current chat interfaces suck, so i built a canvas for llms

LMFAOOOO

current chat interfaces are fundamentally broken: 1. context switching hell - multiple questions = constant scrolling. by the time your first query finishes, you have 3 unread responses below. information gets lost in the chaos

2. memory degradation - LLMs get worse as conversations grow. some queries need no memory, some need temporary context, some need everything. branching solves this

3. parallel queries - chatgpt forces you to wait for the current query to end before you can enter a new one. running queries in parallel saves massive time. why wait for one response when you can get multiple simultaneously?

4. generative ui potential - design is limited as well. apply parallel queries to design and you can iterate through multiple ui variations instantly. not just one design, but multiple high-quality options to choose the best

5. no collaboration tools - teams use ai heavily, when you're researching with someone, shared context is invaluable. but sharing conversations through the share button isn’t scalable, it’s honestly inexcusable. so i added figma collaboration as well

so i built - a canvas for LLMs with parallel execution, branching memory, and real-time collaboration. the future of AI isn't a chatbox. it's a canvas

p.s. i'm 18, previously engineered at 2 VC-backed startups, now looking for my next role. dm me if you're building something ambitious

also on producthunt!

Canvas makes a ton of sense for exploration, particularly creative writing I still use my little canvas project I made awhile ago:

I played with this concept as well, and my conclusion is that it’s not very user-friendly for non-devs.

I built that for images. Pretty sure i'm the only one that uses (not sure havn't added analytics), but I use it almost daily. Need to add nano banana tho

agree with the problem, but this as a solution doesn’t seem great either, after a while you end up in Canvas hell

how'd you make this beautiful UI!!

like this

@ahadj0 can i ask how do you make such videos that zooms and pans with smooth large cursor movement ?

@maxleedev he said he was using

@maxleedev tysm

@maxleedev yup!

This alone, makes this ❤️‍🔥❤️‍🔥❤️‍🔥

this is great!!!

thanks:)

looks so good 🔥 i wish the quoted/selected context stayed visible even after a new chat starts

yay!

w music choice

shamelessly copied @cadenbuild :)

yes!

nice one

Dude this actually looks cool!

it would be super nice if you can give the components a thin border with color based on the model being used. I can zoom out and quickly find the orange claude one !

bro it looks incredible, when are you launching this?

now

This is actually pretty good. @grok explain why brilliant ideas like this many times doesn’t go further and creates an interface / user experience revolution?

Just wanted to point out we can do this in Google AI studio with Gemini right now you can duplicate or fork chats. Very useful! (Sorry I know this is a chatgpt request! I totally agree!!)

the children yearn for loom

very cool i could see a lot of people using this

@sama do this and win the Ai-Race

fastest follow back ever

i really needed this it's so exhausting going back and forth. the more you keep imputing prompts, chatgpt loses its memory and you have to keep reminding, mehn should be a core feature from them, can't lie

That was quick.

this is really cool!

My adhd is going to love this

@praneethpike did it before !

damn, this is good ngl

Really practical and well thought out !

Bro that’s so needed and citations as well for the content generated

MAXLY!

max lee 🤯

Finally someones built an ADHD simulator

THIS is a feature I needed. This is how my brain works and Projects was the closest thing but this is needed.