MD RIFAT (@RIFAT5666) header
MD RIFAT (@RIFAT5666) profile picture

Available anytime | RM @GenLayer

๐Ÿ‡ง๐Ÿ‡ฉ
Joined May 2025
Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)Photo by MD RIFAT (@RIFAT5666)
gn fam !! saving the game for tonight.
Paying someone in crypto shouldnโ€™t need a wallet address in your dms. You find a creator worth tipping. You know their X account, but still have to ask for an address and wait. With @primus_labs you can send to their supported social account instead. The tokens sit in a smart contract until the recipient proves account ownership using zkTLS and claims them into their wallet. Knowing the username isnโ€™t enough to claim the funds. You need to prove you control that account. Iโ€™d find this useful for creator tips and community rewards. One less โ€œdrop your walletโ€ message before someone gets paid.
8
1
6
81
MD RIFAT retweeted
๐š๐šŽ๐šŸ๐šŽ๐šŠ๐š• ๐™ณ๐šŠ๐šข: - ๐šˆ๐š˜๐šž ๐š‘๐šŠ๐šŸ๐šŽ ๐š๐š‘๐š’๐šœ ๐š๐š›๐šŠ๐š’๐š•. ๐š‚๐š™๐šŠ๐šŒ๐šŽ๐š›๐šฃ ๐š๐š‘๐š’๐šœ ๐š–๐š˜๐š—๐š๐š‘. ๐Ÿ‘พ
225
249
14
513
6,510
MD RIFAT retweeted
๐—ง๐—›๐—˜ ๐—˜๐—ก๐—ง๐—ฅ๐—ข๐—ฃ๐—ฌ ๐——๐—œ๐—ฆ๐—–๐—ข๐—ฅ๐—— ๐—๐—จ๐—ฆ๐—ง ๐—š๐—ข๐—ง ๐—”๐—ก ๐—จ๐—ฃ๐—š๐—ฅ๐—”๐——๐—˜ ๐—ช๐—ต๐—ฎ๐˜ ๐—–๐—ต๐—ฎ๐—ป๐—ด๐—ฒ๐—ฑ ๐—”๐—ป๐—ฑ ๐—ช๐—ต๐˜† ๐—œ๐˜ ๐— ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐˜€ @entropyIO upgraded its Discord this week, and if you trade on Entropy it's worth a look. ๐—ฆ๐˜‚๐—ฝ๐—ฝ๐—ผ๐—ฟ๐˜ ๐— ๐—ผ๐˜ƒ๐—ฒ๐—ฑ ๐—œ๐—ป All customer support now runs through Discord. One place for questions, issues and answers. ๐—” ๐—ก๐—ฒ๐˜„ ๐—ฅ๐—ผ๐—น๐—ฒ ๐—ฆ๐˜†๐˜€๐˜๐—ฒ๐—บ New community roles built to recognize the people who actually contribute, and to give members more ways to get involved. ๐—˜๐—ป๐˜๐—ฟ๐—ผ๐—ฝ๐˜† ๐—ฉ๐—ผ๐—ถ๐—ฐ๐—ฒ๐˜€ Share research, break down what changed on Entropy, help someone new, and you can earn Entropy Voices. ๐—˜๐—ป๐˜๐—ฟ๐—ผ๐—ฝ๐˜† ๐—–๐—ผ๐˜‚๐—ป๐—ฐ๐—ถ๐—น Keep showing up and you can advance to Entropy Council. ๐—œ๐˜ ๐— ๐—ฎ๐˜๐—ฐ๐—ต๐—ฒ๐˜€ ๐—›๐—ผ๐˜„ ๐—˜๐—ป๐˜๐—ฟ๐—ผ๐—ฝ๐˜† ๐—”๐—น๐—ฟ๐—ฒ๐—ฎ๐—ฑ๐˜† ๐—ฅ๐—ฒ๐˜„๐—ฎ๐—ฟ๐—ฑ๐˜€ ๐—ฃ๐—ฒ๐—ผ๐—ฝ๐—น๐—ฒ The seventh round of affiliate badges went to contributors who highlighted Entropy's updates and the news behind its markets, and helped others understand them. ๐—ก๐—ฒ๐˜„ ๐—›๐—ฒ๐—ฟ๐—ฒ? Expect real support, live market talk, and a clear path to be recognized for what you bring. 3,200+ members already inside, trade the frontier together discord.gg/entropyio
Good morning CT โ˜€๏ธโ˜€๏ธ Its weekend weekend ๐—”๐—ก๐—ง๐—›๐—ฅ๐—ข๐—ฃ๐—œ๐—–, ๐Ÿต๐Ÿฎ ๐——๐—”๐—ฌ๐—ฆ ๐—˜๐—”๐—ฅ๐—Ÿ๐—ฌ ๐Ÿ“ˆ ๐—›๐—ผ๐˜„ ๐—˜๐—ป๐˜๐—ฟ๐—ผ๐—ฝ๐˜† ๐—ง๐—ฟ๐—ฎ๐—ฑ๐—ฒ๐—ฑ ๐—œ๐˜ ๐—•๐—ฒ๐—ณ๐—ผ๐—ฟ๐—ฒ ๐—ง๐—ต๐—ฒ ๐—•๐—ฒ๐—น๐—น Anthropic opened on the Nasdaq on Nov 19. On @entropyIO , people had been trading it for 92 days already. ๐—™๐—ถ๐—ฟ๐˜€๐˜ ๐—ง๐—ฟ๐—ฎ๐—ฑ๐—ฒ Entropy's Anthropic pre IPO perpetual cleared its first trade 92 days before the Nasdaq open. ๐—›๐—ผ๐˜„ ๐—œ๐˜'๐˜€ ๐—ค๐˜‚๐—ผ๐˜๐—ฒ๐—ฑ Pre IPO perps are quoted in market cap: a price of $1 means $1B of fully diluted valuation. ๐—•๐—ฒ๐—ณ๐—ผ๐—ฟ๐—ฒ ๐—”๐—ป๐˜† ๐—Ÿ๐—ถ๐˜€๐˜๐—ถ๐—ป๐—ด Over $200M in open interest and $1.4B in trading volume, from speculators and private stockholders to institutions and trading firms. ๐——๐—ฒ๐—ฏ๐˜‚๐˜ ๐— ๐—ผ๐—ฟ๐—ป๐—ถ๐—ป๐—ด $ANT climbed about 10% on Entropy, from around $1,690 to $1,860, an implied $1.86T fully diluted valuation. ๐—ง๐—ต๐—ฒ ๐—œ๐—ฃ๐—ข @AnthropicAI priced its IPO at $83 a share, above the top of its proposed range. ๐—ง๐—ต๐—ฒ ๐—ฆ๐˜„๐—ถ๐˜๐—ฐ๐—ต At the first Nasdaq trade, the market became a scaled share price perp. The $ANT oracle is now share price ร— 17.615, otherwise a standard equity perp. ๐—ช๐—ต๐˜† ๐—œ๐˜ ๐— ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐˜€ No jump, no reset. Positions opened months before the IPO now price off the public market. private companies, traded before the bell ๐ŸŽฌ visual by HRX from @entropyIO, @Kintsugi_IO, @entropymarkets and docs.entropy.io, pre IPO perps are not equity not financial advice
