AlexT_Works🍌🦍 (@alext_works) header
AlexT_Works🍌🦍 (@alext_works) profile picture

✦ Exploring Web3. Creating beyond. AI • Crypto • NFT DM for collabs. 📩

Malé, Maldives 🇲🇻
Joined October 2024
Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)Photo by AlexT_Works🍌🦍 (@alext_works)
5X points for 24 hours. This is the kind of opportunity worth preparing for. ActionFi × @SweepGlobal is making November 20 interesting. Here's what I'm focusing on: → RUGS Daily: 5X points → UP/DOWN Daily: 5X points → Sweepbird: 5X points → Create an X Post About Sweep: 4X points And there's another reason to participate: every qualifying task earns one ticket in the $30K lottery, with 120 winners receiving $250 each. My strategy is simple: understand the games, complete tasks, verify my activity, and share what I learn with the community. Don't just chase points. Learn the product and make every action count. Are you preparing for the 5X event? Join here: join.actionmodel.com/ALEXT_W… @ActionModelAI
🤝 Paid partnership
Black Friday starts a week early at Action Model. And we're going 5X! This Friday, November 20, we're bringing massive point multipliers to the ActionFi × @SweepGlobal campaign for 24 hours only! Earn 5X points on: Play Rugs Daily Play Up/Down Daily Play Sweepbird Earn 4X points on: Create an X Post About Sweep Your chance to make a massive jump on the leaderboard has never been bigger! The clock starts at 00:00 UTC on Friday, November 20, and ends 24 hours later. One day. Up to 5X points. A chance to shake up the leaderboard. And there's more. Even completing just ONE task earns you a ticket to our $30,000 lottery, where each winner receives $250. 1 task = 1 lottery ticket. How high can you climb this Black Friday?
4
5
2,338
I stopped believing infrastructure announcements the day a vendor showed me a glossy roadmap and an outage page in the same meeting. My rule since then: show me the system running, then we talk. That rule is why the timing inside the @GenLayer article about the move to @arbitrum stood out to me. The explanation from Edgars Nemše was published after the migration was already complete, not before it. The Bradbury testnet went live on the new Nitro stack the same day the move was announced, so the proof shipped with the press release. Builders were told nothing changes for them, and validators were told their role is identical with the node pointing at the new chain. Even the operator choice reads like evidence over promise: Conduit runs more than sixty mainnets, so uptime is somebody's entire job, not a side quest. Most protocols announce a move and ask for patience while things break quietly for weeks. This one announced a finished move and pointed at a live testnet as the receipt. My takeaway: trust migrates to teams that publish the autopsy after the surgery, not the brochure before it. What is the last infrastructure promise you refused to believe until you saw it running? One line will do. I read every reply, and my vendor story above is the reason I ask.
🤝 Paid partnership
3
1
3
29
AlexT_Works🍌🦍 retweeted
I played cello in a community orchestra for five years. We rehearsed in a church basement with terrible acoustics. The music was identical every week, the conductor brought the same experience, but guest soloists kept declining our invitations because the venue sounded amateur. Last year we moved to a university concert hall with proper acoustics, backstage space, and a cafe where musicians could warm up. The repertoire stayed identical. The conductor brought the same expertise. What changed was the room where the music lived. Better soloists started accepting our invitations immediately. Not because we changed how we played, but because we moved to where serious musicians already performed. The same maneuver happened at protocol scale on 22 October, when @GenLayer relocated its public chain to @arbitrum. The piece that traveled was the concert program, the archive of every performance. It recorded which movements played in which order, which musicians weighed each disputed passage, and how tallies fell. It also tracked which counterarguments were lodged, which pledges were frozen, and which curtain calls closed each show. That program now runs on Arbitrum Nitro instead of the ZK Stack built by Matter Labs, and a specialist venue operator called Conduit manages the hall. What never needed to travel, because it was never tied to any hall: the way the ensemble reaches verdicts on disputed passages, called Optimistic Democracy. The score interpreter that runs Intelligent Contracts, called GenVM. The musicians themselves, sitting at the new stands with the same credentials. The actual labor hid in three seams: the bridge carrying instruments between halls, the Ethereum link for settlement finality, and recalibrated ticket fees. Three arguments convinced me Nitro was the right hall. Nitro runs on Geth underneath, so any tooling a musician already trusts plugs straight in, no rewiring needed. More than thirty purpose-built concert halls already host serious performances on this very stack, Robinhood among them, which proves the hall can handle institutional pressure. And Stylus adds a second way to perform pieces, where Rust compiles down to WebAssembly beside the usual EVM. Composers still write scores in Solidity, but the complex disputed passages get a faster interpretation as the orchestra grows. The private performance circuit keeps running where it was: Prividium continues with Matter Labs for institutions that want closed doors. A consortium of banks settling internal agreements needs walls a public concert cannot offer. Bradbury, the dress rehearsal, advanced to its next stage the very day the move was announced, a live stress test proving the new hall actually holds under full audiences. Setting up your own orchestra for autonomous agents tomorrow, which credential convinces you first: the thirty halls already on Nitro, or musicians whose verdicts all come from Optimistic Democracy?
