Had fun writing this with @kzh — what we learned around prompting GPT, and product design details for AI features, for our launch today of AI-powered headline rewriting
https://t.co/j0MMhE554N
“Near you and Similar to you” to be the seed readers of the post. This is a super effective way for “making friends” since it offers a efficient way to formulate the community dynamically ( no fixed community like subreddit).
An interesting analysis by an A16Z partner.
Basically making friends requires repeated exposure, ppl getting repeated exposure requires share something in common, and “finding one shares something in common with another” requires a place/vibe ppl are willing to show their opions
There are 5+ dating apps that each make more than $100M in annual revenue.
And yet, there hasn't been a single friend-finding app to reach massive scale 🤔
Why products with a stated goal of "making friends" will (probably!) never work ⬇️
That’s why Tinder for friends all fails in the end…bc they use a inefficient data structure.
The Red is interesting here since it uses some strategy to distribute post. Aka immediately after posting, the Red will GeoHash and LookAlike algorhims to recall users
For a social app, how to go viral? The Douyin(TikTok China Version) had less than 100 users in the first month. Several vital events happened to grow Douyin users from 100 to 100M. It's helpful to learn how this happened given Douyin is just a 10 ppl team with a minimal budget.
if you want to work on some new consumer product, one thing to consider is whether the content of the product is more attractive than short videos in any aspect for a considerable population(more than 10%).
Feedback from the leader of computer science group of CICC. I am open to coffee chat and virtual meetings if you are interested in this field. Feel free to DM me.