Luv

researching AI memory and context engine

[Luv @flexaboiL]

I will build at least one new product every day and in public to understand what the fuck does it take to ship something people actually use.

[Luv @flexaboiL]

last element of the project select (github repo clone destination in this case) is clipped in t3 code nightly windows

@theo @t3dotcodes

[Luv @flexaboiL]

Trying something “bot”

[Luv @flexaboiL]

day 7 / turning this into a log instead.
I have realised things look easier on the surface but it takes failure to realise how difficult it was. Travelling.

[Luv @flexaboiL]

T3code split terminal horizontal resize when @theo ?

[Luv @flexaboiL]

jokes aside I am still working on something very interesting, and maybe (or maybe not) I will post about it very soon.

I also want to update on the everyday a "new" product. It was way ambitious than what I had imagined and I quickly got nerdsniped by a very interesting issue that I noticed while working on these "products".
I will be taking my time off of this everyday ship stuff but will occasionally drop here and there.

[Luv @flexaboiL]

day 6 / your agent is way too nice to you
introducing the retort skill. try it on a long session where it made several mistakes, or was even just a tad bit annoying.

[Luv @flexaboiL]

github: https://t.co/kXPXuHaFZz

[Luv @flexaboiL]

day 6 / your agent is way too nice to you
introducing the retort skill. try it on a long session where it made several mistakes, or was even just a tad bit annoying.

[Luv @flexaboiL]

@trashh_dev having dejavu

[Luv @flexaboiL]

@RhysSullivan imagine*

[Luv @flexaboiL]

No Max Payne title? ngmi

[Ben Dicken @BenjDicken]

Scale of 1-10 how hireable am I?

[Luv @flexaboiL]

@julianweisser @DhravyaShah accept it (just saying because I sent you a connection request myself lol)

[Luv @flexaboiL]

I don’t think google is losing the AI battle. I mean not in every sense. They are like the research labs helping everybody advance in the consumer side of the AI.

[Luv @flexaboiL]

Day 5 / still working on what I wanted to ship.
I have yet to conclude the research I am doing and it is taking me longer than usual. Probably good or bad idk. I am trying my best.

[Luv @flexaboiL]

@MoonGotchi There’s something very unsettling about that donut’s texture. It feels like a snake’s skin.

[Luv @flexaboiL]

@sethrosen um RAG?

[Luv @flexaboiL]

@omarsar0 drop the list. also how does your model rank the papers, like on what basis?

[Luv @flexaboiL]

@de1lymoon well most of these products are benchmaxxed. Not their failt entirely. It’s the base standard, like leetcode but for agentic memories

[Luv @flexaboiL]

people around people who actually ship have a real advantage. i unfortunately don't.

[Luv @flexaboiL]

@a16z fire 🔥

[Luv @flexaboiL]

@itsGrizai @yetone well, this is exactly what people building memory infrastructure are doing. I think an eval that benchmarks across multiple coding agents should exists for anything promising cross context. nice observation btw.