residential proxies, at roughly $1 per gigabyte, pushes the same scraper's success rate past 90% with no other code changes. Total spend across proxies, a free search API tier, and cheap AI summarization lands around $2.

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 – 00:40

01 · Cold open

A fast before/after demo of the scraper, then the $249/month problem is named.

00:40 – 02:01

02 · ScrapingBee math and Peter Levels' hack

ScrapingBee's $249 plan works out to about $2 per 1,000 pages; indie hacker Peter Levels built his own scraper for $1 a month instead.

02:02 – 02:52

03 · Drone's ground rules

The scraper being built obeys robots.txt, never logs in, never solves CAPTCHAs, and only sees what a normal visitor sees.

02:52 – 04:29

04 · Claude Code builds the dashboard and fetch engine

A written brief lets Claude Code build and deploy the dashboard and fetch engine to a VPS from short prompts.

04:29 – 06:32

05 · First test: 33% success from a data-center IP

Fetching live from the VPS's data-center IP gets blocked on most major AI tool sites and loses regional pricing entirely.

06:32 – 07:29

06 · Claude Code diagnoses the failure

Given a screenshot of the failed results, Claude Code explains that bot-protected sites specifically flag hosting-provider IP ranges.

07:29 – 09:43

07 · The fix: residential proxies

Data Impulse's residential proxy network, over 90 million IPs across 195 countries at about $1/GB pay-as-you-go, gets wired into Claude Code.

09:43 – 11:14

08 · Second test: success rate jumps

The same scraper, now routed through residential proxies, climbs past 60% success on the first run and costs under ten cents.

11:14 – 14:14

09 · The /goal improvement loop

Repeated prompts and a /goal instruction push Claude Code to keep tuning the scraper toward higher success rates.

14:14 – 16:32

10 · Search and sentiment via Brave and Claude Haiku

A Brave Search API feed plus Claude Haiku summarization adds sentiment scoring, surfacing that smaller tools beat bigger names on sentiment.

16:32 – 18:43

11 · The total bill

Proxy usage, a free search API tier, and AI summarization together land around $2, and the finished scraper can now serve data to other agents.

Atomic Insights

Lines worth screenshotting.

  • A scraping API's generous-looking credit allotment can still cost about $2 per 1,000 pages once JavaScript rendering and premium proxies are factored in.
  • An indie hacker ignored AI's claim that a task couldn't be done and built a working scraper in five days for about $1 a month instead of $249.
  • Sites with heavy bot protection specifically flag IP ranges that belong to hosting companies, not just suspicious request patterns.
  • Hosting a scraper's server in one country hides pricing data from every other country, since many sites serve different prices by region.
  • A scraper that reads and obeys robots.txt, skips logins, and never solves CAPTCHAs stays within what a normal visitor is allowed to see.
  • Swapping only the IP source, with zero code changes, took the same scraper from a 33% success rate to over 60% on the next run.
  • Residential proxies route requests through real home internet connections in a chosen country, which is what makes a scraper look like an actual visitor.
  • A full run of thousands of scraping requests through residential proxies can cost under a dollar in proxy usage.
  • Because Google's robots.txt disallows scraping, routing search through an official API instead is the compliant workaround.
  • Cheap AI summarization of search results can surface that smaller, less hyped tools score better on sentiment than some bigger-name competitors.
Takeaway

Residential Proxies Fixed What Code Couldn't

SCRAPING ECONOMICS

The code Claude Code wrote worked the whole time; what turned a 33% success rate into over 90% was swapping a data-center IP for pay-as-you-go residential proxies.

