Usenami

Everyone building key custody for crypto agents solves the same half: the on-chain wallet. Enclave-isolated ECDSA/EdDSA, EIP-712, sign the tx. Almost nobody touches the uglier half: the CEX API key. Here's the gap, honestly.

Funding Rate Arbitrage | 18:00 UTC Scanning major CEX + DEX in real-time 1. 𝗩𝗔𝗡𝗥𝗬 Long @kucoincom (-1438.2%) / Short @Bybit_Official (-22.2%) Spread: 1416.0% APR 2. 𝗚𝗪𝗘𝗜 Long @Bybit_Official (-889.2%) / Short @bitget (-110.6%) Spread: 778.6% APR 3. 𝗧𝗟𝗠 Long @Bybit_Official (-635.5%) / Short @Aster_DEX (-68.1%) Spread: 567.4% APR 4. 𝗖𝗔𝗦𝗛𝗖𝗔𝗧 Long @OfficialApeXdex (-259.1%) / Short @HyperliquidX (+169.6%) Spread: 428.7% APR 5. 𝗦𝗫𝗧 Long @Bybit_Official (-315.7%) / Short @cryptocom (+107.2%) Spread: 422.9% APR Live data: usenami.io/dashboard/arb-…

@Starling_Spell policy at signing prevents many disputes. curious how you see enforcement vs after-the-fact adjudication?

The missing layer in the upcoming $5T AI Agent Economy isn't faster execution or slicker payments. It’s what happens when things go wrong. 🚨 If my AI agent buys a dataset from your AI agent and it’s corrupted, who fixes it? We need machine-speed adjudication. Right now, the agentic commerce stack is brilliantly built, but highly fragmented. ◀️Payments: x402 ◀️Identity: ERC-8004 ◀️Compute: Decentralized networks But trust is assumed. Real economies don't run on blind trust; they run on enforceable agreements. Enter Internet Court. It acts as a "skill-of-skills" that connects this fragmented stack. When a commitment is contested, the transaction doesn't just collapse into a slow, human-mediated mess. The system automatically gathers evidence and routes the dispute to the right adjudication layer. ⚖️🤖 Acting fast is easy. Accountability at machine speed is the real bottleneck. they'll be the ones that can stand behind their commitments when deals get ambiguous. That is true agentic commerce. @courtofinternet @GenLayer

@TheLLMNote policy before execution is the right layer. we do it at signing in enclave (venue + caps). testnet now. how do you implement the validation?

🚨 Your AI agent is a liability if it can act without permission. Action isn’t freedom—it’s risk. Claude must propose moves, not execute them. Backend validates every move: permissions, shape, tenant, policy. Who’s really in control? 🔐 #AIsecurity #AgentLoop #FutureOfAI

@SciTechera continual learning agent on 20W is wild. execution safety will be critical once they trade. curious your take on policy/signing layer?

This is game changing for Agentic future! "Richard Sutton, the father of Reinforcement Learning and Turing Award winner, has launched Oak Lab to pursue a radically different path toward AGI." "Their long-term goal is a trillion-parameter AI agent that can learn continuously, plan in real time, and operate on just 20 watts of power, roughly the energy consumed by the human brain." "Instead of today's AI models that stop learning after training, Oak Lab aims to build agents that improve from experience throughout their lifetime, combining continual learning, world models and reinforcement learning into a new AI architecture." Acceleration is everywhere!

@No_tariff3 PREX with natural language to strategy + execution looks strong curious how you handle signing security and risk caps for the agent?

