The second shift is happening quietly, but it’s massive. Multi-agent AI. Instead of one system doing everything: → One agent plans your trip
→ One compares destinations
→ One finds activities Parallel intelligence. This is how AI starts to feel less like a tool and more
AI
-
Multi-agent AI: the quiet second shift towards parallel intelligence
By
–
-
Multimodal reasoning: AI that sees, understands, and decides instantly
By
–
The real leap isn’t better answers. It’s multimodal reasoning. → AI that sees images
→ Understands context
→ Makes decisions instantly Example: Point your phone at food → it suggests the healthiest option
Scan a product → it gives a real comparison, not biased reviews -
AI’s next wave: real-time visual decisions with Meta
By
–
Most AI today answers questions.
— Ronald van Loon (@Ronald_vanLoon) 5 juin 2026
The next wave will see what you see and make decisions with you, in real time.
That shift changes everything, from search to commerce to how we interact with the world.
I just broke this down using @Meta’s latest work.
Here’s what most people… pic.twitter.com/meE6gG5mpdMost AI today answers questions. The next wave will see what you see and make decisions with you, in real time. That shift changes everything, from search to commerce to how we interact with the world. I just broke this down using @Meta
’s latest work. Here’s what most people -
Critique of Laurence Devillers’ Contradictions on LLMs
By
–
Laurence Devillers, you say everything and its opposite. For years, you have explained to us that LLMs were hollow, without personality, and incapable of reaching a level comparable to that of humans. Today, you are alarmed that Claude seems to develop https:// x.com/lau_devil/stat /lau_devil/status/2062767846397039062 …
-
Uber driver unknowingly blasts AI-generated music
By
–
I still can’t believe we walked into a living litmus test of AI music in our Uber back from the bar in SF. The driver had no clue that his current favorite music that he was blasting was 100% AI generated. completely oblivious to the fact. he could not wrap his head around it
-
Success with Codex: Convert Writing Samples into a Skill
By
–
it’ll get better sooner thank you think but later than you want! in the meantime I’ve had tremendous success by give codex samples of writings and artefacts that I like and having it convert that into a skill use that skill for whenever it writes (ask codex to add it to your
-
Far more companies than people in AI world
By
–
There'll be far more companies than people in AI world.
-
NVIDIA LocateAnything predicts entire box as atomic unit
By
–
Most VLMs predict bounding boxes one token at a time — X1, Y1, X2, Y2.
— Satya Mallick (@LearnOpenCV) 5 juin 2026
But a box isn't text. It's geometry.
NVIDIA's LocateAnything predicts the entire box as one atomic unit. Parallel Box Decoding > next-token prediction for spatial outputs.
(Part 1 🧵) Breakdown 👇… pic.twitter.com/mQ83uMqtdvMost VLMs predict bounding boxes one token at a time — X1, Y1, X2, Y2.
But a box isn't text. It's geometry.
NVIDIA's LocateAnything predicts the entire box as one atomic unit. Parallel Box Decoding > next-token prediction for spatial outputs.
(Part 1 ) Breakdown -
SaaS to AI: End of Software Publisher Model and Middle White-Collar Workers
By
–
SaaS sold you a tool. AI sells you the result. This shift of three words is the death of the software publisher model. It is also the death of a massive fraction of intermediate white-collar workers. The developer no longer produces code. He validates an architecture.
-
No shortcuts in end-to-end learning for future models
By
–
End to end learning — We chose no shortcuts so we could learn, and build the knowledge and infrastructure to create many more models in years to come.