been looking quite a lot into @octra lately, and I think it is one of the more interesting privacy projects right now
most people hear privacy crypto and think about mixers or hidden transfers, but Octra is trying to do something bigger. they want computers to use your data without ever seeing the actual data
this is possible through FHE. imagine putting numbers inside locked boxes. a computer can calculate with them and return the correct answer inside another locked box, without ever opening any of them
crypto needs this because almost everything on normal blockchains is public. your balance, transfers, trades and defi positions can be tracked by anyone. that works for transparency, but not if we want normal people, companies and banks to properly use crypto. nobody wants their salary, savings, customers and every payment public
Octra is building its own L1, but it can also work as an encrypted compute layer for other chains and apps. this could make private payments, hidden exchange orders, lending without exposing your full financial history, private tokenized stocks and AI using personal data possible
the most interesting part is their own FHE system called HFHE. it uses hypergraphs together with cryptography from the LPN family, which is connected to difficult problems in code based cryptography
the simple idea is that hidden information is passed through a large binary matrix and mixed with noise. producing the result is easy, but reversing it to find the original information is extremely difficult
in HFHE, encrypted values are represented through masked edges in a hypergraph. the network can join and combine these edges to perform calculations without knowing what they contain. addition mainly joins edge sets, while multiplication combines edges and their masks
many edges can be processed independently at the same time. this means the work can potentially be divided across CPUs, GPUs and different machines instead of everything happening through one long calculation
Octra has also added recrypt, which refreshes encrypted data so calculations can continue, and edge compaction to reduce how quickly the structure grows during multiplication
this is different from TFHE systems like Zama, which are mainly built around LWE-style ciphertexts and bootstrapping. HFHE currently has larger ciphertexts and is far less mature, but its possible advantage is that more of the work can happen in parallel
Octra already has a working network, wallet, private balances, stealth transfers, node software, its own VM and public HFHE code. they are also building Circles, which are private environments where apps, files, storage and access rules can exist together
ofc, there are still major risks. HFHE is new cryptography and needs much more independent testing and formal verification. the network is still early, the user experience is complicated and there are not many large apps using it yet
I am not saying Octra has solved private compute, but they are working on something crypto genuinely needs. if crypto is going to be used for real money, businesses, identity and AI, everything cannot stay public forever
ethereum:0x4647e1fe715c9e23959022c2416c71867f5a6e80
Aliaian retweeted
The best project I've found so far this cycle : TensorCash, the verification and settlement layer for global AI inference.
The game of ever evolving liquidity has meant 99% of traders are absolutely rinsed chasing FOMO KOLs , bundled claude coded slop and new narratives daily. Nowadays, there is barely any plays that go up more than 24 hours because every noob in here turned into a deployer and there's more devs than traders.
The game is solved, your make it days are essentially over. Most of you guys will NEVER even make a cent from this game, its obvious . Unless you're a top 0.01% trencher simply put u are not going to make it.
Even if you deploy ain't nobody going to buy your little coins, back to McDonalds applications unless... If you're like me and not interested in this dumb ass fortnite bundle PvP between 100 chains on 100 launchpads this may be the play for you. If you're tired of bundled memes, larp tech and launchpads on 123 chains and want something interesting to research, see tensorcash:native tensorcash.
=========== What is tensorcash:native and why are so bullish mr popseye? =====
Simply put, I look at the narratives in real life. Not some dumb Elon tweet, a zoo meme or whatever silly agent narrative ur KOL came up with on the spot to farm you. Rn its obvious the world's taken over by AI, AI AI, it's all you hear. The big part of AI that most people have to come to realise that’s important is inference.
Recently in crypto inference marketplaces have gone to 100m and projects that actually have real world implications built on blockchain are going to 5b + rn. POD is sitting at 300m FDV , VVV is at billions. Anything inference related deserves to have attention. The market and rightfully so is hungry for inference projects and legitimate ones at that.