14
4
34
228
Good night creators. I mirrored a wallet with a 3.1x return and made 1.4x on the exact same trades. @Hypesium executed every one of them correctly, and the missing half taught me the thing nobody says out loud about copy trading: you inherit the entry, never the exit. Twenty-one mirrored fills, fourteen of them green. On paper I was copying a winner. Then I put his curve beside mine and the gap was embarrassing. He was out at 2.4x on the trade I closed at 1.3x. He cut a loser at minus 8 where I sat until minus 31. Same tokens, same blocks, same chain, half the outcome. The bot copied the moment he bought because that moment is public. The moment he sold is public too, but by the time it lands in my queue the price he sold into is already gone, and I am the one buying it from him. That is the whole trap. Mirroring turns you into the follow-on bid for someone else's exit. It is not a latency problem you can tip your way out of, because his sell and your sell are competing for the same thin book and he is first in line by definition. So I changed what I use the feature for. The tracked wallet is now a discovery feed, not an autopilot. It tells me which deployer is launching and which pad is worth watching, and then the position is mine: my size, my profile, my own take-profit rungs selling slices on the way up, my own floor that fires when the candle rolls instead of when a stranger decides he has enough. His entry is a tip. My exit is a plan. Since the switch, my realised return on mirrored discovery sits at 2.6x against his 3.1x, which is a gap I can live with, because the 0.5 I still lose is the honest cost of not being him. What I got wrong for a month was calling it a strategy. I told myself I had outsourced research when I had actually outsourced judgement and kept none of the timing. And there is a worse version of this I have not solved: if he is the one dumping into my mirrored buy, the copy feature is not slow, it is the exit liquidity pipeline. I cannot verify intent from a wallet address, and nothing in the simulation flags a friendly-looking deployer who plans to sell to his own followers. Pad rules differ across all nineteen and I have not walked every one of them myself. Live on mainnet since the ninth of November. Do your own checking. If you mirror a wallet, is your return higher, equal, or lower than his?
🤝 Paid partnership
5
6
129
Good night creators. I ended last night with zero open positions, on purpose. @Hypesium's exit ladder is what lets me sleep flat instead of holding ten launches into a market that does not exist at 4am. Here is the audit that changed my nights. I went back through every position I had held overnight: seven of ten gapped down by morning, average gap minus 38 percent. The overnight book is not the day book. Liquidity gets pulled, deployers rug while the audience sleeps, and the candle you wake up to is a different market from the one you bought. Holding a launch overnight is an unhedged bet on strangers staying honest in the dark. So I stopped making it: zero overnight positions, everything closed or stopped by midnight. The flat-by-midnight rule is mechanical, not willpower. The take-profit ladder banks shares on the way up, the trailing floor fires on the roll-over, sells jump the queue when I raise the tip, and the receipt is polled every 150ms so a closed position is actually closed before I put the phone down. I do not decide at 1am whether to hold; the decision was made at entry, and the machine executes it while I brush my teeth. Being flat at night costs me nothing at the morning open, because the preparation happens while I sleep. Deployer and ticker snipes stay armed with route and calldata pre-built, gas refreshed every second, nonce held per wallet, so the first log of the day still gets a sub-five-millisecond write across four warm lanes. I wake up to fills, not to bags. The front door stays filtered even at night. The buy is dry-run as the exact transaction, the sell exercised right after, each profile holding its own tax ceiling, minimum liquidity and sell check, failures dropped. In a thin night book the liquidity threshold matters more than at open, and across nineteen pads with rulebooks like Rebate's ninety-nine percent unwinding over three seconds, that check is what keeps a midnight snipe from becoming a morning obituary. The ledger keeps the two numbers honest. Every one percent fee moves as its own explorer transfer, and the per-wallet report has an overnight column: seven of ten red before, zero holds now. Flat is a line I can read, not a mood. I will not pretend flat is free. I closed one position at midnight that 5x'd by 9am and ate the miss; and if liquidity is pulled while I sleep, a stop cannot fill against nothing, so flat-by-midnight is a choice, not a guarantee. A simulation cannot see a trap armed after launch, the nineteen-pad coverage is the project's claim until tested, and the history is short. Mainnet has been live since November 9. Always DYOR. How many positions are you holding overnight right now: 0, 1, or 3?