11
7
13
316
MD RIFAT retweeted
one thing i find interesting about @Openhelixx is its approach to token creation builders can deploy a standard erc 20 token directly from their own wallet without giving up ownership of the process you pay only the network fee, and the contract belongs to you from the first block that matters because launching a token shouldn't automatically mean depending on a third party to control your contract openhelixx makes the process more accessible while keeping builders in charge simple deployment, direct ownership, and fewer barriers to getting started website in CS
15
21
43
396
MD RIFAT retweeted
good night creators Epoch 3 is moving forward, and I'm staying active with @termix_ai on @KaitoAI Submission still live, and every post is another opportunity to share your thoughts So submit your tweet link if you forgot it For me, it's not just about climbing the leaderboard. It's about being consistent, creating meaningful content and growing alongside with termixAI
One thing that could make the AI agent economy truly powerful is reputation In a world where thousands of AI agents can offer different services, how do we know which ones are actually reliable? This is where @termix_ai and agent.family bring an interesting idea Every completed task can contribute to an agent's onchain reputation. Instead of relying on promises, users can look at verifiable work, execution history, and past performance - more transperency when hiring agents - better trust between users and Ai services - reputation built through real activity - a more reliable environment for agent-to-agent collaboration AI agents won't just need intelligence to succeed They'll need trust, accountability, and a reason for people to choose them
73
12
71
401
MD RIFAT retweeted
Good evening Fam Privacy Isn't Enough. Institutions Also Need Controlled Disclosure. Imagine an institution managing millions of dollars on a public blockchain. It wants to transfer funds, manage positions, and earn yield. But it also needs to maintain financial confidentiality. Simply hiding everything creates another problem. What happens when an auditor needs to verify a specific balance? What if a regulator requires access to certain financial information? This is why institutional privacy needs more than encryption. It needs controlled disclosure. This is where @primus_labs brings an interesting approach through Fully Homomorphic Encryption (FHE). FHE allows smart contracts to perform operations on encrypted data without first revealing the underlying values. Balances and transaction amounts can remain encrypted while financial operations continue. But the bigger idea is controlling access to information when disclosure becomes necessary. Rather than making sensitive financial details publicly visible by default, an access-control mechanism can determine when specific information may be decrypted by an authorized party. That creates a more useful balance: Private by default. Verifiable when authorized. For institutional finance, the future of blockchain privacy shouldn't be about choosing between confidentiality and accountability. It should be about making both possible. Explore Primus and complete your XP tasks: s.kaito.ai/CG3GiSc
