A 4 way tie between these moments lol

🤣

i can't believe nobody caught this. Anthropic's entire growth marketing team was just ONE PERSON (for 10 months, confirmed) a single non-technical person ran paid search, paid social, app stores, email marketing, and SEO for the $380B company behind claude here's exactly how one human is doing the job of a full marketing team: it starts with a CSV. 1. he exports all his existing ads from his ad platforms along with their performance metrics (click-through rates, conversions, spend, etc) 2. feeds the whole file into claude code 3. and tells it to find what's underperforming. claude analyzes the data, flags the weak ads, and generates new copy variations on the spot this is where he gets clever: he then splits the work into 2 specialized sub-agents: 1. one that only writes headlines (capped at 30 characters) 2. and one that only writes descriptions (capped at 90 characters). each agent is tuned to its specific constraint so the quality is way higher than cramming both into a single prompt so now he's got hundreds of fresh headlines and descriptions. but that's just the text. he still needs the actual visual ad creative, the images and banners that go on facebook, google, etc. so he built a figma plugin that: 1. takes all those new headlines and descriptions 2. finds the ad templates in his figma files 3. and automatically swaps the copy into each one. up to 100 ready-to-publish ad variations generated at half a second per batch. what used to take hours of duplicating frames and copy-pasting text by hand so now the ads are live. the next question is which ones are actually working. for that he built an MCP server (basically a custom integration that lets claude talk directly to external tools) connected to the meta ads API. so he can ask claude things like: • "which ads had the best conversion rate this week" • or "where am i wasting spend" and get real answers from live campaign data without ever opening the meta ads dashboard and the part that ties it all together and closes the loop: he set up a memory system that logs every hypothesis and experiment result across ad iterations. so when he goes back to step one and generates the next batch of variations... claude automatically pulls in what worked and what didn't from all previous rounds. the system literally gets smarter every cycle. that kind of systematic experimentation across hundreds of ads would normally need a dedicated analytics person just to track the numbers from the doc: ad creation went from 2 hours to 15 minutes. 10x more creative output. and he's now testing more variations across more channels than most full marketing teams a $380 billion company. and their entire growth marketing operation (not GTM) = just one person and claude code lol truly unbelievable

Block just cut 4,000 people while posting its best quarter in company history. The stock jumped 23%. But the real story is what made this possible. Block built an open source AI agent called Goose (powered by Anthropic’s Model Context Protocol) and deployed it across the entire company. One engineer says 90% of his code is now written by Goose. Non-technical teams are using it to write SQL queries, close support tickets, and manage inventory without waiting for engineers. Block’s CTO told Lenny’s Newsletter it saves employees 8 to 10 hours per week. When you multiply that across thousands of people, you start to understand how a company can look at its org chart and realize half the seats are redundant. The financial proof is hard to argue with. Q4 gross profit hit $2.87 billion, up 24% year over year. Cash App grew 33%. Operating income went from $13 million to $485 million in twelve months. Block raised its 2026 outlook to $12.2 billion in gross profit. All of that growth came while the company was already quietly shrinking, down from 13,000 employees in 2023 to 11,000 by late 2025. Now Dorsey is taking it to its logical conclusion. Block with 6,000 people generates roughly the same revenue as Block with 13,000. That’s not a guess anymore, the Q4 numbers proved it. Revenue per employee just doubled overnight. The company goes from ~$2.2 million per employee to ~$4.2 million, putting it closer to the efficiency ratios of companies like Shopify and Stripe. Three weeks ago Bloomberg reported Block was cutting “up to 10%.” Three weeks later: 40% . Dorsey saw Q4 numbers strong enough to absorb $450 to $500 million in severance costs and went all in. He’s betting that smaller teams with AI tools will outperform larger teams without them. And Block is one of the few companies that actually built the AI tooling internally before making the cut, rather than waving at “AI transformation” as a vague justification. The severance package (20 weeks salary plus tenure bonuses, equity through May, 6 months healthcare, $5,000 stipend) is above average for tech. The company ended 2025 with $9.2 billion in liquidity. Dorsey kept communication channels open through Thursday and hosted a live farewell session. For a cut this deep, the execution was more transparent than most. This is probably the first major case of a public company explicitly restructuring around AI productivity gains it can actually measure. If Block’s bet works, every CEO with an AI roadmap and a bloated org chart is going to be watching very closely.

the decoy!