02ScrapingBee math and Peter Levels' hack
  • A scraping API's generous-looking credit allotment can still cost about $2 per 1,000 pages once JavaScript rendering and premium proxies are factored in.
  • An indie hacker ignored AI's claim that a task couldn't be done and built a working scraper in five days for about $1 a month instead of $249.
03Drone's ground rules
  • A scraper that reads and obeys robots.txt, skips logins, and never solves CAPTCHAs stays within what a normal visitor is allowed to see.
  • Treating a robots.txt 'no' as a hard stop is both the ethical move and the one that avoids legal exposure.
04Claude Code builds the dashboard and fetch engine
  • A single written brief lets a coding agent build a working dashboard and deploy it from one short prompt, turning infrastructure work into a single session.
05First test: 33% success from a data-center IP
  • A scraper hosted on a data-center IP got blocked roughly two-thirds of the time, even on sites generally open about being scraped.
  • Hosting a server in one country also hides pricing data from every other country, since many sites serve different prices by region.
06Claude Code diagnoses the failure
  • Sites with heavy bot protection specifically flag IP ranges that belong to hosting companies and block them outright, separate from any other bot signal.
  • An AI coding assistant can read a screenshot of failed results and explain the pattern in plain language before being asked to fix anything.
07The fix: residential proxies
  • Residential proxies route requests through real home internet connections in a chosen country, which is what makes a scraper look like an actual visitor instead of a server.
  • A pay-as-you-go residential proxy plan priced around $1 per gigabyte, with no monthly minimum and traffic that never expires, undercuts a $249-a-month subscription by a wide margin.
08Second test: success rate jumps
  • Swapping only the IP source, with no other code changes, pushed the same scraper's success rate from 33% to over 60% on the very next run.
  • The first full run through residential proxies cost under ten cents in proxy usage.
09The /goal improvement loop
  • Repeating a prompt like 'improve the success rate,' or handing a coding agent an explicit goal, drives iterative gains run over run without re-explaining the task each time.
  • Thousands of scraping requests run through residential proxies can still add up to roughly a dollar in total proxy spend.
10Search and sentiment via Brave and Claude Haiku
  • Because Google's robots.txt disallows scraping, routing search through an official API is the compliant substitute, and it can be free at modest usage.
  • Adding a cheap model to summarize search results into a sentiment score can surface that smaller, less hyped tools score better than some bigger-name competitors.
11The total bill
  • The full build, across proxy usage, a free search API tier, and AI summarization, landed at roughly $2 total plus the cost of the server itself.
  • Owning the finished pipeline means the resulting data can be exposed as an API and handed straight to other AI agents that need it.
Glossary

Terms worth knowing.

Residential proxy
A proxy that routes a request through a real home internet connection rather than a data center, so the request looks like it's coming from an ordinary visitor.
Data-center IP
An IP address that belongs to a hosting or cloud provider's network. Many sites specifically detect and block these because they're associated with bots and scrapers.
robots.txt
A text file a website publishes to tell automated crawlers which pages they're allowed or disallowed to request.
Scraping API
A paid third-party service that handles proxy rotation, CAPTCHA solving, and JavaScript rendering for web scraping, billed by credits or requests.
Success rate
The percentage of scraping requests that return usable data instead of being blocked, timing out, or getting an error.
Claude Code /goal
A prompting pattern where a goal is handed to Claude Code so it keeps iterating a task across multiple runs until the goal is met, rather than being re-prompted manually each time.
Resources

Things they pointed at.

00:55toolScrapingBee
01:51linkCan I Vibe Code It
07:45toolData Impulse
13:04toolBrave Search API
14:22toolClaude Haiku
Quotables

Lines you could clip.

