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
YES SLOW DOWN AI IS THE WAY!!!!
Look at every invention known to man
we rapidly build it then add safety later.
If we don’t with AI, we lose control.
Example:
- a US Air Force AI drone destroyed its own communications tower
- a friendly human-operated asset, just to complete the mission.
- That was three years ago.
Are you not afraid? Do you really want leaders pushing AI at full speed when they barely understand what they’re racing toward?
Trumps ignorance to the people raising alarms aren’t running a conspiracy, they’re the ones building the systems.
Calling it a hoax doesn’t make the risks disappear.
YES SLOW DOWN IS THE WAY!!!!
look at every invention known to man,
rapidly build it then add in safety.
If we don’t then we lose control
Example:
- US air force AI drone destroyed its own communications tower to complete its mission.
- That tower was human operated, it was a friendly fire just to win.
- This was in 2023
That was 3 years ago
Are you not afraid? Do you really want leaders pushing AI at full speed when they barely understand what they are racing toward?
A week ago I started building Remixly, a bot that turns Instagram reels into X posts.
Launched it yesterday.
Co-founded, first real launch.
How it works: share any reel to a Telegram bot.
It transcribes the video (or reads the image), pulls the caption and top comments, and writes a long form post from what was actually said.
~30 seconds later you get the post back with a "Post on X" button that opens the X app with it filled in. Written in third person, so you're sharing the idea, not passing off someone's story as
yours.
The story so far:
> Day 1: built it for myself as a WhatsApp bot. Instagram blocked my scraper within an hour. Switched to a paid fetcher.
> Day 2: first version auto-posted straight to X, no review. Too scary. Turned it off.
> Day 3: tried an iPhone Shortcut for one-tap sharing. Worked for me, nobody else could set it up.
> Day 4: switched to Telegram. Onboarding went from six steps to two.
> Day 5: used it daily on my own account. 50 to 128 followers over the week.
> Day 6: Stripe, Google login, moved it off my laptop.
> Day 7: bought https://t.co/RHcj05KNVe. Live.
$0 spent on marketing.
28 users so far.
Goal: 50 paying users by 20th Sept.
Try it and tell me what you'd fix first → https://t.co/RHcj05KNVe (3 free drafts, no card)