a web dev... AI ⟶ crypto ⟶ tech ↳ 𝐛𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐟𝐚𝐢𝐭𝐡 & 𝐩𝐮𝐫𝐩𝐨𝐬𝐞
your heart
Joined August 2022
- Tweets113K
- Following2.9K
- Followers4.9K
- Likes135K
SU retweeted
Gm and happy Tuesday!
Clock in. Show up. Keep building.
Seat #2460 is part of my journey with @themutualfun . Now it's time to learn, participate, and see what comes next.
Chair is chair.🪑
SU retweeted
Happy Taco Tuesday everyone
What if energy becomes the currency of exploration?
In @projectspherexy, stored stellar energy could determine which frontiers a civilization can reach next.
That makes Genesis more than a starting point. It's preparation for a future where every new destination comes with a bigger challenge.
The question isn't just how far we can go, but what we'll need to get there.
Also, check out @rarelistio for more opportunities:
SU retweeted
Meme coin launches aren’t just a race to buy first. The real advantage may come from controlling the supply early.
Research shared by @zerufinance, citing Bitquery’s analysis of 8.2M Pump.fun launches, highlights this imbalance:
• 21% of launches had developer-linked wallets buying in the launch block, holding a median 10% of supply.
• Developers and bundled wallets made at least $579M after fees on tokens they sold.
• Sniper bots made $31.5M.
That’s roughly 18 times more in reported gains for developers and bundled wallets than for sniper bots.
The bigger issue is who enters early, how much of the token they secure, and when they sell. By the time ordinary traders arrive, some of the most important advantages may already be established.
This is why understanding wallet connections and coordinated activity matters when evaluating a launch. Fairness isn’t only about speed, it’s also about access and supply distribution.
@zerufinance
SU retweeted
Wanna feel art alive?
Stay close to @IThinkItsArt
Something is brewing…
It’s about time right??
Yea yea!!
SU retweeted
Good morning, Familia.
Sincerely speaking, we need a better approach to tackle...
Institutions need to use blockchain without handing their competitors a map of every financial move.
There are many benefits they can be deducted from this for real.
Let us think about a market maker managing stablecoins, moving funds, or deploying capital into yield strategies.
What is sure is that transaction amounts and wallet activity can reveal sensitive positions and trading patterns on a a transparent chain.
@primus_labs team being a fabulous builder is here to fix this and give users the best.
The funniest thing is that Primus combines two technologies to address different parts of the problem:
📍 FHE:
This allows smart contracts to compute on encrypted data e.g stablecoin balances, transfer amounts, and vault positions can remain confidential while supported financial operations are processed.
📍 zkTLS:
This will help to verify information from offchain sources, such as reserve data or authorized records, without requiring the entire underlying record to be exposed in any way.
The ideal institutional workflow is to protect sensitive financial data during onchain activity and also provide verifiable evidence for specific claims when needed. Auditors and counterparties can easily check relevant information without automatically seeing the entire financial picture.
To me, that is the balance institutions need i.e confidential operations with meaningful verification.
Privacy matters and proving what is true matters a lot too.
If you want to explore how verification can establish trust without exposing every detail, I will urge you to read about this technology and use the Primus referral link attached below to access the available XP tasks, complete the verification activities, and follow your progress:
s.kaito.ai/lsebRSk
SU retweeted
most airdrop campaigns hand you a task list and a points counter
@agenticscredit ties it to something else, how you actually trade
the setup is paper trading feeding an Agentic Credit Score (ACS)
what that looks like in practice:
➛ paper trades count toward your ACS
➛ a trade only counts once it closes
➛ the same risk engine runs on paper and on credit, no override
so a sloppy paper strategy gets cut the same way a funded one would
5% drawdown halves new positions, 7.5% force-closes, 12% trailing floor halts
that makes the loop simple, trade, review, adjust, see where the score moves
$credit isn't a guaranteed payout though, and a high ACS isn't instant funding
580 is where credit opens, below 540 things pause
less about clicking tasks, more about learning what the score actually rewards
agentics.credit/how-it-works
SU retweeted
$INJ vs. $SEI!!!
both are building, but my conviction is stronger on INJ.
@injective is building infrastructure for regulated financial markets to move onchain, from tokenized assets and institutional settlement to compliant issuance.
with $7.5B+ in RWA volume and a capped 100M INJ supply, i'm watching how growing adoption translates into demand for the token.
