Founder
@oneinterface
YC S24 | We solve hard problem of AI context. Our mission is to tackle AI consciousness. ready to solve the impossible? | built
@fixkeyai
People are just starting to realize the power of Local LLMs
Especially with new Apple chips. It's a game-changer
Let me show you:
Falcon is 180B LLM. It is the size of GPT3.
How fast would it run on your Mac?
Apple M3 Max: 3.5 tokens/sec
We can compare it to 2 x A100 (2 x…
So I set up BakLLaVA-1 in the llama.cpp, and now it can provide real-time descriptions of the live feed from my camera
check it out!
open source?
cc:
@nisten
@thursdai_pod
@willdepue
#llama
Screenshot & Ask questions about ANYTHING
GPT4 Vision came browser. It can:
- 📑 interpret any table capture
- 🫁 Help you learn visual subjects like anatomy
- 🚗 What is this car element? GPT knows the answer
- ❓pick yours
👇Demo & Beta👇
cc:
@gdb
RAG is most OVERRATED tech on market
Simple example for you why RAG is not here to stay you do real shit
You are lawyer.
You have big long doc with VC deal
ONE single word will change the way the deal is made, POST money or PRE money
I ask your stupid RAG system: hey, what is…
GPT for US immigration and citizenship (visa, green card)
You don't have to pay 10K$ for a lawyer to get an O-1 / H-1 visa!
GPT can answer me straight away using available info with LINKS! Check demo!
Immigration $lawyers$ are past?
cc:
@USCIS
I am now a founding Engineer at Tab.
I sacrificed a lot for it. I said no to every accelerator, investor and millions of worthless job offers from SF fancy startups.
Not because I am soft and can’t do them.
Because I am brutal and I see what is the best future opportunity for…
Openai peeps giving me the cold shoulder so leak time, been holding on this one for a while for a few reasons.
By end of 2025 there should be an openai model in the range of 1 - 10 Billion that is significantly better than GPT4.
Auto-take screenshots of your screen
Send to GPT4 vision at 9PM
Get a summary email with the stuff you have done today.
Automatically. Zero effort
Demo + Github link 🔗👇
If you look deeper,
@GroqInc
and 500 tokens / sec mixtral tech was founded by the same person who created TPU for
@GoogleAI
Tensor Processing Unit – the core thing google AI servers a most likely rely on
Whatever Jonathan Ross is about to do is about to change the AI chip…
one last demo i want to show you
Local LLM to take care of taking notes automatically for you
my chrome extension converts webpage into plain text
and indexes it with locally on running vectorDB
in the end you can ask questions to everything you saw with
@ollama
Open source?
The way people interact with the computer is going to change.
I am working on LLM controller for your Mac
Inspired by LLM OS (
@karpathy
)
But I think LLM first OS will come in 1-2 years.
We need intermediate step of Expanded /w LLM OS first. This is how we get there.
Search for…
Introducing Tilt, a $199 AI wearable that acts as your notes app, transforming your lived experiences into a personal AI to give you perfect recall
Cornell notes
Todo list
Flashcards
calendar events
Open source is in my private repo, I’ll share later
Delivered in summer 2025
Wall-e lives your laptop and remembers everything you see
Processes ideas for you. Creates notes library augmenting your thoughts with AI
- get a recommendation
- catch up
- have personalised learning
or just have fun talking to something which knows you
Who wants beta?
Apple has started showing/leaking some of their LLM developments in public. during last few weeks I saw this:
- WebGPU now available for testing in Safari Technology Preview
- "MLX" – array framework for Apple Silicon
- "LLM in a flash" paper to run efficiently on device
-…
Private LLM to fix grammar with
@OLLAMA
i turned off my internet, triggered fixkey and fixed grammar in a second using local LLM [mistral 7B]
Works perfectly on M1 & M2 with 16GB
Feature will be available next week to set up at
I found it very fascinating the whole story about how
@RewindAI
backed by
@a16z
gets suddenly hurt by small focused team so bad, that they release not ready product to the market just to capture it.
---
Original post to trigger to response of
@RewindAI
by
@AviSchiffmann
…
Concept of LLM Extended OS
I DON'T believe we will be able to jump to LLM first Operating Systems straight away.
Self-operating systems are just like ChatGPT. The need reinforcement learning with human feedback (RLHF). They need it to LEARN how to operate effectively.
How can…
@_vojto
here are top 7 words you need to mute on twitter
🧵👇
1. Here's the most
2. Here are the
3. Here are 10
4. Here's 8 things
5. Don't believe me?
And my personal favourite one:
6. Here are 7
Everyone claiming AI Act is bad, have you even read draft of it?
I did some research
First of all I agree with Nassim Taleb almost controversial thought, the more regulations you have, the more opportunity for small businesses arise, even though it seems the opposite.
So I am…
I made a tutorial for the auto-take screenshots and send to GPT4V for mac.
What it covers:
- lounchd
-automator
-python GPT4+ GPT4V
- bash
- IFTTT
github repo: to follow with me
Auto-take screenshots of your screen
Send to GPT4 vision at 9PM
Get a summary email with the stuff you have done today.
Automatically. Zero effort
Demo + Github link 🔗👇
You know how, like when you use
@obsdmd
, you have to link all the notes to each other
to eventually create a knowledge graph?
so
@yoheinakajima
has created Knowledge graph which auto expands when you throw data at it
check out what is has done to my federated learning notes!
Why is LLM finetuning SO underappreciated?
No one is doing it properly. It's such a big pain for everyone.
