Grok just got superpowers with App Connections + Automation!
Now you can connect: Gmail, Google Calendar, Slack, Notion, Google Drive and more!

In our digital world where every transaction is visible and every interaction can be traced, privacy is no longer optional, it's essential.

Ethereum not only went down it was forked. Stop sugar couting the facts. It is slow and unusable at 13 tps. L2s are not blockchains, they are centralized sequencers that just gather psydo "transactions", zip them up and write them in zips to Ethereum with a 7 day escrow window.

uses the zora subgraph so limited by that but latency is <1s from subgraph index

Aleph can handle complex motion and moving objects. The input video was in daylight, so I asked it to turn the lights off and generate a nighttime scene.

Security / reliability has always been at odds with usability. The design fundamentals of robust systems are well understood. They just tend to be to inaccessible to beginners.
This allows us to start to rebuild basic infra primitives to suck a lot

Programmable data is an interesting evolution for web3 applications. But how does this compare to existing decentralized compute solutions?

i don't think you understand
kimi k2 is such a great model. it's inexplicable how good it is.
i have been taking it for a test drive and it's mind-blowing

a verifiable trust layer for machines is the only way this works

11 years ago, the vision for Ethereum was published in the Ethereum Whitepaper.
The initial paper included ideas like sav…

ELON: EVEN WITH A GUN TO MY HEAD, I CAN'T READ YOUR MESSAGES ON X CHAT
"We're currently really improving the DM system.
So the new system called X chat, which involves audio, video calling, it's all fully encrypted, peer-to-peer, style like Bitcoin.
The test is that even if

Inference - not really, with one caveat (see below). Most models don't need this. The industry "standard" is 48GB meaning most models assume at least that. You'll be able to get even the largest models running on these with fewer parameters. Most image, video models are 48GB.

AI model validation - well most models requiring validators are 1B, 3B and 8B. These will easily fit on 24GB and 48GB GPUs. You'll essentially be able to run a validator on any datacenter GPU.

Clear apis and robust primitives are 100x more leveraged in the age of AI.

It's awesome to see AI generated code starting to push infra platforms towards better CS practices: clear semantics, strong typing, simplified state management etc.

Blindsight is Neuralink's next product after Telepathy. Its goal is to restore vision in individuals who were born blind. The Blindsight implant has already shown success in monkeys.

Kusama hit 143K TPS during Spammening using just a fraction of its potential.
With elastic scaling and shared security, Polkadot's potential appears promising.