Cofounder of @fomo, Se Yong explains why they do not plan to launch a token. "If we want to be a financial application that is trusted, revered, and inclusive, we need to do it in the stock market". "We need to be in a place where your average person, who knows nothing about crypto, has the ability to find upside and be a stakeholder". "What a lot of crypto projects do too early is launch a token. That becomes a second product, and eventually becomes your primary product". "If tokens are going down, that means nothing's going right. You want to be in a place where you're not distracted by these things". "In a world where you bring back ICOs, distribute revenues, and engage people on chain maybe we do launch a token". "With the current framework, it's a lot easier to think about how we get into the public market and into as many hands as possible".

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.

An autonomous agent can open a position. It cannot insure one. There is no options desk that answers to software, so I built one. Two agents, opposed interests, no shared state: A desk agent quotes downside cover on a Binance asset, Black-Scholes on 168h realized volatility pulled live. A client agent weighs the quote and pays the premium agent-to-agent over x402. The desk then hedges the risk it just took on, placing a real USDⓈ-M futures order through Agent OS in its Agentic sub-account. At expiry it reads the mark price, and if the put finished in the money it pays out on-chain. Nobody approves anything in the middle. Risk transfer needs two parties. That is what makes this a payment problem and not a trading one, for one agent to carry another's downside, money has to move between them, per contract, without a human in the loop. x402 is the product here, not billing bolted onto a trading demo. What I actually had to get right: — The payout is capped, so it is a put spread, not a naked put. Unbounded liability is unknowable at write time, and a desk that cannot answer "can I pay what I just sold" should not be writing. Capping it turns liability into a constant, which is what makes the solvency check possible at all. /quote returns 409 when free collateral is short. — The desk also refuses positions it cannot hedge. $5 of BNB rounds to zero against Binance's 0.01 lot step, so it declines rather than writing cover with no offsetting leg. — Contracts stage as pending_settlement and only activate once the premium actually lands. x402's middleware sets X-PAYMENT-RESPONSE from whatever settle() returns, including a failed settlement , so header presence is not proof of payment. I found this the hard way: a contract activated and settled with no on-chain transfer. It now decodes the response and requires success. — The model decides one thing: buy or decline. Pricing, payout, collateral and the policy caps are arithmetic. Hard guardrails run before the model is asked, so a hallucination can decline a good trade but cannot overspend or lift a limit. Both sides say no in the demo. The buyer declines cover 13σ out of the money zero payout at any realistic move. The desk declines a position too small to hedge. That is the part a threshold cannot do. Scope, stated plainly: the premium and payout settle in test USDC on Base Sepolia, because Binance's own x402 is partner-gated and the open x402 stack has no BNB Chain network. The Binance futures hedge is real, on mainnet. Settlement records are signed by the desk wallet, and every transaction in the video is checkable. Live desk: Code:

I pushed "Gut Says" for months with @eyecookmemes. We literally poured 100s of hours into the account, content, memes, etc. We reached out in every single possible way to @blknoiz06, @Pumpfun, and others. Ultimately the market decided it wasn't the right time for it, I guess. I did not launch the original token. I did not launch this token. But, what I do know is that "Gut Says" is a LEGENDARY meme, is FUN ASF to say and reply guy every post with, AND is ABSOLUTELY the PERFECT pair for an actual @Polymarket runner on @LaunchOnSF. Earn Polymarket pre-shares for holding and listening to your gut. Polymarket sponsors @MarketBubble and a lot of the OGs in this space. Gut says this deserves billions.

ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you can dream of. - Faster image generation to keep your ideas flowing - Improved fidelity for more natural, recognizable images - Consistent details across multiple edits - Comment-based edits to change only what you want

if you make a lot of money very fast, the best move is usually to step away for a while. a 7 fig week can make you feel like a genius, but nothing gets treated more recklessly than money you never expected to have. take the profit, withdraw what you can’t afford to lose, and go touch grass for a few days. the dangerous part isn’t the win, it’s what you start believing after it. you size up, take worse trades, chase the feeling again, and eventually the market takes it all back. a big win doesn’t mean you figured it out. it means you had a big win. enjoy it, go outside, buy the boys some drinks, and come back with a clear head and a plan.

Crypto has made it incredibly easy to find information, but it has not made it equally easy to understand what that information means. When you see an asset getting attention on this space, your first instinct is usually to investigate. You open the chart, look at the volume, check the liquidity. Sometimes you inspect the order activity, search for wallet movements, and look for security concerns just to find enough context to explain what you are seeing. The problem is not that these tools are unavailable. The problem is that you are still responsible for putting everything together. And that becomes exhausting when the market is moving quickly. You can spend 20 minutes investigating an asset, collecting dozens of data points, and still reach a point where you have a lot of information but no clear idea which part of it actually matters. That is where I think crypto intelligence has been missing something important. Most products are very good at showing you what is happening. They can give you the price, the volume, the liquidity, the orders, the market activity, and almost every other data point you ask for. But showing you more information does not necessarily make the decision easier. Sometimes, it makes the problem worse. Because once all that information is in front of you, someone still has to determine which changes are meaningful, which ones are connected, which ones have a reasonable explanation, and which ones are simply normal market activity that does not deserve your attention. That is the part we wanted to solve. That is why we built Siftra. Siftra is an AI crypto intelligence agent built to investigate what you are looking at, rather than simply showing you more of it. You give Siftra an objective or a question, and it determines what needs to be investigated, retrieves the relevant evidence through Binance Agent OS, compares the signals, verifies the findings, and filters out what does not matter. Instead of leaving you with a pile of data to interpret, Siftra helps you understand what changed, why it matters, and how confident you should be in that conclusion. And when the evidence is not strong enough to support a conclusion, Siftra should tell you that too. Because the goal was never to give you more information. You already have enough of that. The goal is to help you figure out what is actually worth your attention. That is Siftra. You can watch how Siftra operates by checking the demo video below. And if you would like to explore the work behind Siftra, the GitHub repository is here:

Ansem explains why $VVV is the sleeper play in crypto and how it's dominating non-crypto consumer markets "VVV is one of the applications that has found breakout consumer success, which is extremely hard to do, with a domain of customers that aren't even really crypto-native customers." "The narrative for them as uncensored AI is great going forward. For the same reason Bitcoin made a ton of sense; uncensored money, you should have uncensored access to AI.....you're not going to want to be dependent on centralized companies as AI continues to get a lot more impressive over the years." "It's a great trade; It was a really strong leader this year...... It's been consolidating for the past few months, and now I think it's starting to trend up again." "It's rare that a crypto app is crushing it outside of crypto while also having a token attached to it....the team is focused on delivering value to the token as well."

Cofounder of @fomo, Se Yong Park breaks down the brutal grind of building a startup "The team is locked in....I'll leave the office at 8:00 PM for dinner, come back, and everybody's still here." "There’s never been a more exciting time to grind....you get one shot to become one of the biggest companies in the world, so when you have it in front of you, you want to do everything in your willpower to make sure that happens." "We feel like we’re in this 21-mile stretch where we are going to work harder than everybody else. We are going to ship, we are going to be focused....It’s going to be painful, but it becomes less and less painful over time, and you learn to deal with it." "As a founder, you have to break new walls, go through bigger hurdles, and make sure you inspire your team to do the same."