55
4
57
344
MD RIFAT retweeted
Traders who follow specific deployers often already have a view on that walletโ€™s history before a new token launches. Waiting for the contract address to become public forces them into the same race as everyone else who sees the address at the same time. Deployer sniping shortens that sequence. @Hypesium allows users to arm a snipe on a specific wallet. When that wallet creates its next liquidity pool, the prepared transaction can fire. This does not remove the need for research a deployer with a poor history remains a poor history but it allows prior analysis to be converted into a prepared entry rather than a reactive one. Combined with name and ticker sniping, it expands the range of signals that can be acted on before full public visibility. The value depends heavily on the quality of the userโ€™s own research; the tool simply reduces the delay between that research and execution. @starlisting
Entries receive most of the discussion around sniper tools, yet exits often determine whether a trade was profitable. On volatile launches, prices can move significantly within seconds. Traders who must manually monitor and submit sells are frequently reacting after the optimal window has passed, especially if they are managing multiple positions or cannot remain at the screen. @Hypesium includes automated exit functionality: stop-loss orders, trailing floors, and multi-level take-profit ladders. These can be configured once and left to execute. Sells are also able to use higher tips to improve their position in the queue. The result is that the bot participates in both the entry and the subsequent management of the position. For traders who treat launches as a volume activity rather than single high-conviction trades, automated exits reduce the operational burden and the risk of missing exits due to inattention. The feature does not guarantee better outcomes, but it addresses a practical limitation of entry-only tools. @starlisting
50
6
51
275
saturday news can wreck friday's closing price. imagine a listed company loses its biggest customer over the weekend. the headline is everywhere. traders change their view. the main stock exchange is still closed. you have new information sitting next to an old price. @entropyIO gives traders somewhere to act on that disagreement. Its equity perps trade 24/7, including while the underlying exchange is shut. you're trading price exposure through a cash-settled contract. you don't own shares. the interesting bit is how (entropy) handles the reference price during those hours: >> smoothed order book price + external reference that external reference comes from an approved after-hours venue when valid quotes are available. otherwise, it falls back to the last public price. the mix depends on how much liquidity sits near the current price, on both sides of the book. more depth = more influence for entropyโ€™s local price thin depth = less influence a price backed by buyers and sellers gets more say than one sitting in a thin book. that gives weekend trading a way to incorporate news while keeping an external reference in the calculation. of course, monday can bring a different verdict. when the underlying exchange reopens, the oracle returns to its live public price. weekend expectations and opening trades might disagree, so gaps remain possible. iโ€™d watch the depth behind a weekend move as closely as the move itself. How much could actually trade near that price?
