THANI retweeted
THIS GUY IS A F*CKING GENIUS... he turned a gross cartoon gamer with 30 days to glow up into 116K followers with just 4 videos 🤯
Every episode is one day of the countdown:
→ Day 1 got 5.9M views.
→ Day 2 got 1.1M views.
→ Day 3 got 2.9M views.
→ Day 4 got 878K views.
The whole show runs on two numbers at the top of the screen: days left and glow-up percentage.
In this episode, he finally takes a shower, texts the prom queen he’s trying to impress at the reunion, and cuts toenails that click like claws.
Glow-up: 6%.
The comments write the show.
Under one episode, the top comment said he should take a shower and got 138K likes.
The next episode opens with him in the shower.
So people come back to see if their comment made it in.
One goal, a clear ending in 30 days, and a meter that moves every episode.
You always know how far he has to go.
And look at the bio: open to ads, and a $9.99 playbook on how to make an animated series with Claude.
The recipe is almost too simple:
→ Claude Code writes 30 episodes around one goal and one deadline.
→ Nano Banana creates one main character and keeps him the same every day.
→ Every episode ends with a question, and the best comment becomes the next episode.
If just 0.1% of the 5.9M people who watched day 1 bought that $9.99 playbook, that’s 5,900 sales.
This breakdown shows the full pipeline in Claude Code and how to turn a series like this into $15K a month 👇
THANI retweeted
HOLY F*CK, motion studio in 100% free github skills for you guys.
20 open-source skills for Opus 5.5.
Animation. Shaders. Characters. Audio. Export.
01 Remotion → React video engine.
github.com/remotion-dev/remo…
02 Anime.js → Timelines, easing and choreography.
github.com/juliangarnier/ani…
03 Scene.js → CSS keyframes and animated sequences.
github.com/daybrush/scenejs
04 Konva → Layered 2D graphics and animation.
github.com/konvajs/konva
05 Pencil2D → Hand-drawn animation.
github.com/pencil2d/pencil
06 Zdog → Playful pseudo-3D illustrations.
github.com/metafizzy/zdog
07 Three.js → 3D scenes, lighting and cameras.
github.com/mrdoob/three.js
08 Filament → Physically based rendering.
github.com/google/filament
09 ozz-animation → Skeletal character animation.
github.com/guillaumeblanc/oz…
10 Planck.js → 2D physics and collisions.
github.com/piqnt/planck.js
11 regl → Custom WebGL visuals.
github.com/regl-project/regl
12 WebGL Noise → Procedural noise for shaders.
github.com/ashima/webgl-nois…
13 meshoptimizer → Lighter models, faster rendering.
github.com/zeux/meshoptimize…
14 MeloTTS → Multilingual voiceovers.
github.com/myshell-ai/MeloTT…
15 RNNoise → Neural voice cleanup.
github.com/xiph/rnnoise
16 Rubber Band → Audio time-stretching and pitch shifting.
github.com/breakfastquay/rub…
17 OBS Studio → Screen capture and recording.
github.com/obsproject/obs-st…
18 OpenRV → Shot review and playback.
github.com/AcademySoftwareFo…
19 Mediabunny → Video processing in the browser.
github.com/Vanilagy/mediabun…
20 FFmpeg → Encoding, compression and delivery.
github.com/FFmpeg/FFmpeg
Brief → layout → animate → polish → render → publish.
Bookmark this. Go build something worth watching ↓
THIS IS F**KING INSANE!
Same Claude script. Three times the payout.
A faceless YouTube channel, run by one person with Claude, aiming at $35K a month. The plan starts before any video exists: pick your niche by RPM, not by passion.
What YouTube pays per 1,000 views (upper range):
Finance: ~$30
Documentaries: ~$25
AI: ~$10
Views needed to hit $35K/month (his estimates, not guaranteed):
Finance: ~1.2M
AI: ~3.5M
The script quality is identical. Only the niche changes, and it triples the money.
What he says rarely works: a generic script, a robotic AI voice, and stock footage.
What he does instead:
Claude writes as a documentary scriptwriter: a hook in the first seconds, a new open question in every section, about 1,500 words
ElevenLabs for voice
Midjourney or Flux for images, Runway for video clips
CapCut for editing
His weekly loop: research → script → voice and visuals → edit → thumbnail and upload → check analytics
Which niche would you pick: finance, AI, or documentaries?
THANI retweeted
THE $478 A DAY CLAUDE + SHORTS SETUP COMES DOWN TO ONE PROMPT RECIPE, AND YOU CAN COPY THE STRUCTURE
Not a magic button. A prompt with 5 parts, and the clip shows every one.
First, the setup takes a minute. In Claude, open Customize, go to Connectors, add a custom connector, name it OpenArt and paste the link from openart.ai/mcp. Do it once and you’re done.
Now the prompt. Here’s how the one in the clip is built:
the style: a high energy cinematic Disney-Pixar animated Shorts adventure
the hero and the world: adorable toddler firefighters saving a busy fantasy city
the opening shot: sunrise inside a colorful toy-like fire station
the pacing: a new scene every 2 seconds to keep people watching
the camera: a fire alarm close-up and fast tracking shots beside the fire truck
Paste that into Claude and it sends the job to OpenArt, which renders the clip. Now you have a Short you can post and monetize.
