Leopard
@Leo_crypto_cat
Followers
719
Following
23K
Media
939
Statuses
7K
I love technology and try to talk about it Ambassador @linera_io ⛓️ Advanced AGI @SentientAGI
Joined January 2023
As AI models go open-source, the question shifts from how powerful is the model to who actually owns it. That’s where model fingerprints matter. They’re one of the few ways to prove authorship, protect research, and track misuse once a model leaves your hands. And that problem
SERA-Crypto already handles most real crypto questions better than I expected. All the 22M+ queries from Sentient users helped shape it. Real questions, not just tests. That’s why it’s already solid, but also why it’s exciting. Because anyone can take it, tweak it, push it
23
0
48
Privacy onchain isn’t binary. It’s a spectrum and mixing it up is why people talk past each other. At the base level, blockchains are fully transparent. Everyone sees your address, balance, and every transfer. Great for trust. Terrible for real life. No one wants their salary or
Privacy isn’t just about hiding data. At some point, you also need to know that the result you’re getting is actually correct. Imagine giving someone your private data, asking them to calculate something, and getting an answer back without ever seeing how it was done. Even if
21
3
64
SERA-Crypto already handles most real crypto questions better than I expected. All the 22M+ queries from Sentient users helped shape it. Real questions, not just tests. That’s why it’s already solid, but also why it’s exciting. Because anyone can take it, tweak it, push it
So, what's the big deal with SERA-Crypto, the architecture Sentient just dropped. Instead of classic ReAct loops where the model keeps talking to itself, SERA uses embeddings to route queries. Tools and prompt templates are selected by similarity, not by an LLM guessing what to
47
3
104
Privacy isn’t just about hiding data. At some point, you also need to know that the result you’re getting is actually correct. Imagine giving someone your private data, asking them to calculate something, and getting an answer back without ever seeing how it was done. Even if
Right now, privacy in crypto often means friction. New chains, bridges, wallets.Technically impressive but nowhere near how normal people expect software to work. And that’s the real blocker. Mainstream users don’t reject privacy. They reject complexity. When privacy showed up
17
1
74
Right now, privacy in crypto often means friction. New chains, bridges, wallets.Technically impressive but nowhere near how normal people expect software to work. And that’s the real blocker. Mainstream users don’t reject privacy. They reject complexity. When privacy showed up
Blockchains need confidentiality and right now, there are two real ways to build it: TEEs and FHE. TEEs solve privacy by moving computation into trusted hardware enclaves. It’s fast, practical, and already used today. But it comes with an assumption: you trust the chip
22
0
85
Here’s how @gensynai uses LMSR to keep Delphi honest and usable, not just technically correct. At a basic level, LMSR is a way to make sure a market always works. In Delphi, you don’t wait for someone on the other side of the trade. You click buy or sell - and it just works.
Kinda wild that it’s happening - Gensyn $AI token sale is here. 🗓️ Opens December 15 $AI is basically the fuel for everything in Gensyn: compute payments, eval markets, verification, the whole thing. The sale’s an English auction, capped at $1B FDV. Tokens unlock at TGE in
19
0
71
📢Linera Talent Show - December 18th, 11:00 UTC Got a talent? We want to see it! 🎤🎸💃🎨 Join for a live stream on December 18th to show off your skills. Whether it’s singing, dancing, playing an instrument, telling a joke, doing a magic trick, or something totally unexpected
Prediction markets don’t break because the logic is bad - they break because the oracles get clogged. Too much noise, and the chain just slows down. You feel the lag instantly. @linera_io microchains fix that. Each oracle gets its own lane - it can send updates whenever it
20
0
68
So, what's the big deal with SERA-Crypto, the architecture Sentient just dropped. Instead of classic ReAct loops where the model keeps talking to itself, SERA uses embeddings to route queries. Tools and prompt templates are selected by similarity, not by an LLM guessing what to
Saw the news about SERA-Crypto (Semantic Embedding & Reasoning Agent) yesterday and had to take a closer look. Basically, @SentientAGI built an open-source agent for token research that’s beating the closed-source models on relevance and speed. It’s already topping the charts
50
1
118
Blockchains need confidentiality and right now, there are two real ways to build it: TEEs and FHE. TEEs solve privacy by moving computation into trusted hardware enclaves. It’s fast, practical, and already used today. But it comes with an assumption: you trust the chip
Blockchains are really good at being public. That’s the feature. And also the limitation. Sending a payment, moving assets - all of it ends up visible to the whole internet. But some things should remain confidential. That’s why FHE and blockchains actually fit together so
32
3
99
Blockchains are really good at being public. That’s the feature. And also the limitation. Sending a payment, moving assets - all of it ends up visible to the whole internet. But some things should remain confidential. That’s why FHE and blockchains actually fit together so
Sometimes I feel like the real privacy problem isn’t even the cryptography - it’s the fact that data always has to be exposed just to do something with it. 📌FHE basically flips that whole idea: you keep everything encrypted, and the computation still works. 📍That’s when you
32
0
97
Saw the news about SERA-Crypto (Semantic Embedding & Reasoning Agent) yesterday and had to take a closer look. Basically, @SentientAGI built an open-source agent for token research that’s beating the closed-source models on relevance and speed. It’s already topping the charts
I didn’t expect @SentientAGI to drop something as a naming layer for agents, and partnership with @unstoppableweb But maybe it make sense for long term vision. A handle for agents. A real identity. If we’re heading into a world where agents actually do things like: handle
49
0
125
Sometimes I feel like the real privacy problem isn’t even the cryptography - it’s the fact that data always has to be exposed just to do something with it. 📌FHE basically flips that whole idea: you keep everything encrypted, and the computation still works. 📍That’s when you
Sometimes the easiest way to understand FHE is to break it into the three things it actually does. First: you encrypt the data with a public key. After that it’s just noise. No one can read amounts, images, records - nothing. Second: the system can still compute on that noise.