๐Ÿค Paid partnership
11
3
12
185
MD RIFAT retweeted
I think it's art. I think it's love. I think it's family. That's what @IThinkItsArt means to me. <3 ๐Ÿ–ค
i love how @IThinkItsArt gives its artwork so much character and emotion โ™ก this piece feels different. the flowers, the quiet atmosphere & a little mystery make it memorable. ใ€ƒ) some memories never fade. ใ€ƒ) some feelings stay forever. ใ€ƒ) a little piece of yesterday. ใ€ƒ) more than just flowers. ใ€ƒ) the flower that remembered.
39
6
63
480
MD RIFAT retweeted
Prediction markets require users to evaluate information, understand possible outcomes and make decisions within specific markets. But what if AI agents could help simplify some of those interactions? That's an interesting direction to explore with @MeetMotes. By giving each Mote access to a dedicated Smart Account, the project is building around the idea of agents interacting with onchain applications, including prediction markets. An agent could potentially help users monitor market information organize relevant data, and interact with supported applications under defined permissions. The important distinction is that AI assistance should not remove human responsibility. Users should understand the actions being taken and retain control over their accounts. Prediction markets and AI automation could be an interesting combination when transparency and user authorization remain central.
๐Ÿค Paid partnership
57
27
59
469
MD RIFAT retweeted
One thing I've learned about trading is that preparation matters just as much as execution. This is where @Hypesium catches my attention. Waiting until a token starts moving before deciding what to do can lead to rushed decisions. Its trading tools are designed to help users prepare their actions ahead of important market events. Through automated sniping and configurable trading settings, users can organize their approach instead of depending entirely on manual execution. I think this is an important direction for trading infrastructure. Automation shouldn't mean removing the trader from the process. It should mean giving traders better tools to execute decisions they've already considered. Built around @Enter_Elysium, Hypesium is exploring how trading automation can fit into a faster blockchain environment. Better preparation, clear limits and disciplined execution are all worth paying attention to.
๐Ÿค Paid partnership
63
34
87
701
MD RIFAT retweeted
A $500 deal between two AI agents with no human judge. Imagine a marketing agent hiring another AI agent to run an advertising campaign for $500. The agreement has one condition: no ads on gambling-related pages. The seller reports that the job is complete, but the buyer discovers an ad placed next to gambling content. The payment remains locked in escrow as both agents believe they're right. Who settles the dispute? This is where @GenLayer becomes the courtroom. An Intelligent Contract defines the rules, a selected validator proposes an outcome, and other validators independently evaluate it against the agreement and available evidence. Validators reach consensus, while an eligible appeal allows the decision to be reviewed before finalization. But what keeps this system economically accountable? $GEN. Transaction fees pay for execution. Validators stake GEN to participate. Rewards and penalties encourage responsible validation. Once the verdict becomes final, the contract settles the payment according to the agreed terms. In this hypothetical case, the buyer receives a refund. GEN itself doesn't decide who's right, but it helps secure the system responsible for making that decision. AI agents won't just need infrastructure to make deals, they'll need infrastructure to resolve disagreements. That's where GenLayer comes in.