In any work we do nowadays, we use AI, AI models, LLMs, everything NEEDS and will continue to need AI. That part is undebatable. However, the important part of these AI models is the inference that it uses. But as more and more providers, customers, teams, companies use these AI models, how can they be sure the inference they pay for is what they are getting. Eg claude has several different models with diff amount of tokens, how smart the model is (fable vs haiku). Any provider can just charge u for an expensive model and give u a cheap one (I don’t mean claude but its an example of how differing models cost different).
Actually its proven this is the case for almost half the time you pay for a model, you are being lied to and sold a cheaper one.
The AI model and inference LIE
Researchers auditing third-party API resellers found almost half of their fingerprint tests (45.83%) failed to match the advertised mode. Source: arxiv.org/abs/2603.01919 .
TSC is a layer 1 blockchain that ensures these models are actually what you are paying for. It makes the underlying computation independently checkable. The computation can also serve a paying customer, allowing miners to earn service fees AND mining rewards from the same job. It's a permissionless L1 blockchain where miners, instead of wasting resources in brute forcing the next block, are running trained AI models, a process called inference, and compete for TSC
That gives AI providers more room to compete on prices, while still covering their hardware costs. For the buyers, verification gives a reason to consider an unknown provider: there is evidence of what ran, beyond the company’s name on the bill. It's a form of decentralization of inference, because you don't have to trust anymore, you can verify. (quoted from anapneo)
In summary
- Providers (datacentres, GPU operators, rotating BTC miners) sell AI inference
- Users (apps, agents, businesses) buy it via AskTilly and other marketplaces
- TSC sits underneath as the trust layer, proving the AI work you paid for was actually performed
- Clients pay fiat → AskTilly and other marketplaces buys TSC and pays miners → miners sell less → price up → mining more profitable → more miners, bigger network → more clients
Plenty of Web3 infrastructure projects build impressive technology but ultimately remain stuck inside Web3, waiting for someone to actually use it. TensorCash already has AskTilly being built around its verification infrastructure. AskTilly is being built by Enrico Ottavian, who spent 11+ years at Goldman Sachs and became a Global Head in macro/rates structuring before moving into AI markets.
Again , already there has been INSTUTIONAL players using tensorcash. Asktilly, by Enrico Ottavian ex Goldman Sachs and Samad Javadi ex morgan Stanley, efinancialcareers.jp/news/th… is part the ecosystem with the network produces over $12M a month of inference at market prices, . boblabs.eu/runs a pool. It is of my opinion this continues to get more and more adoption from instituions. Not crypto native audiences because this is a project for real world catalysts not the poop broke world of web3.
On the technical side, Ismael Aadnan / Bob Labs is working directly around the TensorCash ecosystem and inference infrastructure.
So you've got a fairly unusual mix:
AI engineers +PoW/crypto infrastructure + Goldman structuring experience + Morgan Stanley+ institutional capital networks
There have also been conversations with some much bigger names
Again the target isn't crypto native audiences because this is a project for real world catalysts not the small broke world of web3 where we sit there waiting for the dev from the jungles of Vietnam or India to not rug us for 300 bucks.
Putting it simply:
> Provider: “I ran the model you paid for.”
> TensorCash: “Here's the proof.”
Personally at 20m circ and with very low emissions I look at other similar projects going to billions (and probably going higher) and I think this is an insane opportunity and actual tech to get behind for once. Prl rightfully so is getting big attention at token2049, so it looks like inference projects are getting buzz. Criniti just posted an article about AI x crypto. Again the target is tradfi and instituions.
This is solving a projects 300B+ dollar industry going into probably trillions soon at this rate with institutional clients already and continuing to show interest and similar projects already at billions. Imagine the blockchain that solves the biggest pain point for AI and customers and providers atm.