06:11
“This is the big headline. From a data center IP, we had a 33% success rate.”
crisp, numbers-first problem statement→ TikTok hook↗ Tweet quote
12:24
“We've basically paid eight cents to get the data into our site instead of signing up for an expensive plan to do pretty much the same thing.”
concrete cost comparison→ IG reel cold open↗ Tweet quote
16:15
“We've literally gone from 33% success with a data center IP to over 90% using these residential IPs.”
the payoff stat, clean before/after framing→ 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
This is Drone, my own web scraper. 100 AI tools, one button, watch this. From a normal server, red, red, red, red.
Now the same button through residential proxies, green. And even in some cases, prices from different countries. peter levels paid for a scraping service then ai told him that he couldn't build his own well you know what he's an indie hacker he went ahead and built it anyway about one dollar a month to do everything he needed but he wouldn't say which proxies he used and in this video i'm gonna show you everything scraping b can do much of this they claim to be the best scraping api to avoid getting blocked which is good in the age of ai But their business plan comes in at $249 a month.
So let's replace it. Now I hear your question coming straight at me. Why do I even need a web scraper?
Well, my agents need data from the web, things like prices, what changed, what people actually think about things. And most of those things have absolutely no API. So people go ahead and pay a scraping service.
Scraping Bee is one of the big ones at $249 a month, giving you 3 million API credits. Now that sounds like loads, doesn't it? But one page with a premium proxy and JavaScript costs 25 credits.
So that's about $2 per 1 ,000 pages. That can get expensive fast. Peter Levels hit this too.
He runs a site called Hotelist. He tried to build his own scraper. And he said AI kept telling him that it couldn't be done.
Well, five days later, he's got his own scraper. About 90 % success as well. And now he's built something that runs for $1 a month instead of $249 a month.
What's the secret ingredient? Residential IPs, which he wouldn't name. There's also a site called Can I Vibe Code It?
Now it rates over 1 ,000 apps and asks, can you rebuild it with AI? Yes, kinda. Or not really.
And not really means the value is the data. So today, I'm turning a not really into a yes in terms of getting data from the web. Now, one thing before I get into the build, drone that I'm building here is a polite scraper.
It reads every site's robots .txt file. If a site says, no, don't scrape me, then drone says, okay, and flies away. It goes slowly, one page at a time per site.
And there's no logins, no capture solving, public pages only. We're not going inside to any logins and just what any visitor would actually see. Now let's build it.
I've got Claude Code open and I've got an agents .md file ready with a few particulars, the whole plan, the rules, everything, so I can just talk to it and get this coded. So let's do the first prompt. Read agents .md, then build step one, my drone dashboard with 100 AI tools inside a table, please.
Okay, there is my prompt ready to go. Let's hit enter and let Claude Code spin into action. Right, there it goes.
It's starting to spin up everything that's needed for drone. And away it goes, writing the typescript. So very soon we'll have a site.
And there we go. Step one is built and committed. I can even access it on my local computer at localhost colon 3000.
Okay, look at this. Pretty cool. ChatGPT, Claude Perplexity, pretty much everything you'd expect in terms of AI tools listed in a nice top list with categorization.
This is exactly the kind of thing I'm looking for. It's running pretty well here on my local computer, but I don't want to grab all this data from my local computer and get my own home IP band. So I'm going to host this on a VPS and show you how the tool works there.
So let's go ahead and build the engine. I'll say to... Build step two, the drone engine wire up the fetch button so it pulls every tool straight from the server with no proxy.
And also I'd like you to make sure it's deployed on my VPS that I'm hosting right now so I can test it there and not on my home computer. Okay, there's my second prompt. We'll let Claude code work away on this.
Okay, Claude is cooking away, checking out what it needs and whether the VPS I've just set up is reachable. Okay, this is fantastic. So it's going to go onto my VPS that I'm hosting in the cloud.
But as we've heard before, data center IPs are not great at extracting data. It's going onto a server hosted in Germany. Okay, this is great sign.