What can it pay? Google’s overview says 1 million views generally earns between $1,000 and $6,000, as low as $50 to $200, and $15,000 or more for the right long videos.
Swap in your own heroes and your own city, keep the 5 parts, and you can make a new Short in minutes.
the people who learn the recipe once can run a hundred versions while everyone else rewrites a blank prompt every time.
bookmark this and follow, I’m sharing my full prompt template this week.
THANI retweeted
Holy f**k, Opus 5.5 replaced After Effects
It's f**king insane…
Drop the PDF below into Opus 5.5 and see what it makes
Then read the full guide on motion design in the article below ↓
THANI retweeted
Stop writing claude code prompts from scratch.
Anthropic has released a free prompt collection with 52 ready-to-use examples for different coding tasks.
Most developers don't even know this resource exists.
Here's what you'll find inside:
→ Prompts for planning, building and debugging.
→ Code reviews and security checks.
→ Automation and workflow improvements.
→ Copy, customize and use them for your own projects.
→ Each prompt explains the reasoning behind it, so you can learn to create better prompts yourself.
This isn't just another collection of random prompts. It's an official resource from Anthropic.
Save this. You'll definitely need it.
THANI retweeted
You're spending way too much on AI when powerful alternatives are available for FREE.
Qwen3 might change how you look at AI models.
Here's what makes it stand out:
→ Outperforms Claude 4 on certain benchmarks
→ Handles up to 256K tokens of context
→ Works completely offline on supported hardware
→ Offers both reasoning and fast-response modes
→ Available without paying for API access
Stop assuming the best AI tools have to come with expensive subscriptions.
Save this before your next AI bill.
THANI retweeted
will drop a 5x cook here
and maybe some giveaways
so get in while it’s still open!
The best crypto plays never start on the timeline. They start in private rooms.
Bagworkers is one of them.
Doors are open for a short time only. Then it's invite-only.
join here: t.me/+05eBp6Dluw4wNjA0
Let’s cook.
You're spending way too much on AI when powerful alternatives are available for FREE.
Qwen3 might change how you look at AI models.
Here's what makes it stand out:
→ Outperforms Claude 4 on certain benchmarks
→ Handles up to 256K tokens of context
→ Works completely offline on supported hardware
→ Offers both reasoning and fast-response modes
→ Available without paying for API access
Stop assuming the best AI tools have to come with expensive subscriptions.
Save this before your next AI bill.
THANI retweeted
this is f*cking wild, dubbing one video used to cost $1,200 per language. now it's free
MrBeast (the most subscribed creator on YouTube) posted a line that hit 21 million views:
"the next billion views won't come from better videos. they'll come from the same video in 20 languages"
he pays studios to get there. you don't have to. 9 GitHub repos already translate, dub and lip-sync in your own voice:
01 → pyVideoTrans (19.6k stars)
translates any video into another language and adds the dub and subtitles in one run
github.com/jianchang512/pyvi…
02 → GPT-SoVITS (64.7k stars)
trains a model of your voice from 1 minute of audio
github.com/RVC-Boss/GPT-SoVI…
03 → OpenVoice (38.1k stars)
clones your voice in seconds and makes it speak languages you never learned
github.com/myshell-ai/OpenVo…
04 → VoxCPM (41k stars)
reads any script in 30 languages, or builds a brand-new voice from one sentence of description
github.com/OpenBMB/VoxCPM
05 → LatentSync (6.2k stars)
moves your lips to match the new language, so the dub doesn't look dubbed
github.com/bytedance/LatentS…
06 → VideoCaptioner (16.4k stars)
writes, splits, fixes and translates subtitles with an LLM, no manual timing
github.com/WEIFENG2333/Video…
07 → Seamless (11.9k stars)
Meta's model that turns speech straight into speech in another language and keeps your tone
github.com/facebookresearch/…
08 → ebook2audiobook (20.5k stars)
turns any ebook into an audiobook in your cloned voice, in 1,100+ languages
github.com/DrewThomasson/ebo…
09 → translate-book (2.2k stars)
lets Claude Code split a whole book across parallel agents and translate it while you sleep
github.com/deusyu/translate-…
20 languages at $1,200 each is a $24,000 dub for one video. here it's one folder and a night of rendering
full setup explained, step by step ↓
THANI retweeted
Holy f*ck, Photoshop just lost to Opus 5.5
7 free open-source repos that cover every AI tool in a $22.99/month Photoshop plan
Select. Remove. Sky swap. Generative Fill. Lens Blur. Expand. Upscale. Harmonize.