Remember olas at the start u had no one talking about it except two random accounts with 0 followers. Well similar here. There's definitely been some groups that have talked about it but realistically on X only one other person has found it and made a good thread. (borrowed a few quotes as he is so good at writing). nitter.cf/anapneo/status/2106376…
If you are interested in buying its only available on safe trade atm safetrade.com/exchange/TSC-U…
I would also recommend using LIMIT orders, do not market buy as the orderbooks are thin and it takes time to accumulate. I would wait for a dip using limit orders. Please do not market buy click as the orderbooks are thin and get slipped hard.
There are other things about this but I need to write another thread as this got too long
nitter.cf/Tensorcash
Aliaian retweeted
One of the most undervalued projects on Robinhood Chain at the moment.
Insane lore + original dev from 2016 took over the CTO & is actually building something trying to run it back.
And V2 migration coming up pairs NASDANQ w/ QQQ & 100% of fees airdropped to holders feels like the final step to run.
Have been slowly and steadily stacking the last month.
Not sure when it runs, but expect it to run hard when it finally does.
imagine building a joke stock exchange and hearing from Nasdaq
in 2016, Redditors were joking about buying memes before they blew up. NASDANQ took it far enough to become a platform, with fake money, trading firms and people chasing beta keys
then the actual Nasdaq came knocking with trademark paperwork, and the project shut down
a decade later, original creator @DanqDev is back with robinhood:0x51fb76be80ab6daaa345d818f4e06441816b4fea on Robinhood Chain, and trading memes has become an actual industry. pretty good time for the original joke to make a comeback 🫡
Aliaian retweeted
i think ppl are sleeping on $MONEY
regardless of the meta/narrative of the cycle, stablecoin usage increases.
eUSD is dollar minted against SPY
fully collateralized
reserves disclosed
later, they will add other stocks/ETFs, gold in the reserve too.
the current product and yield comes from the trading fees generated with the $money token
- 30% buys the token
- 60% pays real SPY yield to those that mint/stake eUSD and $money
- 10% funds the protocol
this is not their only product, the plan is to keep launching more revenue generating products that increase the yield given to eUSD minters/stakers.
they want to build an ecosystem of fee sources.
dev bg also pretty stacked, being former gitcoin core dev.
last time i saw smth this interesting, defi wise, was $NET
allocated accordingly.
Aliaian retweeted
we've just started testing resource attestation for validators (GPU for circle inference, HFHE actor workloads, storage, etc)
the specific use case is a marketplace of ready made circle droplets, creating an app in one click, an isolated platform, storage, or database for corporate tasks, or a personal cloud for maniacsб etc,
universal tool - oct coin as a bond for validators, as a reward, fees, and as a resource for paying for useful tasks (and this is not about money at all) – at London, England
pretty excited for @GIWA_by_Upbit
one thing i think ppl underestimate with new chains is having a mother cex behind them
> BSC got Binance
> RH chain got Robinhood
> GIWA got Upbit
that gives you a huge existing userbase, liquidity + distribution from day one.
but the listing angle might be even more interesting
Korea is one of the most retail/speculation heavy crypto markets in the world and they fucking love trading alts. an Upbit listing alone can change the attention + volume around a token
so if GIWA gets its own native runners, they have something most new chains don’t:
a natural mother cex where the biggest ecosystem tokens could potentially end up listed
obv nothing is guaranteed, but i think Upbit knows what it takes
compare it to something like Arc. Arc got a ton of onchain attention early, but there isn’t a mother cex sitting behind it with millions of traders and a direct listing venue
think BSC/Binance, RH/Robinhood
that’s why im pretty excited to see what happens when mainnet opens. also most people haven’t even heard about GIWA yet
been digging deeper into own.money / $MONEY and imo the easiest way to understand it is as a degen credit layer for tokenized stocks
Robinhood brought assets like SPY onchain, Own lets you actually make that exposure productive
instead of selling your SPY when you need liq, you deposit it into Own and mint eUSD against it. min collateral ratio is 150%, so youre basically unlocking dollars against your SPY exposure w/o selling it
and thats an important diff vs something like Aave. on Aave youre borrowing stablecoins supplied by other users, on Own youre minting eUSD directly against your own SPY collateral
so basically:
SPY -> collateral -> eUSD liquidity
then $MONEY is what connects the product into the flywheel
$MONEY trades against SPY on Pons V2 and the trading fees are collected in SPY. those fees are split 30/70
30% -> buys + burns $MONEY
70% -> goes into the buffer used to pay SPY rewards to eUSD stakers
but if you want to maximize your share of those rewards, you need to stake $MONEY.
