Dextro

Building @VarticasAI - A multi-agent that plan+previews action across Gmail, linear , Jira and many more . not a chat bot . Not Zapier

[Dextro @loading_X__]

@sriram3720 @X Sure let's connect

[Dextro @loading_X__]

Hey @x

Looking to #connect with what are you building?

💻 SaaS
⚙️ Tech
🤖 Automation
🧠 AI tools
📦 Product Development
🌐 Web apps

Drop what you're working on 👇🏼
Let's #buildinpublic

[Dextro @loading_X__]

@nicklaunches @X Nice one

[Dextro @loading_X__]

@HelloVyom https://t.co/WydUSAbtfD
I won't say much about it just try it it's free

[Dextro @loading_X__]

@LeadFreshAI @X Ohh nice

[Dextro @loading_X__]

Hey all @X

I'm looking to #connect with
people interested in: -

Frontend
Backend
Fullstack
AI/ML
DevOps
Web3
Cloud
Open Source
Tech Writing
Markets
Finance
Crypto
Let's grow together 🤝

[Dextro @loading_X__]

@KenziChenBuilds @X Sure let's connect

[Dextro @loading_X__]

Hey Founders 👋

what do u do when Claude hits limits?

For me it's scrolling X

[Dextro @loading_X__]

@TehnomasterS @X Really love it that you go through my account. Best of luck on your building

[Dextro @loading_X__]

Hey @x

Looking to #connect with what are you building?

💻 SaaS
⚙️ Tech
🤖 Automation
🧠 AI tools
📦 Product Development
🌐 Web apps

Drop what you're working on 👇🏼
Let's #buildinpublic

[Dextro @loading_X__]

@DanielSmidstrup Switch to anti gravity or codex or opencode

[Dextro @loading_X__]

@Itz_HarshXD @X Sure

[Dextro @loading_X__]

@manpo_wang @X Sure let's connect

[Dextro @loading_X__]

@Amani_Thunk @X Sure

[Dextro @loading_X__]

Now let's do the math on what that actually current systems use...
Current system for 6GB model:
OS kernel: ~500MB
Runtime overhead: ~500MB
Memory addressing/pointers: ~1GB
Padding/alignment waste: ~500MB
Cache inefficiency: ~1GB
Actual weights: ~2.5GB

That's 58% waste.

[Dextro @loading_X__]

What does a system actually need to run LLM inference?

Just these things:
-Store the weights — the numbers
-Read them in order — sequentially
-Multiply them — basic math
-Output a result — one token at a time

That's it. That's all an LLM inference engine fundamentally does.

[Dextro @loading_X__]

What does a system actually need to run LLM inference?

Just these things:
-Store the weights — the numbers
-Read them in order — sequentially
-Multiply them — basic math
-Output a result — one token at a time

That's it. That's all an LLM inference engine fundamentally does.

[Dextro @loading_X__]

@ll_marx @X You are right

[Dextro @loading_X__]

@AaronBennat @X Sure i will try

[Dextro @loading_X__]

@RoboRentcc @X What does it do,?

[Dextro @loading_X__]

@LeadFreshAI @X How does it work?

[Dextro @loading_X__]

@marceguerra @X +1

[Dextro @loading_X__]

@sans_builds @X Sure let's connect, i already followed