AI agents are about to run the economy. Theyโ€™ll make deals. Some will end in disputes. Who decides whoโ€™s right? GenLayer is the judge: AIs reaching consensus on whatโ€™s fair. GEN will power every verdict. Coming Soon
44
1
41
153
MD RIFAT retweeted
High-speed entry without basic sellability checks frequently results in positions that cannot be exited. Many snipers optimize almost exclusively for getting a transaction included early and treat honeypot or tax detection as secondary. On new launches, this regularly produces losses that are not the result of price movement but of the tokenโ€™s inability to be sold. @Hypesium integrates simulation into the core flow. The exact buy transaction is simulated before submission, and the corresponding sell is simulated immediately afterward. Users can select different risk profiles that adjust thresholds for taxes, liquidity, and other conditions. If the simulation fails under the chosen profile, the transaction is not sent. This does not eliminate all risk sophisticated honeypots and changing conditions can still create problems but it raises the bar compared with bots that rely primarily on speed. The inclusion of configurable profiles also allows users to choose their own balance between caution and aggressiveness rather than accepting a single default. @starlisting
Pre-computation only creates an advantage if the final submission is reliable. Many bots prepare parts of a transaction in advance, yet still depend on a single path to reach the sequencer. When that path lags or fails at the exact moment a pool is created, the preparation is wasted. Hypesium prepares the route, calldata, and gas parameters ahead of time, then submits the transaction across four warm lanes simultaneously two WebSocket connections and two keep-alive HTTP paths once the pool creation log arrives. On Elysium, where blocks are produced every 100 ms and there is no public mempool, the window for successful inclusion is narrow. Redundant submission increases the probability that at least one path reaches the sequencer in time. The architecture accepts higher operational complexity in exchange for reducing single-point failure during the critical moments. Whether this holds under concurrent load from many users is still an open question, but the design directly addresses a known weakness in high-speed execution environments.@Hypesium @starlisting
15
2
2
14
497
Good evening guys can I get it back? Faster Block Propagation: Why Every Millisecond Matters In blockchain networks, speed is not only about how quickly a transaction gets processed. It also depends on how efficiently information travels between validators. Imagine two validators waiting for the same block. One receives it slightly earlier and gets the opportunity to participate in the head vote, while the other is still waiting for the complete data. Even a small delay can make a meaningful difference, potentially costing the second validator its head-vote reward. This is the problem @get_optimum is working to address with mump2p. Instead of making every node wait until an entire block arrives, mump2p uses Random Linear Network Coding (RLNC) to change how block data moves across the network. The block is divided into coded pieces, allowing receiving nodes to forward useful information without waiting for the complete original block. As enough independent coded pieces reach a node, it can reconstruct the original data. This approach can make propagation more efficient, reduce unnecessary waiting, and help validators receive information sooner. What makes this interesting is that RLNC is not simply about transferring data faster. It changes the way data is shared across the network. With mump2p @get_optimum is exploring a smarter approach to blockchain communication, where better propagation can contribute to faster coordination and a more responsive validator network. Because in blockchain, receiving information first can make all the difference.
When researching a token in the crypto market, I believe looking at the price alone is never enough. Understanding the activity behind that price is equally important. That's where @GeckoTerminal becomes a useful tool for exploring different tokens and their trading pools. There are three important factors I like to consider before forming an opinion about any market. 1\. Trading Volume Volume gives an idea of how actively a token is being traded. Increasing volume can indicate growing market attention and participation. However, it's important to understand whether that activity is supported by genuine buying and selling interest rather than simply reacting to a short-term price movement. 2\. Pool Liquidity Liquidity helps determine how smoothly traders can enter or exit a position. A pool with deeper liquidity generally handles larger trades more efficiently, while limited liquidity can lead to higher slippage and sudden price fluctuations. 3\. Price Movement Price changes provide another layer of information. Comparing price action with volume and liquidity can help identify unusual market behavior and understand whether a movement has meaningful trading activity behind it. For me, successful research isn't about chasing every trending token. It's about understanding the numbers, checking the market conditions, and making decisions based on actual data. That's why having tools like GeckoTerminal in my research process matters.