and imo thats the important part
more eUSD users -> more demand to buy + stake MONEY for the boost -> more MONEY trading -> more SPY fees -> more MONEY bought + burned + more SPY rewards for stakers
so the flywheel actually runs through the token itself
and the rewards are paid in SPY instead of newly printed MONEY, so its not the usual defi setup where you stake the token and just get paid in more inflationary token
SPY creates the collateral, eUSD creates the dollar liq, MONEY increases your share of the yield and MONEY trading creates the fees that fund both the burns + SPY rewards
imo thats the cleanest way to understand it
you keep your SPY exposure, unlock dollars against it, and if you want to maximize the yield on that eUSD you need to buy + stake MONEY
that creates more MONEY trading, which creates more fees, which burns MONEY and funds the SPY rewards that made ppl want the boost in the first place
thats the flywheel
Aliaian retweeted
Soon enough the majority of onchain traders will be holding tokenized stocks
The natural course is then to look at which protocols allow holders to utilise their tokenized stocks. Effectively these projects have the potential to reach very high valuations akin to those we saw in og defi szn.
$Money is well positioned to take advantage of this and is a first mover on rh. The project allows users to borrow eUSD against tokenized S&P holdings and earn yield.
Have been speaking to the team and am looking forward to the updates they have planned.
Aliaian retweeted
The SEC just legalised trading tokenized stocks onchain.
But the most useful thing you can do with a share is not sell it.
It's borrow against it, keep the position, and earn yield.
Nobody wrote a rule about that. So we're building it.
Robinhood supports the @SECGov innovation exemption.
Americans deserve access to crypto technology and all of the financial innovations it makes possible, including instant settlement, 24/7 trading, and fractionalization by default.
Aliaian retweeted
guys I think none of you actually realizes how undervalued $logos is
theres a real difference between telling an agent "dont use leverage" in a prompt and mathematically proving the enforcement code actually blocks it every time
prompts are probabilistic, the model interprets and mostly follows instructions until it doesnt, and with real capital that one time is the whole problem
a rules engine checking trades before execution is already better, thats deterministic code instead of hoping the model behaves
but then the question becomes whether the checker itself is correct, mandates stack multiple conditions at once, position limits, cash minimums, concentration caps, and its easy for enforcement logic to have a gap in some edge case nobody tested for
thats the part @Logos_Layer is actually solving
aristotle generates a formal proof that the checker logic satisfies the mandate across all inputs, not just the cases someone thought to test
then lean's kernel independently rechecks the math and rejects it if it doesnt hold
so youre not trusting a prompt and you're not just trusting untested code either
the enforcement layer is proven to match the spec before it ever touches a live trade
@vladtenev 's own company built the prover they're using here
if robinhood chain becomes the venue for tokenized stocks and agentic trading (which I think it will), the enforcement layer between agents and capital is the bottleneck people are skipping past
this goes so much higher
(also read @Aliaian write-up - chad piece)
This quoted post is unavailable.
Aliaian retweeted
Autonomous trading agents are obviously coming.
The problem nobody really talks about is what happens when you give an AI full control of your portfolio and the only thing stopping it from doing something completely regarded is… a prompt.
$LOGOS is basically building the guardrails.
Instead of telling the agent “don’t use leverage” or “don’t put 50% into one position,” the mandate becomes deterministic rules that every trade has to pass before execution. If the trade breaks the rules, it simply gets rejected.
And this isn’t just AI vaporware.