Every edit in the video is real. All 7 repos ran on one photo
All of it is free:
rembg → Select Subject, cuts anything out of an image
github.com/danielgatis/rembg
LaMa → Remove Tool, fills in whatever you erase
github.com/advimman/lama
SegFormer → detects the sky so you can replace it
github.com/NVlabs/SegFormer
diffusers → Generative Fill and Expand on SDXL, no credits
github.com/huggingface/diffu…
Depth Anything V2 → real depth maps for Lens Blur
github.com/DepthAnything/Dep…
Real-ESRGAN → Super Resolution, 4x the pixels
github.com/xinntao/Real-ESRG…
color-matcher → Harmonize, matches any layer to the scene light
github.com/hahnec/color-matc…
Save this before your next retouch
Drop the PDF below into Opus 5.5 and watch what it builds
Then read the full motion design guide in the article below ↓
THANI retweeted
How does one YouTube channel generate over $60,000 every month?
This creator breaks down the entire process, step by step, showing how to build it using tools that are completely free.
No expensive courses. No paid tools. Just a practical walkthrough.
🔖 Save this, set aside a few hours, and study the process. You might discover your next big opportunity.
THANI retweeted
THIS IS F**KING AMAZING.
Someone is using Claude + AI Visual Studio to build faceless YouTube channels with simple stick-figure animations.
And the numbers on display are worth looking at.
The channel, Zenn, reportedly pulled:
• $62K–$63K in estimated revenue
• 8M+ views in 90 days
• 137K+ new subscribers
Here’s the workflow:
Study successful channels and reverse-engineer their content.
Use Claude to generate video ideas and scripts.
Turn those scripts into simple animated scenes using AI Visual Studio.
Batch-generate visuals and thumbnails.
Assemble and publish consistently.
No complicated animation skills or filming yourself.
But here’s something worth pointing out: the analytics show estimated revenue, not verified earnings. The $100K/month claim and under-five-minute production claim aren’t established by what the video shows.
Still, the workflow is interesting.
The opportunity is in building a repeatable content system instead of creating every video from scratch.
Save this for later.
THIS IS F**KING GOLD!
someone just open sourced a billing system built specifically for AI agents.
it's Apache 2.0. it's free. you can run it on your own infrastructure.
it lets you:
→ track every agent action as a usage event
→ meter tokens, tool calls, tasks, and compute
→ turn usage into invoices automatically
→ set credit limits so agents can't overspend
it handles up to 1M events per second.
traditional SaaS charges by seats or subscriptions.
AI agents don't work that way.
one run might cost $0.02. another might cost $4.
OpenMeter lets you bill based on the actual work the agent does.
bookmark this for later.
THANI retweeted
THIS GAVE ME CHILLS... 10.1M views on a dreamlike flower field, and you can hang that exact world on your wall for anywhere from $35 to $1,000 🤯
→ One reel has 10.1M views.
→ One has 6.6M views.
→ One has 5.7M views.
754K followers here. 620K more on TikTok.
In this video, you rush through a flower field toward glowing people standing under a planet that fills half the sky.
11 seconds. No words.
You don’t watch it. You feel it.
And that’s exactly what he sells.
Not the video, the feeling, frozen on your wall.
His website is a price ladder:
→ Prints from $35 to $65.
→ Framed from $80 to $160.
→ Metal for $250.
→ Full collections for $1,000.
The reel is the free sample.
The print is the souvenir.
And there’s more on the site: separate pages for concerts and games.
Fans buy the art, and artists hire the artist.
Here’s how I’d build my own:
→ Claude Code writes 30 worlds that feel like a dream you can’t remember.
→ Nano Banana creates them all in one glowing palette, in high resolution for print.
→ Kling or Seedance adds one fast, low camera move toward something you can’t reach.
If just 0.1% of one 10.1M reel bought the cheapest print, that’s 10,100 orders.
This breakdown shows the full pipeline in Claude Code and how to turn a page like this into $15K a month
THANI retweeted
ONE GUY JUST CLONED MICROSOFT OFFICE WITH AI
Word, Excel, and PowerPoint completely free and open source.
→ WordCraft → full Word replacement that opens real documents
→ GridCraft → Excel clone supporting 513 of 523 functions (~98%)
→ DeckCraft → PowerPoint alternative with working slides
Built by one developer using AI in a matter of days.
He dropped the Adobe clones first, then released these one week later.
Microsoft spent decades building Office.
He rebuilt it with AI and gave it away.
Comment “free” for the links.
Bookmark this before Microsoft notices.
THANI retweeted
This guy just LEAKED the EASIEST method to make money in 2026
Claude Opus 5.5 + YouTube = $1K/day
Nothing complicated (just 12 minutes).
Ready for you to replicate from home.
Bookmark this for later….
AI agents can now use an actual computer, not just write code 😳
Meet Cua, an open-source toolkit that gives AI agents desktops they can operate across real apps and operating systems.
With Cua, agents can:
→ Click, type, and interact with desktop apps
→ Work across macOS, Windows, and Linux
→ Use isolated computer environments for safer automation
→ Connect through MCP, SDKs, and CLI
→ Test computer-use agents with benchmarks
→ Run tasks across code, APIs, and graphical interfaces
The interesting part?
Your AI agent can switch between writing code, calling APIs, and interacting with a GUI when a task requires it.
That opens the door to automating workflows that ordinary scripts struggle with especially when an app has no useful API.
Cua has already attracted 28K+ GitHub stars 👀
Repo: github.com/trycua/cua
The next step for AI agents isn't just generating answers. It's being able to use the tools and software we use every day.