We're finishing up here so I can log into my drone site here and everything is intact and ready. All right, we can see Claude is finishing up saying that everything's working, plain fetches via a headless Chromium render. This is what we want to get real data from the web.
And now the moment of truth. Fetch straight from the VPS that I'm hosting on a data center IP. Let's click the button and watch this go.
Okay, some are already blocked due to robots saying don't do it. Look at this, red crosses. Well, we got Claude.
We got Microsoft Copilot. Yes, we got Cursor. We got GitHub.
We didn't get Pi. We didn't get Character AI. Lovable refused to be grabbed.
Look at all of these red crosses everywhere, including pricing. It's struggling to grab pricing from places. But still, most of my data is really looking very, very patchy.
Let's watch this all the way down. We're actually nearly halfway. A lot of sites are very happy in the age of AI to have their data extracted, but we can see tons of them up here, especially the bigger ones, are not very happy to have their data extracted.
What we're getting here is the inability to get pricing, okay? So some sites are actually pulling in pricing, and I can hover over, and when I hover over a tick, we get an HTTP. 200 that means the price was pulled into my database for my drone site but prices from all around the world like i'd like to know what chat gpt costs in the us in the uk in germany india and other countries because of course some websites actually charge you differently depending on where you're based and we'll get into that in a moment But what I want you to notice here is because my VPS is based in Germany, some sites are allowing with green ticks the pricing information to be extracted.
But really, if we look at this chart, it is a flurry of red crosses. Nothing really hung or got held up. Everything is finished.
71 .9 megabytes of data pulled down. But look at this. This is the big headline.
From a data center IP, we had a 33 % success rate. It's just not good enough. And yes, if we look at the columns, of course, every other pricing column is red.
Of course, we can't see what users are paying in other countries because my server lives in Germany. It can't see what someone in India pays. But let's actually ask Claude Code what just happened.
I'm going to take a screenshot of, say, the top 15 sites here. And let's go ahead and drag it into Claude Code. And we'll just put a very simple prompt here.
This is the result I got running this from my server on the VPS in Germany. As you can see, it's not that great. Summarizing just a couple of lines what you think happened here.
Don't try and fix anything. Just give me a brief summary. Okay, so I've posted the image and that question to Claude Code.
Let's see what comes back. Where it failed. It's mostly the data center IP.
Sites that sit behind. behind heavy bot protection like ChatGPT, Perplexity, Pi, Grok, and all of these treat IP ranges from hosting companies like bots and refuse them. And every other country, US, UK, India, Brazil, Japan, cell is red because your server is in Germany.
And those grey robots things there, the cells with robots in, that's where sites actually asked not to be fetched and drone obeyed. But I'm getting blocked around 67 % of the time. And here's the thing, as you can see, real people don't browse from data centers.
So lots of sites already know that and treat data center traffic like a bot, as Claude Coates just said there. A server can only ever be in one country, but now this is where the solution comes in. Data Impulse have kindly agreed to support me with this build and they do do all kinds of proxies including those needed residential proxies.
And what that means is all of your data extraction goes through real home internet connections in any country that you choose. They've got over 90 million IPs and if we look at locations here we can actually scroll down and see all locations over here. We've got top residential proxy locations listed here with millions of IP and, well, pretty much every country in the world is listed here.
So you can be certain if you want to see data as it's shown to residents of a certain country, you can do it with something like Data Impulse. But how much does it all cost? I hear you screaming.
We don't want to pay $249 a month like Peter Levels did for a scraping service. You can pay as little as $1 a gigabyte. That's really pretty good.
And there's absolutely no monthly plan. It's pay as you go and your traffic never expires. The important thing to see here is it's supported in 195 countries and there's 90 million plus ethical residential IPs.
And this part is important. They tell me that their IPs come from people who actually opt in and even get paid for it. Right, so I'm going to log into my account.