The product is live, the reference vault is deployed on Robinhood Chain, and the core checker has 591 theorems proved in Lean with zero sorry/axiom shortcuts according to the project. They’re also very clear that the contracts themselves are not audited yet, which is worth knowing.
HOOD. Base. On-chain equities.
What do they all increasingly have in common?
Agentic trading.
Feels like one of those parts of the cycle nobody gives a fuck about until suddenly everyone does.
Nice Robinhood ecosystem punt for me and definitely a longer-dated hold.
If autonomous agents become a serious part of on-chain finance, something needs to stop them going full Skynet with your portfolio.
Aliaian retweeted
A lot of people asking me about $BEM, since most information is in chinese and no one seems to know what it is.
First of all this is a great article by @blueshirt666 ex @binance on Tapeout $BEM.
A lot of progress has been made since then, but it’s still one of the best reads out there. I’ve been doing due diligence on this ever since I came across it and have bought multiple coins to get exposure to the project.
Now it looks like the official token, $BEM, for the protocol is finally out.
Some numbers on $BEM since nobody seems to be tracking this properly. (credits to @leyten, he helped me a ton doing due diligence on this)
Mining went live 4 days ago. 24,338 $BEM has been mined since. Emissions are 7,200/day, all of it coming from mining. No premine, no team bag, nothing to unlock.
$BEM is at $57 with a $1.4M market cap and did $1.5M in volume over the last 24h. That ratio is not normal.
The part people are missing is the gate market. You burn gates (ERC-1155s) to deploy a circuit, and they’re gone forever. Lifetime gate volume is 2,553 BNB, with 527 BNB in the last 24h.
Behemoth gates were minted at 0.0005 BNB and last traded at 0.1035 BNB — a 200x in a week, pricing the 100K supply at ~10,000 BNB. Standard processor gates went from 0.0005 to 0.0064 BNB with a 1M supply. At the last trade, the gates are worth multiples of the token.
Network stats right now: 400+ mining wallets, 5,000+ live circuits competing across 268 tasks, 763 processors deployed, 16,066 circuits taped out in total, and 20,000+ gate-holder positions.
Roughly $400/day flows to every 1,000 verification weight. Some of the top miners are running SAT solvers just to hold their slots. It’s an actual compute war over a $1.4M token.
@Blonskr is shipping daily, and his work has already been acknowledged twice by @cz_binance.
Aliaian retweeted
I BET A WHOLE BUNCH OF PPL STARTED GRAVEDANCING ON THE NO1 STOCK LAUNCHPAD ON BASE $BSTONK AFTER IT HAD A LIL DIP AFTER THE FIRST RUN UP
NOW WE GOT RUNNERS LAUNCHING ON THERE AND I AINT SEE ANYONE CHIRPING ANYMORE
DEV KEPT SHIPPING UPDATES CONTINUOUSLY EVERY SINGLE DAY WHILE 50 OTHER RUGGER DEVS RUGGED DAILY ON LAUNCHPADS
BUT HEY COS THE BRANDING AINT UR TASTE AND NAME WASNT UR LIKING IT WONT PULL A PONS AYE ?
POSTED DAILY ON MY TG ABOUT BSTONK AND UPDATES
AHHHHHHHHHHHHHH
People still haven’t realised the $PONS and $STONK of base launched a few hours ago and still under 1m
Just like we rode PONS 100x to 70-80m and riding stonk to 10x and higher, we ll be doing the same for this while everyone’s begging Cobie to interact with a cat (I’m in too dw, just thought it’s funny the whole reason base is reviving their chain is literally for their stocks launch but everyone’s not even looking for plays there still and late as always)
Aliaian retweeted
$BSTONK pays you to hold.
🔥 8.12% of supply burned - forever
💵 Every trade's fee → USDC, straight to your wallet
🧺 The $BSTONK basket pays out in kind, automatically - various RWAs & coins, direct to your wallet.