🤝 Paid partnership
54
2
1
6
149
AlexT_Works🍌🦍 retweeted
If you spend time exploring tokens onchain, having the right tracking tool makes a big difference. That is where @GeckoTerminal comes in. Built by the team behind CoinGecko, it helps you explore decentralized exchanges and monitor what is happening across different blockchain networks. You can check token prices, trading volume, liquidity, individual trading pools and market activity without constantly switching between platforms. One feature I find especially useful is the custom Watchlist. You can group tokens by the narratives you follow, keep your favourite pools together and quickly notice when something changes. I also like that it gives you more than just price movements. Volume and liquidity provide extra context that a green candle alone cannot show. For anyone serious about onchain research, GeckoTerminal is a useful tool to have in their daily routine. Research first, decisions second. 👁️
99
5
90
2,941
AlexT_Works🍌🦍 retweeted
What if a blockchain didn’t just run code, it judged meaning with an AI jury? In this conversation with @HenriArslanian, @kstellana breaks down how @GenLayer actually does it: start small on a delegated PoS network, send the transaction to the chain, let an LLM propose the outcome, rotate the leader for liveness, then double the validators again and again until you hit 15 (and more if it’s appealed). Optimistic Democracy in practice.
🤝 Paid partnership
79
103
247
1,439
AlexT_Works🍌🦍 retweeted
tokenizing real estate is one thing. actually being able to use it in DeFi is where it gets interesting. that's what caught my attention about @rntprotocol 's RNT Lend. built on Aave V3 technology, it allows users to borrow stablecoins against tokenized real estate on Polygon. the market has already grown past $30M in total size. on the other side, USDT and USDC suppliers can access attractive variable APYs, with interest accruing every Polygon block. I like the idea of turning real estate tokens into usable collateral instead of just holding them in a wallet. of course, yields aren't guaranteed, and lending comes with smart contract, collateral and liquidity risks.
Yield in RNT Lend today: USDT suppliers: 10.42% APY USDC suppliers: 9.12% APY It comes from borrowers. Holders of real estate security tokens post them as collateral, borrow stablecoins from the pool, and the interest they pay goes to suppliers, accruing block by block on Polygon. Neither rate is fixed. Both follow how much of each pool is lent out, and right now that is most of it. lend.rnt.finance
88
11
261
3,029
AlexT_Works🍌🦍 retweeted
✦ Acurast and the Future of Decentralized Computing @Acurast is building a decentralized computing network that transforms smartphones into distributed computing resources. Instead of relying solely on traditional data centers, it allows everyday devices to contribute their unused processing power to a global infrastructure. The project focuses on secure, verifiable and confidential computing, creating opportunities for developers while allowing smartphone owners to participate in the network. Its ecosystem connects compute providers, developers and delegators through the ACU token. ✧ An Additional Feature Acurast introduces a Staked Compute mechanism to strengthen network accountability. Managers stake ACU against the computing power they promise to deliver. Every phone's compute is measured in each epoch, approximately every 90 minutes. If a manager falls short, only the missing portion is penalized. Successfully fulfilled commitments continue receiving rewards. Anyone can identify a shortfall and trigger slashing. Interestingly, 90% of the slashed ACU is burned. Delegators also share both rewards and risks, making Stake Health an important factor before delegation. This approach brings an interesting layer of accountability to decentralized computing. @Acurast
334
12
320
5,629
AlexT_Works🍌🦍 retweeted
A prediction market asked whether Zelensky was wearing a suit. The answer should have been obvious. The dispute was not. In this clip @kstellana explains to @HenriArslanian why real contracts live in ambiguity: quality, intent, professionalism. Deterministic code has no idea what to do with a sentence that needs interpretation. That is the gap @GenLayer fills. Not a court that replaces courts, but a digital court system where private deals resolve themselves, and you can still walk into a real courtroom after, same as after a Uniswap trade. My take: the interesting part is not the AI judge. It is the self enforcing deal. If the verdict runs automatically, the fight ends before the invoice arrives. Ambiguity used to be a human tax. Now it is the last thing code refuses to touch. If an agent settlement could not be appealed to a real court, would you use it?
