The current generation of Agents are Reactive Prompt → React → Output The next generation of Agents will be Proactive Anticipate → Plan → Initiate As Memory, Extended Context Windows, and Local RAG systems get better, they'll unlock this new paradigm

You're gonna wanna check these guys out, one of the most cracked teams working at the frontier of decentralized ai Prime Intellect

This comment section is funny cause this is the part of ai twitter that farms engagement And everyone here clearly has the same model and prompt set for their reply 😂🤣

Applications and Frontends capture the value

Models are like servers, generic runtimes for compute Apps are like websites; tailored execution environments for context, with agents acting as dynamic interpreters In the end, the app layer will dominate

App cycle confirmed

Multi-Human to Multi-Agent Collaboration is the next frontier for Agentic UX

so back to the good ol' days? im in

Yup, the next unlock is multiplayer agentic experiences, we're actively working towards this at Zo | zo.me

It's so obvious I'm genuinely surprised noones approached this space The future is humans and teams operating Networks of Agents in multiplayer experiences We've unlocked a v1 of this at Zo | zo.me

Just one more L1/L2 bro, I promise this is the one for mass adoption

Wang Huning from the CCP Something akin to Kissinger x Thiel with true political power In Chinese culture, there's always the Emperor and the Emperor's Advisor He's never been rotated from his position and has advised all of the three last premiers of the CCP; Xi Jingping, Hu

Fascinating storyline for the decade ahead Don't agree with all the assumptions but overall the picture points towards mass bifurcation from the systems we know and rely on today I also think the writer underestimates the darkhorse that is AGI, especially one that may live on a

Nominative Determinism Sam Altman building Alt to Man Sundar Pichai pitching AI Stephen Schwarzman(Black in German) of Blackstone Many such cases

Paper is out. Congrats Pluralis! Decentralized networks can now train frontier AI models at scale.

This is a true 0 -> 1 in Decentralized AI and compute in general We can now pool heterogenous consumer grade hardware into a mesh training network for billion parameter models that are owned by the collective DeAI Supercycle is real

angga maulana You need to be bicoastal between NYC and SF Densest intelligence and ambition in the world in these two spots and gives you variance in actors

Here’s an accessible breakdown of Pluralis Research’s incredible paper. When we train large models on decentralized networks, the idea is to break them down into pieces and have different nodes process the different pieces. There are a few ways to do this. One way is low hanging

If you made it here, you can make it in crypto...