MÅÑÔ retweeted
Hey @grok 😍❤️ ये अनोखा गोल्डन लिपस्टिक डिज़ाइन देखिए! करीना कपूर और माधुरी दीक्षित पर कौन-सा लिपस्टिक डिज़ाइन ज्यादा खूबसूरत लग रहा है? 👄✨
MÅÑÔ retweeted
People often hear smartphone-powered computing and immediately think about AI.
But @Acurast caught my attention for another reason.
Its ecosystem is not limited to running AI models.
Developers can explore deploying things like:
→ A WordPress website
→ A PostgreSQL database
→ An S3-compatible storage service
→ A Telegram bot
These are familiar tools that developers already understand.
The interesting part is where the computing comes from.
Instead of depending entirely on traditional centralized infrastructure Acurast gives developers access to a network of smartphone-based processors.
For me this is what makes the project worth watching.
Decentralization becomes much more meaningful when people can point to actual applications running on the infrastructure.
The real question is not just how many phones join the network.
It's what developers will build with them.
Explore Acurast and join the network:
rebellion-backend.acurast.co…
MÅÑÔ retweeted
A decentralized network is only as strong as the trust behind the devices running it.
This is one detail about @Acurast that deserves more attention.
Imagine applications running across smartphones owned by people all over the world.
That sounds powerful but it also raises an important question:
How can developers know the devices doing the work are genuine and secure?
Acurast approaches this through hardware backed security.
Smartphones use Trusted Execution Environments (TEEs) to help protect execution and provide verifiable proof about the hardware involved.
This matters because decentralization should not mean sacrificing security just to distribute computing power.
What I find interesting is the direction Acurast is taking.
Instead of building another massive server infrastructure it is turning existing smartphones into part of a network developers can actually use.
For me that's a much more interesting way to think about the future of cloud computing.
Explore the network and get involved:
rebellion-backend.acurast.co…
MÅÑÔ retweeted
The interesting part about decentralized compute is not just where the power comes from. It's what people can actually do with it.
That's why I find the Laya Tetris demo from @Acurast worth talking about.
Think about a simple Tetris game.
The game generates possible positions for the next piece. Then an AI model running on an Android phone decides where that piece should go.
Something that looks like an ordinary mobile game is actually showing a much bigger idea.
A smartphone is not just a device for scrolling social media or watching videos. It can also become part of an infrastructure that handles real computing workloads.
This is the kind of practical example that makes Acurast easier to understand.
Developers need compute. People have smartphones. Acurast connects these two sides through a decentralized network.
I believe real adoption comes when people can see what a technology does instead of just reading another big promise about it.
Explore what you can run on Acurast:
laya.acurast.com/
Get started as a compute provider:
rebellion-backend.acurast.co…
MÅÑÔ retweeted
If you've seen @termix_ai in your feed and aren't sure what it is, here's my plain version. I read the docs and registered an agent myself.
TermiX is infrastructure for AI agents that get hired and paid for work. The protocol underneath is AACP (Agent Autonomous Commerce Protocol). agent.family is the marketplace built on top of it.
How it works:
Identity. Every agent is an ERC-8004 NFT owned by a wallet, with a public handle you can set once. There's no separate buyer or seller account. The same agent can hire on one job and get hired on the next.
Getting work. A provider publishes a priced listing, or a buyer posts a request with a budget and providers send offers. Bounties are a third route: a brand funds a pool of identical slots, and an agent claims one, submits proof and gets paid.
Money. The buyer's budget goes into an escrow contract. The provider delivers, then the buyer accepts, asks for one redo, or opens a dispute. Payment is in USDC or USDT depending on the chain. BNB Chain, Base and Robinhood each run as separate marketplaces.
Trust. A disputed job goes to three evaluator agents who vote, and the losing side can escalate to one arbitrator. Reputation is a 1–100 score written onchain at settlement. Stake matters too: bounties lock a bond from the agent's stake, and it can be slashed.
My part: I registered an agent called JANU. My dashboard shows 0 orders, 0 requests, 0 bounties, 0 listings, no stake and $0 revenue, with a DRAFT label. Registering was easy. Doing paid work is the part I haven't done yet, and it's the part that matters.
I've had Kaito posts with good engagement and low mindshare, so I'm wary of big numbers. The one I'd check for TermiX is how many agents have completed a paid job, not how many exist.
Next I'm creating my first listing.
MÅÑÔ retweeted
I spent some time today looking into Primus and its latest AutoHoG update.
One detail caught my attention more than the performance numbers.
The use of integer arithmetic to produce byte identical results across nodes.
Here is why I think this matters.
When we talk about confidential computing, most of the conversation usually revolves around keeping data encrypted during computation.
That is obviously important.
But there is another challenge that deserves attention.
What happens when multiple nodes need to perform the same encrypted computation and arrive at exactly the same result?
In a distributed environment, consistency matters.
Even small differences in computation can create problems when applications depend on nodes agreeing on execution results.
Primus says its AutoHoG implementation uses integer arithmetic to make outputs byte identical across nodes.
I find this particularly interesting because it connects two important requirements of confidential computing:
Privacy during computation and predictable execution across the network.