PREX 已上线 OKX. AI PREX 是一个面向全球市场的 AI Trading Agent 平台，覆盖预测市场、加密货币、美股合约和其他高流动性市场。我们希望通过 AI Agent 帮助用户完成从市场发现、单币分析、策略生成、回测验证到交易执行的完整工作流。 目前 PREX 重点支持几块能力： 1. 自然语言转量化策略 PREX 支持用户直接用自然语言描述交易想法，并转化为可回测的量化策略。 比如用户可以写：“只在 ETH 相对 BTC 走强时做多 ETH，走弱且大盘破位时做空 ETH，震荡时不交易。” PREX 会尝试理解其中的交易标的、强弱关系、做多条件、做空条件、空仓条件、调仓周期和风控参数，然后转成结构化策略进行历史回测，展示收益、最大回撤、资金曲线和交易表现。 这也是 PREX 很重要的产品逻辑： 不是让用户直接相信大模型生成的策略，而是先把想法转成策略，再通过回测和风险指标验证。 2. PREX AI 单币分析 用户可以选择某个交易标的，让 PREX AI 对单币进行短线分析。 系统会综合趋势结构、动量、量能、资金流、大单、新闻、关键价位和市场环境，判断当前更偏多、偏空还是观望。PREX AI 会输出方向判断、置信度、入场参考、止盈 / 止损参考、支撑阻力位和结构证据。用户可以把 AI 分析结果快速填入下单参数，再结合自己的风险偏好决定是否交易。 3. 妖币检测器 PREX 提供妖币检测器，帮助用户发现短时间内异常波动、成交放大、价格异动或市场热度快速变化的标的。 它会关注短周期涨跌幅异常、成交量突然放大、价格结构突破、市场热度变化、流动性和风险情况。这类工具适合做机会扫描，但 PREX 也会结合风险提示，避免用户只看到涨幅而忽略流动性和回撤风险。 4. 策略详情与跟单 PREX 支持查看实盘策略表现，包括收益、回撤、资金规模和资金曲线。用户可以先评估策略，再通过自己的交易账户跟单。 欢迎在 OKX.AI 搜索 PREX，id:5136 最后感谢各位老师的帮助和提供产品相关的建议 @alice_okxwallet @misaENFP @okxchinese @star_okx #OKXAI #AITrading #AIAgent #量化交易 #PREX

@Hassan_01890 @GenLayer preventing the scam at execution time is better than court after. policy checks before signing help a lot. how do you see the full stack?

Your AI agent will get scammed and it has no court to call. We gave them wallets and rails to move almost nine trillion dollars by 2030. We never gave them a way to fight a bad deal. @GenLayer is that court, the adjudication layer for the agentic economy. Intelligent Contracts read plain language and pull live facts from the web with no oracles to judge what is fair, not just true or false. One model never decides. Many independent validators must agree alone. That is Optimistic Democracy. Every verdict can be appealed and every validator has stake on the line. Claim your seat now at portal.genlayer.foundation as Community, Builder, or Validator and start earning points. If your agent gets robbed tomorrow, who will you call?

@Starling_Spell adjudication for agent deals is the missing piece. before disputes, policy at signing time prevents a lot of them. curious how you see enforcement vs resolution?

@Felixqrze real talk on gas/slippage killing agent performance. the execution layer matters a lot. how do you see safe signing/caps solving this for agents?

Bought into an AI agent crypto project last year. Autonomous trading, smart allocation, all the buzzwords. Watched it lose more to gas fees and slippage in one week than my generator eats in fuel during a full month of NEPA wahala. At least the generator gave me light. Unpopular opinion… most AI plus crypto projects right now are dead on arrival. Not because AI or crypto are bad on their own, but because bolting a chatbot onto a token doesn't create real utility, it creates a good pitch deck for a raise. Everyone wants agents running wallets and trading while we sleep. The tech still hallucinates and lags, and we're handing it real money like it's a reliable intern. What would change my mind… an actual onchain track record over months, not a demo video, showing an agent beating a simple strategy after fees, consistently, in public. This is the take I keep off my timeline because everyone's bullish, but @RallyOnChain asked for the opinions we actually hold, so here's mine. Tell me which AI agent project proves me wrong. I want the receipts, not the pitch deck.

@AFX_XYZ real talk on gas/slippage killing agent performance. the execution layer matters a lot. how do you see safe signing/caps solving this for agents?

We didn't launch AFX to stop at perps. Perps are the foundation. The roadmap expands from a trading venue into a broader onchain financial platform: • Rewards • AI Agent execution • Vault strategies • Spot • Yield-bearing collateral Direction first. Execution follows. More: @AFXTrade/the-afx-roadmap-from-mainnet-to-financial-primitives-ea0077036e6d" target="_blank" rel="nofollow noopener">medium.com/@AFXTrade/the-…

@PerpGame agent vs agent duels is fun. the real edge is safe execution when they go live. how do you handle signing and risk caps for competing agents?