Just compare those two results
People deserve to have their personal fine-tuned models
I asked it Question about Apple swift documentation. And it answered it wrong:)…
who is working on personalised arxiv papers digest email?
i am ready to pay for such service
want to define my interests areas and receive email every 2-3 days about the best papers on topic
I finally got my hands to
@fixkeyai
and added
@GroqInc
support there
and damn its fast
it translated 5k characters (600 words) to german from english in < 2 seconds
thats crazy
comment if you want beta | free for everyone while groq API is free :)
@lexfridman
@elonmusk
Loved the interview so much that I have created Elon GPT inside chatGPT which was trained on 100+ YouTube interviews
Free and Available now
Introducing the 01 Developer Preview.
Order or build your own today:
The 01 Light is a portable voice interface that controls your home computer. It can see your screen, use your apps, and learn new skills.
This is only the beginning for 01— the…
So I set up BakLLaVA-1 in the llama.cpp, and now it can provide real-time descriptions of the live feed from my camera
check it out!
open source?
cc:
@nisten
@thursdai_pod
@willdepue
#llama
This is FUCKING hard tech
Our latency at fixkey is 400ms on avg
No local models can compete with us
good luck, losers
if we use
@GroqInc
it gets to as close as 300ms to fix the whole paragraph
Fixkey has just pushed improvement in latency
0.3 seconds per request 🤯
Please reply if you face any errors!
You don't even feel anything.... it's so blazingly fast
What if you connect emotions to colors?
and add a beautiful gradient on top?
You will get Feeling gradient
My Aurea[Berlin] hackathon project
> Voice to emotion to color <
Open sourced:
..hacked into data
~/Library/Application Support/com.memoryvault.MemoryVault
And there is SQLite, which search_content contains raw OCR results
Who is doing better RAG over it?
I was wondering what are the top green flags for VCs
Got a list
- Mattress on the floor
- single bedroom
- lowercase
- does blueprint protocol
- built startups before
- YC
- has strong community
- uses focus tool & blockers
- has big x profile
- won hackathons
what else?
My past work I am extremely proud of is
@fixkeyai
I really pushed core principles of building clean, native and aesthetic macOS app
as a nerds me&team went into details to make every part of it feel native
we pushed the limit
today we launch it on PH
chat with page was just publish in chrome extension store!
Its 100% free & open source
Plug in your OpenAI API key and play!
Chrome extension link:
Github link:
e/acc in Berlin ⏩ - more than 100 builders turned up at our AI Tinkerers event today, showcasing work-in-progress prototypes and ultra-polished products with millions of users. What a crowd!
See some impressions below 👇🏽
@Karmedge
with
1/n
This Next step to killing modern Hollywood & Disney.
I give it 1 year to start people working full time on animated movies and earning money from publishing it
Soon creators will solo publish animated series in less than a week
Creativity is about to explode
Introducing Pika 1.0, the idea-to-video platform that brings your creativity to life.
Create and edit your videos with AI.
Rolling out to new users on web and discord, starting today. Sign up at
Text to UI explanation
I found a cool connector from ChatGPT to Figma Jam.
Ask ChatGPT something very complex
copy the output to Figma Jam AI
it will generate nice visual charts to visually explain it for you
Meet GPT of
@elonmusk
I have collaborated with and created GPT which was trained on 100+ public YouTube interviews
Free and Available now:
@OpenAI
I want to comply with your policy, please contact me to take it down if required
🇩🇪/acc in Munich - one of the OG machine learning cities where LSTMs and Stable Diffusion were born - was on fire yesterday.
more than 100 builders showed up at our
@AITinkerers
event yesterday, with 12 folks showcasing projects ranging from no-code pipelines to fully finetuned…
Is there a tool to automatically create knowledge graphs based on my text i want to send?
happily will pay for an API to do it for me or quick to set up lib
goal: send it raw text throughout the day and make it embed it and connect the dots for me
MLX exploration experiments. Thread with screenshots of speed and my benchmarks
what I will explore
- mistral 7B instruct
- phi2
- finetuning
lets start.
100% GPU is based
I transcribed
@OpenAI
conference with whisper and asked gpt4 to summarize it. Here is what I got:
🚀 OpenAI introduces GPTs: customizable AI models with instructions, knowledge, and actions.
💡 Create your own GPTs to perform specific tasks and integrate with 6,000+…
Go from hand-drawn sketch to ready logo in 30 seconds using the new chatGPT
now DALLE can get your input image and generate anything based on it
- 🦹 avatars
- ✏️ logos
- ✂️ any kind of modifications
My today aictivity minute by minute
want to know yours?
local, private, yours
📊📊🔮🔍🔀📊📊📊🔍🔍🔀🔍🔍📊🔒🔥🔍📊📊🔍🚀🔍📊📊🔍🔥🔍🚀📊🔍🔮🔒🎯📊🤔🔍🔄🔁📊👨👨🕰👨👨👨🔍👨👨🤖📊📊🕒👀📊👨📁📊📊👀📊👀📊⚙📊👨🚀🎯🔍🎮🕳📊🔥🎯📊📊🤔🎯📊🎮
🌶️How to 5x your typing speed and be in a flow state while you type?
(no joke)
Use my customGPT - The Flow Speed Typist.
(link in the next post)
It’s a fascinating concept.
Bookmark to think about it later.
One of the AI’s Superpowers is its ability to understand extremely…
Feels like
@realGeorgeHotz
was right all the way with GPT4 size
Mistral release of mixture of experts model somewhat proved it
Mistral is 8 x 7B (feels like gpt3.5)
->
GPT 4 is 8 x 220B