Here is my account and you'll see I've signed up for a residential proxy plan right here. I'll grab the proxy access details here, which for obvious reasons I've blurred out. Then I'll click docs down here to open up all their documentation, which by the way, Claude Code can read.
There's no blocking on data impulse docs, so it can set up all of those residential proxies for me. Okay, I've popped all the credentials into a file that Claude Code can read, so now it's time for the clever bit. I'm going to wire in Data Impulse residential proxies, and also, this is really clever, I can actually get Claude Code to wire in so that it can view sites from the US, the UK, Germany, India, Brazil, Japan, all on one VPS server.
So let's do step number six. Build step six. I've added my data impulse login to a file that you can read, and then I'd like you to re -upload it to my VPS, restart drone, and then make it ready for me to click the button to fetch data via data impulse.
Okay, it's as simple as that. We'll let Claude Code cook this in. It's just finalizing the data impulse run job so that this all works for me, and soon we'll be able to refresh here on my VPS server and test via residential proxies.
Boom, look at this. You can see Claw Code is wiring in those different country names. It's enabling the data impulse button and the proxy check panel.
And finally, it's committing and then copying everything to the VPS so we can test it. Okay, looks like we're good here. Green, Claude Code has deployed.
So remember that 33 % success rate. Let's refresh my site. And look, now we've got this lovely big green button, fetch via data impulse.
I'm really curious how this is going to work. So let's maximize this screen, click the fetch button and see what happens. Right, as we can see, it's kicking into action.
And wow, it's a lot faster. We are getting tons of data coming back. It's really interesting to see the time elapse, the data transferring, the success rate that's all.
Already going above the 33 % that we suffered with earlier. We can see immediately some sites block by robots .txt. But we can also see that we're grabbing prices from around the world.
And it's interesting to see that Gemini does price differently in different locations. Right here we're getting a 64 % success rate. now the interesting thing here and i think this is a real stark comparison to actually paying for a service to do this using the residential proxies on data impulse how much does it cost and look at this we've been charged less than 10 cents to do our very first run look at this residential proxy plan one has pulled in all of these different hosts these are indeed different ai tools and the amount of requests made and the amount of traffic that went through the proxies this is incredible stuff we've basically paid eight cents to get the data into our site instead of signing up for an expensive plan to do pretty much the same thing that we're doing here it seems sas is officially dead if you can vibe code it with a service like data impulse but let's just ask claude code what can you do to improve the success rate and look at this it's rolling again it's going immediately over all the sites we're seeing a lot of green stuff come in now this is amazing just one prompt to claude code improve the success rate
And as you can see, constantly asking Claude to improve the success rate means that it goes around in a loop. You can even put forward slash goal here until Claude Code works with Data Impulse using their docs and their terms and conditions to get you as close to 100 % success as possible. And I love this way of prompting AI.
And now this is obviously taking me time, but I'm basically replacing a $249 a month SaaS with pay -as -you -go data. I absolutely love this Claude slash goal. Every run, the success is getting better and better.
And this is in thanks in no small part to the residential proxies that I purchased. Right now we're getting a good success rate. Let's go ahead with Claude code.
Build step seven, search and sentiment for every tool. And I'd like to use the Brave API. Okay, there is my prompt ready to go.
I'll just go ahead and access the search API from Brave. Copy the API key here and set Claude Code on this task. Prices are nice, of course, but what do people actually think?
That's data that's super valuable. Now, of course, you can pull this from Google, but Google's own robots file says, no, don't do that. And Drone respects that, so I'm using an official API like Brave instead.
Again, it's just a few dollars a month for really valuable information. Claude Code is writing its code and about to deploy this to my server, Sentiment Scorer, and it's run, ready to go. And by the way, while this is going, I wanted to just scroll through and show you all the results that have come in green so far via my data impulse residential proxies.
