simran sachdeva

???

i love the concept of borrowed personalities, when you think about it everyone has a borrowed personality. from the books you read, the movies you watch, the people you meet, the places you travel to etc. makes you the person you are

@amritwt wtf

your fav egirl on the Internet is just five .md files on a random dude’s laptop

@davidliu8473 😭🤚 ikr

@simranrambles lowk this is me with claude. you ship a working feature before imposter syndrome can even spin up 💀

vibe coding is so so addictive because the feedback loop is sm faster than your fear response

I’ve always used replit remotion for motion graphics but never tried building anything with it before. actually the most fun 24 hours ! Thanks guys @MannyBernabe @Franciscocrz

building Seen for the @Replit Buildathon :) where you match with your potential partner based on context from your chatgpt/claude histories I’ve never had so much fun sitting at a stretch and building something really excited for this join the waitlist: seen-match.replit.app

it scratches the same part of my brain that video games do

that’s actually a really good point, though the goal isn’t to find your clone, it’s to find someone cognitively compatible. compatible is doing a lot of heavy lifting here lol. similarity, resonance, balance. some traits should align, others should complement each other. more “fits your mind”. someone can think like you yet bring a whole different energy to your life about being with polar opposites, I’ve not really thought of it yet. but i could probably use some input here. what do people optimise for when they date someone who’s completely different from them ?

@simranrambles @Kannav02 @Replit What about “opposites” so what if I like certain activities or live a certain way but would prefer a partner who didn’t share my way of life for example, less about symmetry and more about completing each other if that makes sense.

as of now the matching is less ‘people who both like X’ and more mapping how people think across many signals simultaneously. something like curiosity patterns / emotional cadence / thinking styles etc etc. and yeah raw chats never persist long-term, sensitive info gets filtered/redacted before the analysis, and only derived embeddings/traits go through. the goal is matching on cognition, not storing people’s diaries :p

Cool stuff, 2-3 questions though -> the grouping/similarity score, is it divided into buckets/categories, where they are then compared within those buckets, or is it global similarity matching via the subject graph that you create based on chats? -> is there a privacy classifier layer to avoid including sensitive info as part of the analysis, and also was wondering if the final analysed data from chat context is stored somewhere or if its one-off data really interesting idea

yeah but that’s kinda the point, dating profiles are currently optimised for perception. the way people talk to ChatGPT/Claude is way more unfiltered, so that matching would be optimised for honesty and the “real” deal. And realistically, if you’ve already shared that much context with AI, the ‘private’ line already got blurry a while ago.

@simranrambles @Replit do you think people will share their gpt/claude chat histories/context? personally, i talk about alot of personal things and would not want to share any of that with a third-party app.

replit remotion (it’s free for 24 hours rn btw, do what you will with this information)

24 hours without my laptop and using my mac mini full-time feels weirdly prehistoric like why am I tied to one physical location

just frontendmaxxed and uxmogged

traces on neatlogs are now cleaner and richer ✈️

actively

considering switching to codex

rewatching money heist and drooling over professor as i write this tweet

@simplydt 🤞

@simranrambles nice move, stay tight on prep and nail the launch

going to do my first product hunt launch in a couple weeks. should be fun for everyone involved #excited