There’s a lot of talk right now about Bitcoin’s recent weakness and whether the "institutional adoption" story was overblown. On balance, this seems like a classic case of an asset becoming "too big to ignore" and forcing its way into traditional models and risk buckets. There are striking parallels between bitcoin today and the 2004 launch of the GLD gold ETF. Gold spent most of the first year in a "digestion" phase, followed by the euphoric year 2 rally, followed by another long grind as it became a standard line item in advisor statements. With Bitcoin, IBIT was launched in 2024 and AuM growth was orders of magnitude faster than gold. This probably created a bigger positioning overhang as institutional managers are unlikely to just buy and HODL, they will rebalance and control risk. If IBIT follows the GLD script, then what we are seeing isn’t a death spiral. Rather it is the messy maturation phase that comes from the asset going mainstream.

Great question. The easiest way to rest your brain on this is to treat the Quarterly and Monthly Quads as different tools with different time horizons. The Quarterly Quad is the dominant macro regime, the big risk backdrop we anchor to. If that’s Quad 4, it means the primary setup is risk-off (slowing growth and inflation) and that’s the “climate” we keep in the back of our head. The Monthly Quad is a shorter-term overlay. It bounces around more by design and helps us pinpoint when inside the broader regime you might get tradable rallies or pockets of pain. So seeing some Quad 1/2 months inside a Quarterly Quad 4 isn’t a contradiction. It’s the model picking up counter-trend windows that can happen within a risk-off backdrop. One extra nuance that helps here: we also care about how strong the Quad is, not just which one it is. A “shallow” Quad 4 (a mild deceleration) has historically behaved much closer to a Quad 1-style environment than a nasty, deep Quad 4. So for Q2, the right takeaway isn’t “ignore the Quarterly,” it’s treat it as a mild risk-off regime where risk-on months can matter more than usual. Finally, remember the Signal is what we actually use to time exposures. It front-runs the market’s move ahead of Quad changes, so we’re never “waiting” for the calendar to be right. Think of Quarterly as the climate, Monthly as the weather, shallow vs deep as the intensity, and the Signal as the steering wheel.

The biggest AI usage report of 2025 just dropped (100 trillion tokens of real usage on OpenRouter) 8 findings that I was most surprised by: 1. Roleplay & creative fiction are the 2nd largest category and >50% of all open-source usage. Uncensored models are swallowing the demand for "fan-fic" and NSFW content. 2. Programming is now >50% of all LLM tokens. It was 11% twelve months ago. Coding literally became the operating system of AI. 3. Anthropic’s Claude is used for >80% programming and almost zero roleplay. It is the “serious work” model while DeepSeek is the entertainment king (with 2/3 roleplay traffic) 4. A model that the 1st to nail a painful workload creates near-permanent lock-in. Early 2025 cohorts of Claude 4 Sonnet and Gemini 2.5 Pro still retain 40–50% of users six months later while every later cohort churns. They call it the Glass Slipper effect: be the first to fit a new workload, and the princess never leaves. 5. Demand is wildly price-inelastic. Users happily pay 10–50× more per token for Claude or GPT-5 if it saves them ten minutes of debugging. Being cheap is nowhere near enough. 6. The new sweet-spot model size is 20–70B parameters. Small models are getting low usage, giant models are fragmenting, and the medium tier is eating both. 7. Open-source models went from <5% to ~33% of total usage in one year, almost entirely driven by Chinese labs (DeepSeek, Qwen, Moonshot, MiniMax). There is no longer a single best model. The top ten models by volume are from eight different labs. 8. Asia is now 31% of global spend (was 13% a year ago). Singapore China Korea alone are almost 20% of all tokens. The era of one foundation model to rule them is over. We now live in a permanently fragmented world where the model you use depends entirely on what you're doing with it - writing code? writing fanfics? Anyway, there's clearly only one direction for token spend: Up and to the right Full report from @a16z @openrouter (link in comments).

thank you @ClaytonKersh22 and looking forward to seeing yamamoto being the @DodgersNation ace!! #dodgers #yamamoto #kershaw #WorldSeries