It's pretty incredible to see a page that was, well, completely red, now completely green. And the best thing is information like regional pricing, using those residential proxies from around the world where it's available. Not all sites do regional pricing.
We can see it right here in our table. look at all of these green boxes i mean this is just incredible over 90 success rate very few blockages now most of them coming in green and okay this is exactly what we wanted to see We've literally gone from 33 % success with a data center IP to over 90 % using these residential IPs.
We can see here Claude Code is building the sentiment tracker and we're going to get nice scores on the next run over here. And the best thing is I've also put in a very lightweight Claude Haiku API key so we can get those summaries and sentiment scores created by AI. Look at this.
Fantastic. Step seven is deployed and ready for me to press. So let's make this really big and let's refresh my drone site.
And yes, we have search and sentiment, which I can now add into every single site. So we're going through, grabbing the sentiment on each AI tool. We'll wait for this to happen.
Obviously, this will take longer because it's AI processing each tool. And look at this, we've got stuff coming in. We've got sentiment scores coming in right next to the data.
This is absolutely fantastic. So again, we're using those residential IPs to go out and get real search results from sites that live online. Plus, we're using the Brave API key to get some search that way.
And this is absolutely fantastic. We've got a sentiment score coming in on everything here. We're seeing that it boosts productivity, but is less accurate than some competing LLMs right there on GitHub Copilot.
This is fantastic with a link out to the tool. I've basically built an overpowered dashboard that grabs worldwide pricing, data on tools, and online sentiment in just a matter of minutes. So now if I look at the top 10 at least and the sentiment scores, which AI model should I be using?
And oh my goodness me, actually it seems Mistral, Lechat, and also DeepSeek have way better sentiment than the top AI tools. I would never have known that unless I'd run this experiment. On Mistral, it says speed, EU data residency, affordable pricing is good, but reasoning capabilities are behind the top ones.
DeepSeek, unmatched price to performance. but a few security vulnerabilities. This is ultimate sentiment that has come in in minutes, and I can immediately trade off which tool I want to use based on the amazing data that I've just pulled in.
Every tool gets a score and a one -line verdict right here. Now the bit though you've been waiting for after running all of this, what is the actual bill for everything? Now I've been hammering those residential proxies non -stop and look at this I'm struggling to go over a dollar for all of the searches I've done and I've probably run those searches a few thousand times since starting this video so Peter Levels one dollar a month to get all the data I need is really pretty accurate.
Then we can look at the billing for my search API from Brave. Well I've made 200 requests for my experiments and paid nothing for it because Brave gives me free usage. So it's way better than scraping Google when I can just use Brave for free.
It's worth also looking at my Claude account, a million tokens in, 55 ,000 or so just out, all via Haiku. And my estimated bill for Claude to do all of that sentiment analysis with AI, again, just over a dollar. So I've basically spent $2 to get an awesome dashboard that gives me everything that I need.
Every request, every site cost me... pennies or cents now to be fair i am paying for a server as well and my own time and you can get convenience if you buy a sas but ultimately we're asking can i vibe code it can i replace it and with tools like residential proxy ips from something like data impulse it's absolutely fantastic if you need them for your agents the link will be down below i'll definitely take drone i've built it now i can even make an api of that site to give to my tools like claude code and codex when they need information what they're doing i can point my agents at it and get the elite stuff that i need one command and claude code or codex has all the data it needs so that's the full picture one very cheap server drone built with claude code in an afternoon data impulse residential proxies for pages that a server just can't see and a search api and ai for the sentiment Data Impulse is linked down below.
You can start with just $5. So I'd really like to know what would you build with this stack that gets you 90 % plus success rate getting data from the web. AI needs it and you can get it.
Let me know in the comments down below. Thank you so much for watching. And YouTube is showing a video on your screen now.
You should watch next. Thanks.
The Hook