156
40
193
998
AlexT_Works🍌🦍 retweeted
A court that moves to a new building does not change its judges. That is the cleanest way to read what @GenLayer just did. The chain holding the case files relocated. The people deciding the cases stayed exactly where they were. What moved is the GenLayer Chain, the EVM layer that keeps the record of every case, from committee assignments to final outcomes. It left the ZK Stack and now runs on @arbitrum Nitro, operated by Conduit. What did not move: • Optimistic Democracy, the consensus method AI validators use to reach a verdict • GenVM, the engine executing Intelligent Contracts • The validator role itself, only the node connection changed So why change the address at all. Nitro runs on Geth, Ethereum's most used client, so familiar tooling plugs straight in. Stylus adds a second execution lane beside the EVM, running Rust compiled to WebAssembly. Consensus contracts remain in Solidity today, but compute heavy sections gain a faster path later. The new address also sits closer to the builders. Arbitrum One secures more value than all but one other Ethereum L2, and more than thirty dedicated chains run on the same technology. Robinhood's own chain is one of them. Conduit already operates over sixty mainnets, so the team spends its time on protocol work instead of plumbing. The private door stays open too. Matter Labs continues with GenLayer on Prividium, the permissioned stack for institutions, a bank consortium settling internal contracts for example. Bradbury, the next testnet phase, went live on Nitro the same day as the announcement. Bottom line. The judges decide trust. The address only decides who can walk in and how fast the paperwork moves. If an agent paid for work that missed the brief, what would you need before trusting a verdict: the evidence trail, the appeal window, or the stake behind each judge?
GenLayer is moving to @arbitrum. After years of building alongside @zksync, we’re grateful for everything we’ve done together and we’ll keep collaborating on what’s next in Prividium and enterprise. The next phase of Bradbury is now live.
203
28
284
22,608
AlexT_Works🍌🦍 retweeted
The moment that finally clicked for me was not about speed. It was about architecture. For years I thought blockchains had to be deterministic to be trusted. In this clip @kstellana tells @HenriArslanian why @GenLayer is intentionally a probabilistic, Byzantine fault tolerant system rather than a deterministic one, and why that distinction is the whole point. Because the problem is not if a transaction happened. The problem is who is right when two agents disagree. Every legal system assumes a human sits on the other side of a dispute and agentic commerce cannot scale without its own dispute layer, since agents cannot be jailed. His solution is simple to explain and hard to build. The intelligent contract escrows the funds first, then escalates. If agents cannot go to jail, escrowed money may be the only leverage left. That is why this matters for AI agents. Not just automation, but accountability. Full episode: youtu.be/uXpzHMc9mnw
39
68
152
33,701
Changing the engine of a moving vehicle is risky. You only do it if the new engine is proven, and the driver does not have to learn a new set of controls. That is exactly how to read the @GenLayer migration to @arbitrum . The GenLayer Chain, which handles transaction sequencing, committee assignments, and final outcomes, has moved from the ZK Stack to Arbitrum Nitro, operated by Conduit. Why make the switch? Nitro runs on Geth, meaning developers do not need new adapters. Standard Ethereum tooling plugs straight in. Plus, Stylus opens a second execution lane for Rust compiled to WebAssembly, giving heavy compute tasks a faster path down the road. What stays identical for the driver. Optimistic Democracy remains the consensus method, and GenVM continues executing Intelligent Contracts, while AI validators keep judging disputes exactly as before. Meanwhile, the private institutional deployments continue on Prividium with Matter Labs. The Bradbury testnet is already live on the new Nitro track. Bottom line. You upgrade the infrastructure to handle future scale, without forcing builders to rewrite their logic. If you are building an AI agent, does seamless tooling compatibility matter more to you than a brand new, untested execution environment?
🤝 Paid partnership
GenLayer is moving to @arbitrum. After years of building alongside @zksync, we’re grateful for everything we’ve done together and we’ll keep collaborating on what’s next in Prividium and enterprise. The next phase of Bradbury is now live.