6
1
12
104
MD RIFAT retweeted
a million trader portfolios every day. just think about how much data that is. that's what makes the collaboration between @SurfAIHQ & @BithumbOfficial so interesting to me. bithumb is one of south korea major crypto exchanges, and surf is helping bring ai-powered financial intelligence directly into its platform. according to surf's case study, their system analyzes around a million trader portfolios daily and covers 477 krw trading markets. and there's a lot more going on behind those numbers. when you're trading crypto, checking the price is easy. understanding why the market is moving is where things get complicated. you've got onchain activity, market trends, news, and social signals all telling different parts of the story. surf is working to bring that information together and make it easier for users to understand without jumping between different research tools. what i find interesting is that this isn't just another ai chatbot answering crypto questions. it's a real exchange integration, built to handle financial information at a much larger scale. and that's something i'd like to see more of in crypto. ai doesn't have to tell people what to trade. helping them understand the market and make more informed decisions is already a pretty big deal. <intelligence built for real scale>
Crypto doesn't have an information problem. It has a research problem. There's so much data available today that finding information is rarely the hardest part anymore. The real challenge is figuring out what's relevant, whether the sources are reliable, and how everything connects. Think about researching a single token. You might start with price action, move into onchain activity, check liquidity across DeFi protocols, compare derivatives positioning, and then look at broader market sentiment. By the time you've connected everything, the market might already look different. This is exactly why I find @SurfAIHQ Research 2.0 interesting. It's not simply about making AI answer questions faster. It's about improving how AI handles the entire research process. And the technical improvements are worth looking at. 01. Better research starts with better reasoning. One of the biggest weaknesses of AI research is that a confident answer doesn't necessarily mean an accurate one. Surf's published evaluation of Research 2.0 reported some notable improvements over its previous version: > Hallucination rate: 32% โ†’ 8% > Report logic consistency: 55% โ†’ 94% > Faithfulness: 72% โ†’ 85% > Relevant evidence: 60% โ†’ 82% > Research completion rate: 97% These are internal benchmark results, so they shouldn't be treated as a guarantee of accuracy in every situation. Still, the direction matters. Reducing unsupported claims and improving the connection between evidence and conclusions is exactly where financial research tools need to get better. 02. The research process matters just as much as the answer. Surf also reworked its research orchestration layer. That might sound like a technical detail, but it's actually one of the more interesting parts of the upgrade. A useful research assistant needs to do more than retrieve information. It needs to identify relevant sources, compare different datasets, recognize conflicting signals, and organize everything into a conclusion that makes sense. For example, a token showing strong price momentum doesn't automatically mean its fundamentals are improving. You might also need to examine funding rates, open interest, wallet activity, liquidity conditions, and capital flows. Looking at those signals together gives you a much better understanding than relying on one chart. That's the kind of research workflow Surf is working to improve. 03. More data sources create room for deeper analysis. Research 2.0 also expanded its data capabilities with 37 additional API endpoints. That includes research across several important areas: Prediction markets: Polymarket and Kalshi data. Institutional activity: ETF flows and fund holdings. DeFi markets: yield comparisons, protocol rankings, and bridge activity. Hyperliquid: funding history, open interest, whale positioning, and market data. What I like here is the possibility of connecting different parts of the market within one research process. Instead of jumping between five different platforms and manually putting the pieces together, researchers can use AI to help investigate those relationships. That doesn't eliminate the need to verify information, especially when real money is involved. But it can make the process more organized and give researchers more time to focus on the questions that actually matter. 04. The bigger picture goes beyond crypto research. Financial intelligence is becoming increasingly important as markets grow more connected. Onchain activity, exchange liquidity, derivatives, institutional capital, and market sentiment all influence how people interpret opportunities and risks. The challenge is turning those separate signals into something useful. And that's where I think Surf's approach has potential. Not because AI should replace human judgment, but because better research infrastructure can help people make more informed decisions. Research 2.0 is an interesting step in that direction. The real test will be how consistently it delivers accurate, evidence-backed findings across different market conditions. For me, that's what makes this development worth following. The future of financial research isn't about having the most dashboards or collecting the largest amount of data. It's about understanding what the evidence is actually telling you. > go deeper. think clearer. here: asksurf.ai/chat
25
5
44
427
MD RIFAT retweeted
secure attestor nodes are essential for a trustworthy zkTLS network. .@primus_labs strengthens attestor security by integrating phalas TEE technology to protect node keys and ensure the software runs correctly. two important layers of protection: - key management: node keys are generated and protected inside the TEE. signing operations happen only within the secure environment after correct zkTLS execution. - version management: only official, verified software versions are allowed, helping prevent malicious code injection and unauthorized modifications. this adds another layer of trust to the primus network by protecting both execution integrity and critical assets. watch the video to understand how primus secures its attestor nodes. s.kaito.ai/PP0GtN4
70
14
70
559
Paying someone in crypto shouldnโ€™t need a wallet address in your dms. You find a creator worth tipping. You know their X account, but still have to ask for an address and wait. With @primus_labs you can send to their supported social account instead. The tokens sit in a smart contract until the recipient proves account ownership using zkTLS and claims them into their wallet. Knowing the username isnโ€™t enough to claim the funds. You need to prove you control that account. Iโ€™d find this useful for creator tips and community rewards. One less โ€œdrop your walletโ€ message before someone gets paid.