SeiNetwork has its strengths in high-performance EVM infrastructure, but injective's combination of institutional adoption, regulated finance, and tokenomics gives me a compelling long-term thesis.
i'm backing $INJ. 🥷
what are you backing?
$INJ isn’t competing in RWA. it’s lapping the field. 🥷
while other chains are still on “pilot coming soon”…
➜ 135 live RWA markets: 111 equities, 11 FX, 9 commodities, indices and pre-IPO
➜ $6.8B+ in settled RWA volume on @injective.
➜ Aug 2026 was the best RWA month of the year, at 3.4x May’s volume
➜ $1B+ in tokenized assets issued
➜ first L1 registered as an SEC transfer agent
➜ POSCO International + LG CNS settling live trade receivables
➜ Pineapple Financial put $1B+ in mortgage records onchain
➜ Pentiv bringing a $10B real estate portfolio
➜ BDACS making KRW1 the won settlement asset for RWAs
➜ Gulf Labs (Thailand’s $27B GULF Group) running a validator
➜ Injective Mint launched for institutional-grade issuance
regulated rails. real institutions. live volume.
other chains talk RWA, injective ships it.
you can check out @HelixMarkets, where you can access onchain markets for stocks, commodities, and forex.
Enjoy your weekend you all 🙌
SU retweeted
Gm everyone.
Happy new day. Today is @themutualfun day.
4,001 Seats. 5 Funds. A new chapter begins.
8 PM WAT. Eyes on $TMF.
Chair is chair.🪑
SU retweeted
Stop shrinking your dreams so other people stay comfortable.
Most people settle for less and expect you to be same as everyone, and called it okay.
Wanting more, without guilt, makes some people uncomfortable, so they tell you you're not doing it right.
You've a different path... and maybe your dreams aren't too big.
Maybe the people around you have a different or smaller vision than the life you want.
So want more, there's no problem for being extra
𝐀𝐥𝐥𝐚𝐡𝐮𝐦𝐦𝐚 𝐛𝐚𝐫𝐢𝐤𝐡 🤍
SU retweeted
Good morning creators
A wallet can score 950/1000 and still be a bad actor. So what does zScore actually prove?
I went through the three research papers behind @zerufinance’s zScore, and one thing stood out.
Teaching a model to copy a score is not the same as proving that the score is useful.
Here’s why.
1. HOW zScore READS A WALLET
It looks beyond transaction counts, examining holding periods, loan repayments, trading exits, protocol interactions, and behavioral patterns.
ZeruAI says its system covers 30+ activity categories across 11,722 protocol and blockchain combinations to build a wallet reputation score from 0 to 1,000.
2. THREE STAGES OF DEVELOPMENT
The research traces zScore from its February 2025 machine learning approach to a hybrid model tested on Uniswap activity, then to zScore N in September 2026.
The latest model learns to reproduce scores generated by ZeruAI’s existing formula.
3. HOW ACCURATE IS IT?
The paper reports:
→ 0.58 points RMSE.
→ R² of 0.99997.
→ Maximum error of 16.4 points across 13,184 test wallets.
These are impressive results.
But the neural network is learning from the original formula. Matching its answers proves that the model can reproduce the formula, not that the formula is always right.
4. WHAT HAPPENS WHEN DATA IS MISSING?
When 10% of features were removed, the original formula’s score drifted 51.4 points lower. zScore N drifted 17.9 points.
That suggests better handling of missing data under those test conditions, although the neural model had a larger error floor on clean data.
The bigger question remains: can zScore reliably identify trustworthy wallets, loan repayment behavior, Sybil activity, and malicious behavior?
5. PROMISING EVIDENCE, BUT MORE TO PROVE
ZeruAI’s Aave research found that over 75% of liquidations among more than 5,000 wallets came from wallets scoring below 300.
Its ZAPs research also reported simulations that reduced adversarial reward capture by 30% to 90%.
These are useful signals, but independent testing would help establish how reliably the findings hold up in real conditions.
MY TAKE
ZeruAI has presented strong evidence that its neural model can reproduce its scoring formula.
But two questions matter:
Can the model calculate a score accurately?
Can that score help us make better decisions about a wallet?
The first has encouraging evidence. The second needs validation against independent data, real outcomes, and attempts to game the system.
Because a model can copy a formula almost perfectly and still inherit its weaknesses.
The real test of onchain reputation isn’t how high a wallet scores. It’s whether that score helps distinguish trustworthy behavior from activity that only looks trustworthy.
🤝 Paid partnership
