Pinned Tweet
I have created 10-video playlist to help you learn how to integrate Supabase with Bubble.
Check it out: youtube.com/playlist?list=PL…
Perfect for the beginner.. Enjoy.
If my agent works all day while I sleep all day, does that count as taking a vacation or working 9–5?
People using Railway are acting like that manager we all hate, the one who fires people after just one bad quarter.
The way these companies are giving away credits - it should be called "vibe creddittiing"..
Just send my way too
The new open-source scam: a “Free Agent API” that’s just a router to paid APIs under the hood. Bring your own APIs and use them.
Bro, if I have to pay anyway, why not just plug those APIs directly into my agents? I’m not even writing the code myself. Why would I use your GitHub library?
All this just to get a few stars and impress HR. Kuch bhi chal raha hai! 😂
In the finance ecosystem, building a transaction auto-categorization app is like building a social media scheduler. Everyone has built one.
Interesting challenge! Let me try this too. Let Claude + Hyperframes cook! 🔥
by Dec 31st, I'll hit 1M+ views on a new YouTube channel that my AI agent runs for me
calling it my last challenge of 2026 💪
a few days ago, one of my explainer videos did 600K impressions on X, so I started a YouTube channel that runs completely on its own
it’s called "How It Actually Works"
every day, it picks a topic, writes the script, makes the video and posts it
so far, it has covered interesting topics like how Shazam recognizes a song in seconds, what really happens when you delete a file, and how a contactless card works without a battery
and this is my calendar - every green check is a video that went out by itself, and the blue ones are already queued up
the whole setup is Revid making the videos and Squad running the agent that picks topics, writes and schedules them via PostSyncer
this is going to be fun 🤩
xitter.cf/tibo_maker/status/2105…
Things I learned while building a research agent. Maybe these can help someone:
1. Every search must name the subject.
2. More data doesn’t mean a better report.
3. Test APIs/MCPs before adopting them.
4. Use each platform’s own API. If unavailable, use Apify.
5. Verify every sentence against its evidence.
6. Design the report after analyzing the data, not before.
If you’ve built a research agent, share what you’ve learned too!
Ran an UltraCode effort on Opus 5.5, It run for 55 minutes and used 4.3M tokens.
It exhausted 10% of the 5-hour limit and 3% of the weekly limit.
I think its pretty good, what you think?
The a16z team perfectly summarized what I’m trying to do with NocodeTalks:
Helping finance teams become finance engineers.
Anyone interested in advertising on clipwipe.com/? - gets an average of around 200 views per day.
$50 for next 15 days. Just testing the stuff, not sure how it will turn out.
Comment or DM me.
The Hyperframes team did a pretty good job with their desktop app. It’s so easy to generate videos with it.