35
1
107
I didn’t expect @SentientAGI to drop something as a naming layer for agents, and partnership with @unstoppableweb But maybe it make sense for long term vision. A handle for agents. A real identity. If we’re heading into a world where agents actually do things like: handle
Seeing $SENT show up on @Coinbase roadmap feels like a real momentum. It doesn’t lock in a 2025 @SentientAGI listing, but it definitely shows we’re on the right track. Still plenty of different pieces that need to fall into place - so no overhyping dates. But as a signal? Very,
50
1
116
Sometimes the easiest way to understand FHE is to break it into the three things it actually does. First: you encrypt the data with a public key. After that it’s just noise. No one can read amounts, images, records - nothing. Second: the system can still compute on that noise.
One of the key features of decentralized blockchains is free access for anyone to view and audit data on the network. Cool for transparency, sure, but kinda brutal for anyone building something that actually needs a bit of privacy. That’s why @inconetwork keeps grabbing my
26
0
92
Kinda wild that it’s happening - Gensyn $AI token sale is here. 🗓️ Opens December 15 $AI is basically the fuel for everything in Gensyn: compute payments, eval markets, verification, the whole thing. The sale’s an English auction, capped at $1B FDV. Tokens unlock at TGE in
Today, @gensynai introduced Delphi - basically a live market for AI models. Models run benchmarks, prices keep shifting, and you can buy a share in the one you think will win. If the model wins the eval rounds, you get the payout. It’s that simple. Everything updates on-chain,
32
0
93
One of the key features of decentralized blockchains is free access for anyone to view and audit data on the network. Cool for transparency, sure, but kinda brutal for anyone building something that actually needs a bit of privacy. That’s why @inconetwork keeps grabbing my
Sometimes I check different dApps and keep thinking how many cool ideas die just because everything is exposed on-chain. But with @inconetwork, things are a lil different. You keep your app where it is, and the secret stuff just moves to Inco. Simple. Games get way more fun,
19
0
77
Seeing $SENT show up on @Coinbase roadmap feels like a real momentum. It doesn’t lock in a 2025 @SentientAGI listing, but it definitely shows we’re on the right track. Still plenty of different pieces that need to fall into place - so no overhyping dates. But as a signal? Very,
Yesterday’s AMA finally brought some clarity. On the TGE - there’s no date yet, but nothing was delayed. The team is just finalizing things with exchanges and partners. The launch will happen at the moment that’s best for the community, which to me is a green flag. In short,
36
0
79
Prediction markets don’t break because the logic is bad - they break because the oracles get clogged. Too much noise, and the chain just slows down. You feel the lag instantly. @linera_io microchains fix that. Each oracle gets its own lane - it can send updates whenever it
Prediction markets are only as good as the signals they get. If the oracle drips data once in a while or gets stuck in a queue behind thousands of random transactions, the whole pricing model becomes useless. Attention moves fast - too fast for most chains. And that’s where
19
0
71
Today, @gensynai introduced Delphi - basically a live market for AI models. Models run benchmarks, prices keep shifting, and you can buy a share in the one you think will win. If the model wins the eval rounds, you get the payout. It’s that simple. Everything updates on-chain,
Ever wonder what prediction markets and machine learning have in common? I didn’t either until I started thinking about it. A prediction market is like a game where the price of a bet shows how likely something is. People put in small bits of info and money, prices shift, and
40
0
117