Twitxy Saved

Ajay PV

Verified account This account is verified. Learn more

@ajaypv4

Co-founder, @sloosh_ai @shoposai
Previously SVP @0xPolygonLabs, Co-Founder @scapicxr (Acq. @flipkart), Head of @Flipkart Labs, Founding Engg @gridraster, @cisco
Bengaluru, India shopos.ai Joined January 2010
747 Following 1.7K Followers 2.1K Posts

Latest posts 5

Ajay PV Verified account @ajaypv4
Been testing HTML-in-Canvas and ended up making a website that knits your designs.

It’s called Thread :)

Upload an image, type something or draw your own design. Move things around, pick a colour and watch it get knitted, stitch by stitch.

You can pull the fabric, add wind and export the result as a video too.

Try it here: thread.vercel.app
0:20
Ajay PV Verified account @ajaypv4
What better way to validate a product than this! We've been on a mission to enable 'Single person billion dollar brand'

- One MCP that makes your business creative
- One plugin that makes you more sales
- One subscription that gives you the tools, models and skills to become AI pilled

Try @sloosh_ai & supercharge your brand 🚀
Ajay PV Verified account @ajaypv4
This is scary!! someone reverse-engineered TikTok’s app and says it has more than 25 AI models running directly on our phones.

Here’s what they found:
• TikTok can collect a message you typed even if you never sent it.
• One model uses your phone’s sensors to work out whether you’re walking, lying down or sitting in a car.
• The app sends information including your battery level, whether your phone is charging, screen brightness, whether it’s muted and more than 50 other variables.
• Videos come with scores predicting whether you’ll skip, finish, like, open the comments or follow the creator.
• Another model predicts whether you’re going to skip the remaining videos in your feed. If it thinks you will, it requests what’s called an ‘emergency batch’ of new videos.
• Keep skipping and the app can decide you’ve lost focus, then insert a livestream.
• One model tries to tell when you’re falling asleep, and another predicts when you’ll open the app again so it can start getting ready in the background.

And it gets weirder. The extracted models also estimate a person’s age, gender, mood and attractiveness. In the developer’s tests, the attractiveness model scored darker-skinned and older faces more negatively.

The amount of work that goes into keeping someone scrolling is kind of mind-blowing. It’s not hard to see why people find it so addictive.