I build things with AI and explain the things I’m learning Agents, coding, video, visual creation and the latest shifts in AI
Joined February 2021
- Tweets31.7K
- Following1K
- Followers1.2K
- Likes35.4K
Pinned Tweet
Built a Gemini Live adaptation of this @HeyGen liveavatar demo.
I replaced the OpenAI realtime layer with @GeminiApp Live, while keeping the @TryLiveAvatar + LiveKit experience intact.
Live, tested end to end, and open source:
github.com/mahdijnt/liveavat…
Real-time AI experiences are now solved AND open sourced.
We worked closely with @OpenAI to build the best framework for it, combining GPT-Live-1, @TryLiveAvatar, and @HyperFrames_
Check out the demos below and show us what you build with it.
github.com/heygen-com/liveav…
Mehdi retweeted
. @SVGatorApp challenge on @contra
A detective board, rebuilt as a motion ad: connected clues, fast camera moves and one by one reveals.
I designed it to turn a platform’s story into an investigation—with replaceable images and editable text in a reusable template.
#svgatorchallenge
SVGator workflow :
part1 : app.svgator.com/editor#/cc74…
part2 : app.svgator.com/editor#/71b9…
. @SVGatorApp challenge on @contra
A detective board, rebuilt as a motion ad: connected clues, fast camera moves and one by one reveals.
I designed it to turn a platform’s story into an investigation—with replaceable images and editable text in a reusable template.
#svgatorchallenge
SVGator workflow :
part1 : app.svgator.com/editor#/cc74…
part2 : app.svgator.com/editor#/71b9…
this one hits harder
still short
still human
rising with hazels
real creators don’t wait, they claim their spot
studio.hazels.io/?ref=HZL-OH…
@0xhazels #WEAREHAZELS
140K people in less than 36 hours
but imagine what happens if even a fraction of them deploy Jev into real products
$0.042 / 1M input tokens
free output
193.6× faster
444.6× cheaper
that's the interesting part
not another chatbot
a decision primitive that can run everywhere
@typesafeai 's workflow evals
it's been < 36 hours
we've jev-ed 140k off the waitlist (our platform team trying hard to get EVERYONE off ASAP)
I've been told we're 9th biggest launch video of the year (and in very good company)
SO many cool demos! evals! ideas!
and still the most important number is going to be how much true automation y'all deploy!
Mehdi retweeted
it's been < 36 hours
we've jev-ed 140k off the waitlist (our platform team trying hard to get EVERYONE off ASAP)
I've been told we're 9th biggest launch video of the year (and in very good company)
SO many cool demos! evals! ideas!
and still the most important number is going to be how much true automation y'all deploy!
Mehdi retweeted
maybe AI agents don't need a bigger brain
maybe they need a faster decision layer
LLM:
"here's what I think we should do"
Jev:
"these are the probabilities of each valid action"
code:
"done"
that means you can put intelligence before and after an LLM call
route the model
choose the tools
check the output
decide whether to retry
then let code execute
that's a pretty interesting architecture
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
maybe AI agents don't need a bigger brain
maybe they need a faster decision layer
LLM:
"here's what I think we should do"
Jev:
"these are the probabilities of each valid action"
code:
"done"
that means you can put intelligence before and after an LLM call
route the model
choose the tools
check the output
decide whether to retry
then let code execute
that's a pretty interesting architecture
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
Mehdi retweeted
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
Mehdi retweeted
Hazels x Nabulines 👁⃤
50 WL spots up for grabs.
How to enter:
❤️ Like
🔁 Repost
👥 Follow @nabulines & @0xhazels
Drop your EVM address below.
50 winners.
Good luck.
Mehdi retweeted
Built a Gemini Live adaptation of this @HeyGen liveavatar demo.
I replaced the OpenAI realtime layer with @GeminiApp Live, while keeping the @TryLiveAvatar + LiveKit experience intact.
Live, tested end to end, and open source:
github.com/mahdijnt/liveavat…
Real-time AI experiences are now solved AND open sourced.
We worked closely with @OpenAI to build the best framework for it, combining GPT-Live-1, @TryLiveAvatar, and @HyperFrames_
Check out the demos below and show us what you build with it.
github.com/heygen-com/liveav…
Mehdi retweeted
AI is moving faster than most of humanity can adapt
So I asked three of the biggest AI systems a simple question :
What is the real goal of this AI transition, who is it being built for, who gets left behind, and what does the future actually look like?
I asked @ChatGPT , @claudeai and @grok to answer from the perspective of their own companies.l
Here are their answers :
Mehdi retweeted
GPT-6 Astra turned complex farm blueprints into a realistic walkthrough video.
It handled every step:
> Detailed drawings in AutoCAD
> 3D visualization in Blender
> Lifelike scenes with Higgsfield
AI is moving faster than most of humanity can adapt
So I asked three of the biggest AI systems a simple question :
What is the real goal of this AI transition, who is it being built for, who gets left behind, and what does the future actually look like?
I asked @ChatGPT , @claudeai and @grok to answer from the perspective of their own companies.l
Here are their answers :
But here's what I really want to know:
If AI agents eventually become capable of doing most intellectual work…
Who owns the intelligence?
Who decides who gets access?
Who benefits from the productivity?
Who pays the cost of displaced work?
And most importantly:
Who gets to decide what benefiting humanity actually means?
Your turn
@OpenAI @AnthropicAI @spacexai
@ChatGPT @claudeai @grok
Did I understand your position correctly?
What part of these answers would you defend and what part would you challenge ?
And to everyone else :
Are we building AI for humanity, or are we building a new layer of power that humanity will have to adapt to?
Reply with your answer
I want to hear both sides of the argument