The bait, then the rug-pull.

A scraper built live with Claude Code fails two-thirds of the time from a normal cloud server, then clears 90% success on the very next run with nothing changed but where its requests came from.

Frameworks

Named ideas worth stealing.

02:02list

Drone's polite-scraper rules

  1. Reads and obeys every site's robots.txt
  2. One page at a time, one site at a time, no bursts
  3. No logins
  4. No CAPTCHA solving
  5. Public pages only

A short, explicit rule set the scraper follows so it only ever sees what an ordinary visitor sees and backs off whenever a site says not to scrape it.

Steal forany scraping or data-collection agent that needs a defensible, non-adversarial posture
CTA Breakdown

How they asked for the click.

VERBAL ASK
18:21link
“Data Impulse is linked down below. You can start with just $5.”

Sponsor plug placed after the result is already proven live on screen, not before, so the pitch lands on top of a demonstrated 90%+ success rate rather than a promise.

Storyboard

Visual structure at a glance.

open
hookopen00:00
33% success
promise33% success05:29
residential proxies
valueresidential proxies08:17
sentiment scores
valuesentiment scores15:13
CTA
ctaCTA18:35
Frame Gallery

Visual moments.

open00:00Frame at 00:15 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS00:15Frame at 00:33 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS00:33Frame at 00:49 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS00:49Frame at 01:03 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:03Frame at 01:17 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:17
open00:00Frame at 00:15 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS00:15Frame at 00:33 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS00:33Frame at 00:49 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS00:49Frame at 01:03 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:03Frame at 01:17 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:17Frame at 01:31 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:31Frame at 01:45 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:45Frame at 01:59 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS01:59Frame at 02:13 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS02:13Frame at 02:26 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS02:26Frame at 02:41 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS02:41Frame at 02:55 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS02:55Frame at 03:09 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS03:09Frame at 03:18 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS03:18Frame at 03:32 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS03:32Frame at 03:56 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS03:56Frame at 04:00 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS04:00Frame at 04:13 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS04:13Frame at 04:26 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS04:26Frame at 04:47 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS04:47Frame at 05:01 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS05:01Frame at 05:15 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS05:1533% success05:29Frame at 05:40 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS05:40Frame at 05:57 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS05:57Frame at 06:14 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS06:14Frame at 06:24 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS06:24Frame at 06:40 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS06:40Frame at 06:54 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS06:54Frame at 07:08 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS07:08Frame at 07:26 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS07:26Frame at 07:29 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS07:29Frame at 07:45 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS07:45Frame at 07:57 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS07:57residential proxies08:17Frame at 08:38 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS08:38Frame at 08:42 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS08:42Frame at 08:57 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS08:57Frame at 09:09 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS09:09Frame at 09:28 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS09:28Frame at 09:42 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS09:42Frame at 10:00 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS10:00Frame at 10:10 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS10:10Frame at 10:24 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS10:24Frame at 10:38 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS10:38Frame at 10:52 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS10:52Frame at 11:06 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS11:06Frame at 11:14 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS11:14Frame at 11:34 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS11:34Frame at 11:47 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS11:47Frame at 11:58 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS11:58Frame at 12:19 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS12:19Frame at 12:31 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS12:31Frame at 12:40 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS12:40Frame at 12:59 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS12:59Frame at 13:06 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS13:06Frame at 13:26 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS13:26Frame at 13:45 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS13:45Frame at 13:48 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS13:48Frame at 14:09 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS14:09Frame at 14:20 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS14:20Frame at 14:35 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS14:35Frame at 14:54 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS14:54Frame at 14:58 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS14:58sentiment scores15:13Frame at 15:33 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS15:33Frame at 15:42 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS15:42Frame at 16:01 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS16:01Frame at 16:12 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS16:12Frame at 16:26 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS16:26Frame at 16:41 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS16:41Frame at 16:52 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS16:52Frame at 17:06 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS17:06Frame at 17:25 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS17:25Frame at 17:39 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS17:39Frame at 17:53 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS17:53Frame at 18:07 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS18:07Frame at 18:21 from Claude Code Built a Web Scraper to Replace a $249/Month SaaS18:21CTA18:35
One-click upgrade to your Google

Get more breakdowns in your search results

Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.

Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
Watch next

More from this channel + related breakdowns.

13:14
Creator Magic · Demo

Jev Sorted 250 Emails for 1 Cent

A silent AI model that can only return typed decisions triages 250 real inbox emails for about a cent, while Claude Opus 5 takes 14 minutes and $1.39 to do the same job.

September 21st
24:53
ByteGrad · Tutorial

AI Web Scraping with Jev and Playwright

An AI coding agent explores a review site, writes its own Playwright scraper, and a fast classifier model sorts the results, no human touches a CSS selector.

September 28th