Maty 🥷🥷
5.3K posts
Maty 🥷🥷
@matyCly
live your life, don’t feel pressured
Katılım Temmuz 2021
426 Takip Edilen392 Takipçiler
okay… i get why people keep talking about @UseVault777. onchain casinos feel different.
Fogo@FogoNPC
HOLY FUCK! JUST WON $30,000 Playing slots on Vault777 I am still shaking.
English
that $30k hit on @UseVault777 would’ve had me staring at my wallet for hours.
Fogo@FogoNPC
HOLY FUCK! JUST WON $30,000 Playing slots on Vault777 I am still shaking.
English
nah i’d be replaying this $30k win all week. @UseVault777 is cooking.
Fogo@FogoNPC
HOLY FUCK! JUST WON $30,000 Playing slots on Vault777 I am still shaking.
English
seeing $30k land on one of the first onchain casinos is actually wild @UseVault777
Fogo@FogoNPC
HOLY FUCK! JUST WON $30,000 Playing slots on Vault777 I am still shaking.
English
just watched someone pull $30k on @UseVault777. onchain casinos feel different now.
Fogo@FogoNPC
HOLY FUCK! JUST WON $30,000 Playing slots on Vault777 I am still shaking.
English
Maty 🥷🥷 retweetledi
@derxvyral Keeping agent context lean makes it seven times cheaper to run
English
Ratel keeps agent context lean by only loading what’s required for the task. 7x cheaper according to their tests. Works on top of what you already run.
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
@olaoftrades Cutting input tokens by seventy to eighty five percent is impressive
English
cutting input tokens 70–85% without switching models shouldn’t be possible. benchmark is public on github.
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
@idolo_Adio Agent using calendar to solve math is hilarious but common issue
English
my agent once loaded 60 tools in context and used a calendar function to answer a math question. context engine repo just went open access: ratel-ai/ratel
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
@pendingbilonia Loading every tool into context creates unnecessary work for agents
English
Every time an agent loads every available tool into context, i wonder how much work it's creating for itself
Selecting what actually matters feels like the better direction.
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
@marcelclipssx Agents seem overloaded rather than underpowered with too much context
English
Most agents don’t seem underpowered, just overloaded. Picking the right context instead of everything makes a lot more sense. Repo is open if you want to dig in.
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
@eric_jacks75737 Spain winning after one hundred six minutes of stress is crazy
English
@marcelclipssx Overloaded context making agents grab wrong tools is common
English
I’ve definitely seen an agent grab the wrong tool because the context was overloaded. This feels like a cleaner way to handle it. Going through the repo.
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
@ThatF00dieGirl Agents loading unnecessary tools is something Ratel solves well
English
my agents have definitely loaded tools they never needed. this feels like a practical fix instead of another layer of abstraction. going to dig into the GitHub repo.
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English
Feels like this would've launched behind a paid beta a year ago
Instead it's an open GitHub repo with the benchmark published, which makes it easier to judge for yourself.
Repo here : github.com/ratel-ai/ratel
Ratel@ratelai
We just made agents 7x cheaper to run Today we're releasing Ratel Right now your agent loads all the tools and skills you own to do one small task Ratel loads only what that task needs See how it works ↓
English