By
–
This is a seven character swap across 19 clips maintaining perfect environment & movement consistency 🤯
— Justine Moore (@venturetwins) 16 octobre 2026
If you had told me six months ago that AI video would get here this year, I never would've believed you… pic.twitter.com/gg8UEZk4Al
This is a seven character swap across 19 clips maintaining perfect environment & movement consistency 🤯 If you had told me six months ago that AI video would get here this year, I never would've believed you…
By
–
Spent years watching models ace a single image and completely whiff on anything that moves or changes angle. Today we’re fixing that.
— Andrew M. Dai (@AndrewDai) 16 octobre 2026
Try Glimpse now, our first visual thinking model with competitive frontier performance compared to Astra on key public visual reasoning… pic.twitter.com/KpoVLpn5Gm
Spent years watching models ace a single image and completely whiff on anything that moves or changes angle. Today we’re fixing that. Try Glimpse now, our first visual thinking model with competitive frontier performance compared to Astra on key public visual reasoning
By
–
Built a real-time forest walk in the browser 🦌
— Bihe Xi (@xbh_artist) 16 octobre 2026
Inspired by @MengTo 's scene work, I used Midjourney for scene refs, refined them into detailed prompts, then let Claude Opus 5.5 build the world.
Deer by Tripo P 2.0 API: model → rig → walk.
Play: https://t.co/sgVvr1qCSR https://t.co/BGDWM4oLca
Built a real-time forest walk in the browser 🦌 Inspired by @MengTo 's scene work, I used Midjourney for scene refs, refined them into detailed prompts, then let Claude Opus 5.5 build the world. Deer by Tripo P 2.0 API: model → rig → walk. Play: https://
developers.tripo3d.ai/usercases/fore
st-deer/
…
By
–
🦖 Gemini 4 Argon looks at your LEGO DUPLO pile, designs from exactly those bricks, and @specsfordevs walk you through building it. Lens built with CLAD.
— Stijn Spanhove (@stspanho) 16 octobre 2026
How it works:
1. Photograph your bricks
2. Gemini 4 Argon counts them and designs one buildable figure, step by step.
3.… pic.twitter.com/K1bdtR5yBi
🦖 Gemini 4 Argon looks at your LEGO DUPLO pile, designs from exactly those bricks, and @specsfordevs walk you through building it. Lens built with CLAD. How it works: 1. Photograph your bricks 2. Gemini 4 Argon counts them and designs one buildable figure, step by step. 3.
By
–
And @A2RLeague goes 250 mph with AI. Proving your point about speed wrong too.
By
–
Humans kill 40,000 people in USA every year. Every year. AI will be way way safer
By
–
Builders are getting such fantastic new tools.
— Robert Scoble (@Scobleizer) 15 octobre 2026
Won't this help many people build new things?
Which leads to new companies?
Which leads to jobs? After all, even though AI can do a ton of work it can't go and setup a booth at a tradeshow yet.
So you gotta hire people to do… https://t.co/98OTATPizQ
Builders are getting such fantastic new tools. Won't this help many people build new things? Which leads to new companies? Which leads to jobs? After all, even though AI can do a ton of work it can't go and setup a booth at a tradeshow yet. So you gotta hire people to do
By
–
Ask an AI agent to find you a hotel in New York and you get back paragraphs to read.
— Robert Scoble (@Scobleizer) 15 octobre 2026
In the @openuidev demo, the same request comes back as hotel cards with photos, a nightly budget slider, and buttons to narrow it down. Developers choose the building blocks, and the agent… https://t.co/03tzZSWart
Ask an AI agent to find you a hotel in New York and you get back paragraphs to read. In the @openuidev demo, the same request comes back as hotel cards with photos, a nightly budget slider, and buttons to narrow it down. Developers choose the building blocks, and the agent
By
–
Every robotics team I talk with says the same thing. Collecting data is the easy part. Knowing whether it's any good is the hard part. @BuildOnSapien is making a bet on that. A team sends its hardest dataset and its own rules for what good looks like. Sapien gets three days to
By
–
Phone switchboards worked fine until too many people started calling. The fix wasn't a faster operator. It was a new system.
— Robert Scoble (@Scobleizer) 15 octobre 2026
Nick's point is that CRMs are hitting the same wall, except now it's agents calling, nonstop and in parallel.@attio's answer is to look like a plain SQL… https://t.co/qMBmDcdnmR
Phone switchboards worked fine until too many people started calling. The fix wasn't a faster operator. It was a new system. Nick's point is that CRMs are hitting the same wall, except now it's agents calling, nonstop and in parallel. @attio
's answer is to look like a plain SQL