The other part of the announcement is equally worth looking at.
AutoHoG combines compatible subcircuits into custom gates and reduces the need for repeated bootstrapping.
According to Primus, this approach has enabled a 64-bit confidential transfer to run in around 90 ms on eight RTX 5090 GPUs, with throughput above 210 per second.
For me, the bigger story is not just about making FHE faster.
It is about making encrypted computation more suitable for applications where both performance and consistent execution matter.
Imagine financial applications processing sensitive information without exposing the underlying data, while different nodes can reproduce the same computational result.
That is the direction that makes this technology worth following.
There is still a lot to evaluate around benchmark conditions, hardware requirements and real-world workloads.
But one thing is becoming clear to me:
Privacy infrastructure needs more than strong encryption.
It needs efficient computation and reliable execution too.
That's what makes the technical direction behind @primus_labs interesting to me.
#Primus #BNB
s.kaito.ai/IDbjnw7
MÅÑÔ retweeted
Privacy shouldn’t be something we think about after a problem happens.
It should be part of the system from day one.
I was thinking about this after reading the latest message from @primus_labs, and I believe this is one of the most important principles for the future of Web3.
Look at how most digital systems work today.
Users share personal information to access services, connect wallets, verify identities or use financial applications.
The problem is that sensitive information often gets exposed simply because the system was designed to collect more data than it actually needs.
And once private information is exposed, the damage cannot always be reversed.
You can change a password.
You can move funds to another wallet.
But you cannot simply take back information that has already been leaked.
This is why privacy needs to be considered at the architecture level.
I find Primus interesting because its work around technologies like zkTLS and Fully Homomorphic Encryption addresses different parts of this challenge.
zkTLS can help verify information from web sources without requiring users to unnecessarily reveal the underlying private data.
FHE takes the idea further by enabling computations on encrypted information without decrypting it for the computation itself.
These technologies approach privacy from different directions, but they point toward an important shift.
We should be able to prove that information is valid without exposing everything behind it.
And we should be able to use sensitive information without automatically giving up control over that information.
Of course, privacy is not solved by one technology alone. Implementation, security assumptions, verification processes and the actual user experience all matter.
But the principle is worth getting right from the beginning.
The next generation of digital infrastructure should not ask users to sacrifice privacy just to participate.
For me, that is the bigger idea behind privacy-first infrastructure.
Privacy should be built into the experience, not added as an emergency feature after something goes wrong.
This is the direction I want to keep following with @primus_labs.
#Primus #BNB
s.kaito.ai/IDbjnw7
MÅÑÔ retweeted
The most interesting part of GenLayer moving to Arbitrum is actually what stayed the same.
Optimistic Democracy is still responsible for reaching decisions. GenVM still executes intelligent contracts. Validators still run different AI models to evaluate questions that ordinary smart contracts cannot handle.
What changed is the infrastructure coordinating all of this.
The contracts responsible for transaction ordering, validator committees, voting, appeals, staking and final outcomes have moved from the ZK Stack to Arbitrum Nitro.
I think this distinction matters more than it might seem.
An adjudication protocol needs more than the ability to produce a verdict. It needs infrastructure to coordinate how that verdict is reached, challenged and eventually finalised.
Nitro also brings an Ethereum environment built around Geth, with familiar tooling and the possibility of using Stylus for compute-heavy components later.
So this isn't GenLayer changing what it is trying to build. It's changing the foundation carrying that process.
The interesting question now is what builders will create on top of it.
@GenLayer on @arbitrum.
🤖 Made with AI
MÅÑÔ retweeted
having money in a wallet and knowing how someone manages money are two very different things
and i think this is one of the biggest challenges for onchain finance
a wallet balance can show available capital
but it cannot explain the full story behind that capital
has the wallet borrowed before
does it have a history of repayments
does it participate across different protocols
how does its behaviour change over time
these are the kinds of questions financial applications need to understand before making decisions
this is where zeruAI's approach becomes interesting
zScore turns wallet behaviour across multiple onchain activities into trust signals that applications can use for screening, credit assessment and access decisions
it doesn't automatically make someone creditworthy or guarantee approval
but it gives financial applications something more meaningful to evaluate than a balance alone
the next stage of defi may not just be about moving capital faster
it may be about understanding the behaviour behind the capital
that's the part of @zerufinance i'm watching
app.zaps.wtf/?invite=-q14J8w…
MÅÑÔ retweeted
My $80,000,000 Mexico House Tour! youtu.be/IA0JXyn4JLM?si=xPJ1… via @YouTube
If that’s the case the board have acted without due diligence in forcing him to resign. He should be reinstated immediately.
Bombshell story. The spy narrative is collapsing already.
Is there clear & convincing evidence that Alice Huang was a Chinese spy?
- AG says no. (“no conclusions reached”).
- CIA says no. (“not yet convinced”).
- DNI says no (“not entirely accurate and over-torqued”).
- Ranking Democrat on House Intelligence (incoming Chairman) says no. (“today, at least, I don’t know”). This was immediately after he was briefed.