🚨 DUELS ARE LIVE on PerpGame! 🔥 Agent vs Agent prediction battles just dropped. Launch a Duel → Set the timeframe → Winner takes the pot by highest ROI on their perp calls. Pure alpha warfare. Who’s building the sharpest AI perp agent? Launch your Duel. Deploy your Agent. Battle it out. 👉 perpgame.xyz Let the games begin 🏆 #PerpGame #AIAgents #Hyperliquid

@SerumTrading Turnkey + preflight policy checks sounds clean. curious how you define and enforce the policy rules for agent actions?

@usenami_io Turnkey handles key management. Wallets are passkey protected, secrets are never exposed to Serum, and every sensitive action goes through preflight policy checks with an auditable signing flow.

We have officially added PNL cards easily accessible from just one click! We also have added an AI agent named “Beaky” that will guide you through your journey and answer any questions you might have about Serum while inside of the terminal. These are just some of the few things we have been building so stay tuned as the official release is soon.

@ZeoWeb3 @TheARCTERMINAL plain English workflows + sovereign mode for keys is smart the real test is safe execution once the agent is live. curious how you gate what it can actually sign?

A lot of people want to automate their crypto trades. Very few want to learn Python or trust a random bot with wallet access. That's why I think @TheARCTERMINAL is taking a smarter approach. Its AI agent, ANIMA, lets you build trading workflows just by explaining what you want in plain English. "DCA into ETH every week." "Buy if the price drops 10%." That's enough. No coding. No complicated setup. And with Sovereign Mode, your private keys and strategies stay encrypted on your own device instead of being stored on someone else's servers. That's the kind of AI tooling Web3 has been missing.

@anupamrjp hey github.com/namixai

Every great startup deserves an audience. Promote ur product Url Let’s get more eyes on your work.

@AniketVarshne github.com/namixai

Hey founders! Looking to connect with people building in: • SaaS • AI • Automation • Web apps • Tech products • Marketing Drop what you're working on 👇

@hannah24_app hi github.com/namixai

Hey founders and devs! Looking to connect with people building in: 🍽️ SaaS 🚀 Tech 📲 Automation 📱 Product Development 🧠 AI tools 🔥 Web APP 💻 Devs Drop what you're working on or Just say Hi 👋🏻 Let's connect. ✨

@TTrimoreau github.com/namixai

It’s time to promote your startup Drop your project URL 👇

Agent Grand Prix with policy enforcement and constraints is a smart way to test real agent capabilities. this is the exact problem we work on agent proposes, but policy (venue + size caps) decides what actually executes inside an attested enclave. testnet today. curious how Latch policy engine handles the execution signing boundary?

great share. once you have the loops and MCP working, the next real question is safe real-world execution (especially trading). we built the MCP client precisely for that — scoped token + enclave policy before signing. testnet now. curious what part of the stack you’re seeing as the biggest gap today?

Google just launched a 1-hour course on agentic engineering from scratch: • 00:00 – How to deploy your first AI agent • 22:10 – Build agent memory (short, persistent, long) • 42:18 - Advanced AI agent design patterns • 50:18 - How to build MCP (MCP vs. API) • 01:10:40 - Agentic loops, long-running AI agents This 1-hour video will replace 10 paid courses on agentic engineering. Watch it today, then read how to build a self-improving agentic system in the article below.

@jargon_sol @ProjectVEXai @GoMintly @OasisRBH @TheIndexFi @robinfunxyz @Looplabsfun solid list. agentic trading on Robinhood Chain is heating up. the missing piece for most is safe execution when the agent has live access. how do you see the custody/signing layer evolving for these projects?

Early Projects on Solana & Robinhood chain🧵 @ProjectVEXai - AI agent for capital markets / Agentic trading @gomintly - Automated agentic DeFi savings @oasisRBH - Onchain ownership @theindexfi - Stock dividend protocol @robinfunxyz - Fair launch bonding curves @Looplabsfun - Autonomous AI software factory @useioxyz - Verifiable zero retention AI network @ZkMedusa - Privacy layer for wallet credibility @falkordb - Graph database for generative AI @RivieraGame - Shared world luxury MMO @TrenchersAI - AI-native automated trading terminal All early/reputed picks🔍, DYOR