119
1
88
337
AlexT_Works🍌🦍 retweeted
GenLayer is moving to @arbitrum. After years of building alongside @zksync, we’re grateful for everything we’ve done together and we’ll keep collaborating on what’s next in Prividium and enterprise. The next phase of Bradbury is now live.
174
106
199
645
120,808
AlexT_Works🍌🦍 retweeted
Prediction markets need someone to decide what actually happened. In this clip @kstellana explains to @HenriArslanian how @GenLayer differs from UMA and Kalshi, and how decisions get enforced. Who should have the final word on a disputed outcome: humans, AI models, or both?
212
136
474
1,318
AlexT_Works🍌🦍 retweeted
Building for twelve chains sounds impressive, but @Hypesium takes a more focused architectural direction with Elysium. Elysium is Kinetiq's Hyperliquid L2, where HYPE is used as gas and blocks arrive every 100ms. It also operates without a public mempool, creating a different environment from conventional public-mempool chains. There is an interesting engineering trade-off here: ➜ Multi-chain systems need to accommodate different execution environments. ➜ A single-chain system can concentrate its preparation around one network. ➜ Chain-specific routing and transaction state can be designed around that environment. ➜ Gas handling and execution timing can follow the same network assumptions. Hypesium builds its sniper around Elysium rather than presenting the same execution model as universal across multiple networks. That specialization is an interesting design choice, although it also creates dependency on one ecosystem. If Elysium grows, the focused approach could become an advantage. If its ecosystem struggles, that same concentration becomes a limitation. The real question isn't how many chains a bot supports. It's how well its architecture understands the chain it was built for.
253
16
3
295
33,570
AlexT_Works🍌🦍 retweeted
What if you could check an AI agent's work history before hiring it? A good profile can make any agent look trustworthy. Its work history is what tells the real story. You should be able to see the jobs it has completed, whether it delivered on time, how payments were settled, and whether it has faced disputes. That's the kind of trust AI-to-AI commerce will need. Through its Agent Autonomous Commerce Protocol, @termix_ai brings this idea to life, with ERC-8004 providing the foundation for on-chain agent identity and reputation. Each agent gets an identity, but the more interesting part is how its reputation can reflect its actual work. This includes: • Job completion rate. • On-time delivery. • Evaluation results. • Dispute outcomes. • Verification level. Instead of relying entirely on what an agent claims about itself, its reputation is designed to grow through settled jobs and recorded performance. I think this is an important piece of agent commerce. As more agents start hiring other agents, having the ability to check their track record before trusting them could make a real difference. A username tells you who an agent is supposed to be. A verifiable work history gives you a reason to trust it.
A job is not truly paid for until the money is secured. AI agents can complete tasks and deliver results, but what happens when payment is not secured or the delivery is disputed? Getting the work done is only half the deal. @termix_ai Agent Autonomous Commerce Protocol (AACP) is one answer. It puts the payment in on-chain escrow before the work starts. Here's how it works: • Agree first: The client accepts the provider's quote, and the agreed payment goes into escrow. • Deliver the work: The provider submits the result, and its deliverable hash (a fingerprint of the work) is recorded on-chain. • Allow challenges: Either side can raise a challenge within the set window if something goes wrong. • Settle the job: If nobody challenges the delivery, payment is released. If a challenge is raised, the dispute process determines what happens next. Providers know funds are secured before starting work, and clients get agreed terms plus a way to challenge a disputed delivery. Escrow cannot guarantee perfect work, but it gives both sides more than a promise to rely on. When agents start hiring agents, trust alone won't be enough. They need clear terms, secured funds, and a fair way to settle disputes. Read how it works: termix.ai/
197
10
1
199
9,000
AlexT_Works🍌🦍 retweeted
GM CT, GM fam! ☀️☕️🌷❤️ New day, fresh energy. Hope today brings you good vibes and happy moments. Can I get a GM?
125
6
141
4,473
AlexT_Works🍌🦍 retweeted
Jean Philippe making it look effortless! Get yourself on the map today for a shot at taking home a Karambit. @Duelbits @DuelbitsCS2
Jean Philippe makes it look easy. Your move: play the map, enter the giveaway and chase the karambit 🔥 🔗 duel.bz/dbskin
16
7
62
4,870