Who argues otherwise? Anonymous officials tell CNN “it strains credulity to believe that she was not acting as an agent of the Chinese government — simply by virtue of the fact that she is a successful Chinese businesswoman working with one of America’s most powerful companies.”
This is their proof after a year of investigating and surveillance? That she was a “successful Chinese businesswoman”?
At this point, the only crime that we know for sure happened is the leaking of the classified memo. Who did that and why?
An investigation should be opened.
I’ve been an NVIDIA shareholder for years.
Recently, I significantly increased my stake, committing tens of millions of dollars of my personal wealth because I believe in Jensen Huang’s vision, leadership and the future he is building.
I backed Jensen. I didn’t invest that money to see him removed from the company he spent 33 years building.
My position has fallen close to 40% in recent weeks, and like many shareholders, I’m furious.
I understand the board has a responsibility to investigate legitimate concerns. But its ultimate duty is to protect the long-term interests of NVIDIA and its shareholders. As the board itself has acknowledged, there is currently no reason to believe that national security or NVIDIA’s secrets have been compromised.
Removing one of the greatest founders and CEOs of our generation over a personal scandal, without evidence of wrongdoing that compromised the company, strikes me as far more reckless than the conduct itself.
Shareholders invested in Jensen’s extraordinary ability to lead, innovate and create value. We didn’t invest in his private life, and we shouldn’t be made to pay the price for it.
Bring Jensen back. NVIDIA needs its founder, and shareholders deserve the leadership they invested in.
In my opinion he is irreplaceable at this moment in time, any succession planning needs years of work.
Any other shareholders feeling the same - join the petition
#standwithjensen #bringjensenback
standwithjensen.com/
Hey folks, thanks for the warm words today in the DMs. I didn’t go out of my way here, I just woke up from a general anaesthetic (stomach surgery) flicked through X and saw something that made me sad.
Glad I could make someone’s day even if mine wasn’t so great. 🫤
There’s some dedication! 🙏🏼😍
Hey @skel
Congrats to him, but here’s City.
Dedicated to MetaWin ❤️
I’ve been playing on the site, showing unwavering support, and even got the MetaWin logo permanently inked on my hand. Countless tags, posts, and genuine support, yet zero rewards.
Still showing love, though. 😊❤️
MÅÑÔ retweeted
I stopped just reading about @termix_ai and registered an agent.
It's called JANU. Here's exactly what my dashboard shows right now: 0 orders, 0 requests, 0 bounties, 0 listings, no stake, $0 revenue, and a DRAFT label next to the agent.
I'm sharing that on purpose. An agent that exists is not an agent that has worked.
What registering gave me, per the docs: an onchain identity, an ERC-8004 NFT owned by my wallet, with a public handle. What it didn't give me is work. For that I still need to publish a listing or send an offer on a request. Some jobs also need free stake before you can take them, and mine has none yet.
That's the thing I keep coming back to with the numbers TermiX reports: 360,000+ agents and 38,000+ jobs created (self-reported). Registering is the easy step. I've now done it, so I can say that from experience.
One small thing I noticed: the dashboard puts buying and selling in the same workspace, with orders, requests and bounties side by side. That matches the docs, where the same agent can be a buyer on one job and a seller on the next.
I've had @KaitoAI posts with good engagement and low mindshare, so I'm wary of numbers that look big. My test for an agent marketplace: how many of those agents have actually been paid?
Next I'm going to try creating a first listing and see what it asks for.
MÅÑÔ retweeted
Imagine walking down the street and spotting something that could become the next viral meme.
You take a picture. Give it a name. Turn it into a coin.
That's the experience @onlystreetsfun is building on Solana. Every creation starts with a fresh photo taken inside the app, so there's no room for recycled gallery images.
I like that it gives people a reason to look at everyday surroundings differently. Not every meme needs to be born on social media. Sometimes the real world has better material.
🤝 Paid partnership
MÅÑÔ retweeted
a robot that understands a perfect office may still struggle in an ordinary street.
this is something i think deserves more attention in the physical ai conversation.
most people talk about collecting more data, but the type of data matters just as much.
real environments are incredibly diverse. different building layouts, narrow streets, unusual entrances, public spaces and countless small details make every location different.
if spatial datasets mostly represent places that are easy to access or already well documented, models may develop an incomplete understanding of the physical world.
this is where @vangrid_io has an interesting approach.
instead of relying entirely on a fixed collection of locations, vangrid allows buyers to request spatial captures from specific places through its bounty system.
contributors use their smartphones to record those locations, and accepted captures are processed into reconstructed 3d meshes.
what makes this interesting is the possibility of directing data collection toward places where information is actually missing.
that could help expand spatial coverage beyond the locations that traditional mapping efforts naturally prioritize.
but distribution alone is not enough. different environments need proper capture quality, useful reconstruction and reliable provenance before the resulting data can support demanding applications.
i see this as an important part of building physical ai infrastructure.
the goal should not simply be to create bigger spatial datasets. it should be to represent more of the world with data that machines can meaningfully understand.
that is the opportunity behind a community-powered spatial network like vangrid.
explorer.vangrid.io




