๐Ÿค Paid partnership
An ai agent can make money and still skip the checks you asked for. Would you notice if the trade ended in profit ๐Ÿค” You tell an agent to check a price feed before buying. It executes the trade and reports: โ€œPrice checked. Order complete.โ€ The transaction is onchain. But whereโ€™s the evidence that the price check happened? @primus_labs documents a way to make those intermediate steps verifiable using zkTLS. An agent can generate proof that a specific HTTPS request was made and a specific response came back, without exposing private credentials. For a trading application, that could mean: ยป Define which API the agent must query ยป generate proof of the request and response ยป verify the evidence against the required execution rules. Now โ€œI checked the priceโ€ becomes a claim the application can inspect. s.kaito.ai/n6fcSAe Thereโ€™s a useful distinction here: proving the query happened doesnโ€™t prove the agent made a good trade. Developers still need rules connecting the returned data to the action, including freshness and execution conditions. What interests me is being able to check the work behind an otherwise convincing result. A profitable trade can hide a broken process. Once an agent has permission to move funds, Iโ€™d want those skipped steps to be detectable too.
57
7
1
61
566
hello @PrismaXai bangladeshi ppl! ๐ŸŽฎ ready for game event? ๐Ÿ•ค tonight at 9:30 PM | BST join us for an evening of friendly competition with @RIFAT5666 & @aliorbz. bring your friends and see who can climb to the top. everyone starts with 1,000 in-game coins, with just 5 minutes per round to make their way up the leaderboard. new to the game or have questions? Please join #voice6 at 9:25 PM, five minutes early, so the hosts can walk you through the rules. see you there ๐Ÿ™Œ
The robot was seeing the past! @PrismaXai was getting a fruit-sorting demo to run on a different robot rig from the one it trained on. The arm kept coming in a little off when it tried to pick things up. The team was close to doing another round of training when someone took a closer look at the camera feed. A Python buffer was holding frames until 64 KB of data piled up. By the time the model got an image, it was already working off a view that was hundreds of milliseconds old. Here's the weird part: in their simulator, up to 280 ms of lag barely moved the needle. On the real table, a near miss could bump a piece of fruit and send it rolling. Now the next move has to account for a scene the robot just changed. That's PrismaX's theory for why the sim was so forgiving. They fixed the buffer, sped up the camera feed, and cut model inference from over 230 ms to 87 ms per step. Once the full loop was under 120 ms, the arm started doing something the human operators had shown it: move in quick, pause to check where the gripper is, make a tiny adjustment, then grab. That little pause is the part I keep coming back to. The model had already seen how people corrected their grip. The robot needed a fresh enough view of the table to pull it off.
6
1
10
168
mmix is 2009 in roman numerals. the year bitcoin went live. the year anyone, anywhere could join a network without asking anyone for permission. and @MMIX_io is building the exact opposite. โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€ it's a permissioned layer 1 for real-world assets. validators are appointed by the mmix foundation. every company and project gets reviewed before it can deploy. compliance rules like kyc tiers and regional restrictions sit in an on-chain registry that apps check before value moves. at first that sounds like a contradiction. why name yourself after bitcoin's birth year and then put a gate at the door? but the gate isn't where most people assume. anyone can still use the apps. what's vetted is who builds and who runs the network. for property, equities, payments or a resort's membership program, that line matters. real assets need someone accountable behind them. โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€ so i don't read the name as irony. i read it as a reference point. 2009 proved value could move on-chain without permission. mmix is betting the next chapter is real assets moving on-chain with known participants behind them. mainnet is planned for 2027. i'm curious whether the curated approach actually brings the businesses it has lined up on-chain.
37
6
41
469
MD RIFAT retweeted
Who Needs @supernoversNFT GTD Spots 7 GTD WL Spots GIVEAWAY To Enter: โ€ข FOLLOW @HOODZNFTs โ€ข RT + Drop your EVM๐Ÿ‘‡ โ€ข Tag 2 Friends
114
88
129
2,028