#!process Letta interview for Ai memory

1 messages · Page 1 of 1 (latest)

turbid ivy
#

orz

frail mica
#

no way

vagrant wren
#

Yes way

frail mica
#

i think letta has a really cool story and is a cool idea but when i tried to use the product it was horribly broken

#

few months ago, over the summer

#

but better than supermemory

vagrant wren
#

That’s why they need me

frail mica
#

trueeee

#

fix it

vagrant wren
#

My latency is 3000x better than theirs

#

For memory retrieval

#

Btw

frail mica
#

i’ll be no 1 customer

#

do you seriously have your own system?

#

oh wow nice

#

that’s sick

vagrant wren
frail mica
#

do you do the memory rearranging stuff while the user isn’t querying

#

i thought that was super neat

vagrant wren
#

My system is completely different crime all theirs idk what that is ngl

#

I didn’t know ai memory was a thing

frail mica
#

oh lol

vagrant wren
#

I was making my own ideas right

frail mica
#

o are u trolling

#

oh

vagrant wren
#

Then realized UCB made memGPT

#

Which is now Letta

frail mica
#

ah gotcha

#

yeah it’s a cool story

vagrant wren
#

Same abstractions even were the same as mine

#

All because

#

My OS exam asked what’s the next phase in IS

#

OS*

#

I really started thinking

frail mica
#

hahah

vagrant wren
#

But tbh I’m pivoting away from ai memory

#

Hive minds are cooler

#

Still need a real job thoprayge

frail mica
#

wouldn’t that still need memory

#

or is it just not memory as a product

vagrant wren
#

Lightweight layer yes

#

We have it

frail mica
#

sick well gl

vagrant wren
#

Check us out on LinkedIn

#

We’re ab to pop off

frail mica
#

hmmmm

#

i’ll look

vagrant wren
#

This is because they’re too research orientated

#

They just research

#

They don’t execute

#

That’s why their system is the slowest

#

They def got the brightest minds tho

#

They’re back by Google deep mind tho so

frail mica
#

yeah the team seems obviously cracked

vagrant wren
#

They’ll do good regardless I feel

#

My college is selling though

frail mica
#

idk how good the system /product people r tho

#

lol wdym

vagrant wren
#

UT Austin had the chance to make their own ai OS with me

#

They let Berkeley get me instead

frail mica
#

hahaha i see

vagrant wren
#

Absolute salesmen

frail mica
#

ay top talent does what it wants

vagrant wren
#

My professor for OS is a genius

#

I still think she’ll help

#

Prob over summer

frail mica
#

oh that’s cool

vagrant wren
#

We got a good bind

#

Bond

#

Search up Alison Norman

#

She’s the goat

frail mica
#

wait so r u planning on working on it still?

vagrant wren
#

Yes

frail mica
#

or work ft

vagrant wren
#

Mycelic is a hive mind distributed system

#

Letta is an AI OS

#

I’ll try to aim for a partnership

frail mica
#

ohh i see

#

that’s sick

vagrant wren
#

Because my software competes with monitors

#

Kognitos

#

They suck ass

#

Edge native distributed knowledge system

#

Beats Kognitos

frail mica
#

every class be using OSTEP ong

vagrant wren
#

Ima delete it

frail mica
#

oh i never heard of them that’s a cool angle

vagrant wren
#

They do knowledge systems for enterprise

frail mica
#

why would you want to build on letta if latency is ads tho

#

ass

vagrant wren
#

Because

#

My hive mind needs a Agentic state layer

#

Right now it’s only capable of sharing info

#

Using NATS

#

I’m in touch with their xeo

#

CEO

frail mica
#

would it really benefit from the letta memory rearrangement tho?

vagrant wren
#

Idk what that is but

frail mica
#

i mean hey sounds really cool either way

vagrant wren
#

It will extremely benefit

#

From Letta

#

And they will see my vision

#

I’ll inprove their memory layer

#

They’ll improve my Agentic state layer

#

Aka deep memory layer

frail mica
#

this is what i’m referring to

#

ay i’m following on linkedin so excited to see where it goes

vagrant wren
#

Lemme read

#

Yeah we alr do that

#

In a different way tho

#

Very interesting approach