No staking. No claiming. Just hold.
basestonk.io
did another dive into $SUMMER today, and there’s some new updates
Binance Chinese Community just released Summer episode 8:
binance.com/en/square/post/3…
this episode matters more than the earlier episodes because Summer is literally explaining how an unknown movie became a meme, how memes turn into memecoins and what actually prices attention onchain they even linked the Niu Lai meme covered in the episode directly to its Binance Alpha page
so Summer clearly isn’t being left as some random one off comic character like most ppl thought at first
Binance is continuing to build her as their AI Companion Officer, and now they’re using her to explain memes + actual Binance Alpha assets
at the same time, $SUMMER was added to Gate Alpha HOT yesterday
the bet here is that the market has selected one dominant Summer token before Binance finishes building out the character/IP, similar to how CT tried to frontrun PALU, which reached a $120m ath
the difference is PALU had 550+ days of official content before it got listed on Binance Alpha
Summer has existed for around six weeks and already has:
> 8 official episodes
> 5 comics
> a direct PALU handoff
> the title AI Companion Officer
> an episode specifically explaining how memes become memecoins
i don’t think Binance introduced a new AI companion, handed the role over from PALU, gave her an ongoing series and started teaching meme + Alpha culture through her for no reason. they’re clearly building Summer for the long run
0xeeb3d73d4dd44e6e4d957d89492f05d5ccee7777
been digging a lot more into $SUMMER and imo ppl are reading the Agent OS selloff completely wrong
the Summer story didn’t start yesterday. on july 10, Binance officially introduced Summer as the Binance Academy “AI Companion Officer”. her job is to explain market events, trends and Binance products in simple 1 minute videos
binance.com/zh-CN/square/pos…
since then they’ve already released 7 episodes with her + started a separate Summer comic series
then it got a lot more interesting.
on august 5, @binancezh posted Day 606 of the PALU series with PALU panicking because Summer was shown as the new chief intern:
nitter.cf/binancezh/status/20849…
two days later they posted Day 607:
intern Summer is coming online soon
nitter.cf/binancezh/status/20856…
that’s basically a direct handoff from PALU to Summer. PALU was Binance Chinese’s main character for hundreds of posts. the community token later got listed on Binance Alpha and ran to around 120m
now Binance is introducing a new character inside the same universe, but this time she’s built around AI from day one
and they’re not just mentioning her once
> official AI companion reveal
> 7 educational episodes
> direct PALU handoff
> coming online soon teaser
> 5 Summer comics in 10 days
> constant Binance Square posts
this is the perfect playbook to slowly introduce a character before a bigger rollout
Binance also just launched Agent OS, which gives AI agents access to Binance market data, trading, payments, wallets and onchain tools
the Agent OS release did not mention Summer. Binance currently calls AI Pro its in app assistant, so i’m not going to pretend Summer becoming the main Agent OS interface is confirmed. but the setup is still pretty obv imo
Binance has now released the infra needed to build these agents, while simultaneously introducing its own branded AI character to millions of Chinese users
if they eventually want a consumer facing character sitting on top of that infrastructure, Summer is already there
so the narrative basically connects:
> Binance’s new AI character
> direct successor to PALU
> Binance pushing Agent OS
> Summer already explaining stocks and Binance products
> token rewards holders in SPYB, launched on Flap
> PALU previously ran to 120m + Binance Alpha
my guess is Binance gives Summer a much bigger rollout within the next few weeks, probably late august or early september
i don’t think they go straight into releasing her as a fully interactive trading agent. more likely they expand her role as Binance’s AI companion first, then slowly connect her to more products
Binance said Summer was coming online soon, started posting her much more frequently and then launched Agent OS shortly after
none of that confirms Summer will be connected to Agent OS, but imo it makes a bigger rollout far more likely than her just being a short comic series
but rn the trade is simple:
you’re betting Binance continues turning Summer into its next major character before the rest of the market fully understands what they’re building
0xeeb3d73d4dd44e6e4d957d89492f05d5ccee7777
