I turned Claude into a 24/7 research assassin that scrapes Reddit,...
J.B.@VibeMarketer_
5 views
Jun 28, 2025
~2 min read
1
I turned Claude into a 24/7 research assassin that scrapes Reddit, YouTube, and TikTok on command.
now it feeds me viral angles, customer pain points, and winning copy… while I sleep.
here’s the unfair setup I use 👇
now it feeds me viral angles, customer pain points, and winning copy… while I sleep.
here’s the unfair setup I use 👇
2
apify just dropped a Multi-Call Protocol (MCP) API.
it lets you stream structured data from any Apify actor directly into Claude - in real time.
that means:
– no more scraping scripts
– no more copy-pasting comments
– no more guessing what your customers care about
just scrape, feed to Claude and get hooks, angles, and offers instantly.
prompt-native research is here
it lets you stream structured data from any Apify actor directly into Claude - in real time.
that means:
– no more scraping scripts
– no more copy-pasting comments
– no more guessing what your customers care about
just scrape, feed to Claude and get hooks, angles, and offers instantly.
prompt-native research is here
3
I hooked Claude up to Apify’s MCP using a custom function call that grabs results from actors like:
– youTube Comment Scraper
– tikTok Trend Scraper
– google SERP
– reddit Threads
– amazon Reviews
(and 100s more)
now I just ask, and Claude does the digging
– youTube Comment Scraper
– tikTok Trend Scraper
– google SERP
– reddit Threads
– amazon Reviews
(and 100s more)
now I just ask, and Claude does the digging
4
here’s what this looks like in practice:
I just tell Claude:
“scrape the latest YouTube comments from Andrew Huberman. summarize the most common objections, pain points, and emotional triggers. then give me 3 ad angles based on viewer psychology.”
claude hits the MCP pipeline, pulls real customer data, and turns it into ad-ready insights in seconds
I just tell Claude:
“scrape the latest YouTube comments from Andrew Huberman. summarize the most common objections, pain points, and emotional triggers. then give me 3 ad angles based on viewer psychology.”
claude hits the MCP pipeline, pulls real customer data, and turns it into ad-ready insights in seconds
5
I see it as building a full-stack research assistant that runs 24/7.
you can literally scrape real conversations from Reddit, YouTube, TikTok…
feed them straight into Claude…
and turns it into marketing insight you can actually use:
– Customer pain points
– Viral hooks
– Objections and desires
– Swipeable copy and angles
you can literally scrape real conversations from Reddit, YouTube, TikTok…
feed them straight into Claude…
and turns it into marketing insight you can actually use:
– Customer pain points
– Viral hooks
– Objections and desires
– Swipeable copy and angles
6
the best part?
you don’t need dev skills.
just call the MCP actor endpoint from Claude with the right input config.
you don’t need dev skills.
just call the MCP actor endpoint from Claude with the right input config.
7
what this unlocks:
– deep market research in minutes
– competitor analysis on autopilot
– consumer psychology pulled from the comments section
– copywriting inputs based on real pain, not guesses
– trend detection across platforms
– deep market research in minutes
– competitor analysis on autopilot
– consumer psychology pulled from the comments section
– copywriting inputs based on real pain, not guesses
– trend detection across platforms
8
this is imo the only way forward to write banger ad scripts, scroll-stopping hooks and viral angles.
If you’re still writing copy without scraping real comments, reviews, and posts first (or worse… still doing it manually)…
ngmi.
If you’re still writing copy without scraping real comments, reviews, and posts first (or worse… still doing it manually)…
ngmi.