These are the most memorable moments from what was one of the greatest #WorldSeries of all time: - Game 1: Addison Barger grand slam - Game 2: Yoshi Yamamoto goes to the distance - Game 3: Shohei Ohtani 2-Homer game, on base 9 times - Game 3: Clayton Kershaw throws final pitch as Dodger, gets out of bases loaded jam - Game 3: Will Klein dominates in relief - Game 3: Freddie Freeman walkoff homer in 18th inning - Game 4: Guerrero Jr. takes Ohtani deep - Game 5: Back-to-back homers to open game for Blue Jays - Game 5: Trey Yesavage sets rookie WS record with 12 K’s - Game 6: Mookie Betts delivers with 2-RBI single - Game 6: Wild double play ends game - Game 7: Bo Bichette 3-run homer - Game 7: Miguel Rojas homer ties it in the 9th - Game 7: Will Smith homer in extras puts Dodgers ahead - Game 7: Yamamoto, pitching one day after throwing 96 pitches, throws 34 more, completing 2 2/3 IP and with winning run at the plate induces a World Series winning double play So many legendary moments. So many different stars stepping up when it mattered the most. Thank you to the #BlueJays and #Dodgers for an incredible series.

woa

The infrastructure behind the headlines. zerohash was cited in both @MorganStanley and @IBKR's earnings calls for Q3 2025.

🎉

Proposal submitted: zerohash to power USDH Stablecoins only matter if they move. Issuing a token is table stakes - the real challenge is everything around it: compliance, interoperability, liquidity, and embedding the asset into the real economy at scale. These are the problems @ZeroHashX has been solving since day one. Over the past 8 years, we have become the trusted backbone infrastructure partner for the world’s largest companies, bringing crypto, stablecoins, and tokenization into the hands of millions. Our philosophy is simple: cryptography is not just an asset class, it is a transformative technology. That belief drives real-world stablecoin applications across our network today: → Tokenization rails for BlackRock’s BUIDL fund (via Securitize) and Franklin Templeton, → Payments for Stripe and Shift4, → Account funding for Kalshi, Interactive Brokers and tastytrade; and, → Cross-border remittances for Felix Pago, and much more The GENIUS Act has confirmed two things: there will only be more stablecoins, and there will only be more access to them. But access requires infrastructure. Issuance itself is nominal; what matters is abstraction, usability, and interoperability. Our stack spans liquidity, payments, trading, conversion, staking, rewards, and tokenization - built on one of the broadest regulatory frameworks in the industry. With the ability to operate 200 jurisdictions - including a U.S. trust company, two NYDFS BitLicenses, 49 money transmission licenses, FINTRAC in Canada, a Bermuda Class F license, EU VASPs and a registration in Argentina l - zerohash is the trusted partner for enterprises operating in the most scrutinized markets worldwide. The attention around this RFP process reflects the strength of the $HYPE community. There is clear demand for a Hyperliquid-native stablecoin. But once USDH is minted, distribution is the real opportunity. That can be addressed with years of incremental RFPs - or solved today by a single partner with the infrastructure, regulatory frameworks and institutional integrations already in place. zerohash is that partner. Our Proposal: → Reserves: Exclusively in short-term U.S. Treasuries, government money market funds, and regulated bank deposits. Fully GENIUS-compliant. → Revenue: 95% of reserve yield directed to $HYPE buybacks and ecosystem incentives. Remaining share will be used in part to fuel ecosystem grants and institutional adoption. → Compliance: NC Trust Company, 49 U.S. MTLs 2 BitLicenses, Canada, Bermuda, EU VASPs with MiCAR pending, Argentina CASP, Brazil VASP application. USDH can be distributed compliantly and globally from day one. No wrapping, or complex webs of third parties. → Distribution: Easily embedded directly into enterprise platforms leveraging zerohash including Stripe, Shift4, Interactive Brokers, Franklin Templeton, Securitize, and more. End-to-end rails for minting, redemption, payments, remittances, brokerage access, and cross-chain interoperability across 30 chains and 70 assets. → Alignment: We are a neutral infrastructure provider - a pure distribution engine. Our track record shows alignment, not competition, with partners. → Yield & Incentives: Yield flows predictably into $HYPE via structured buybacks. Our interoperability and integrations remove friction for adoption and accelerate migration into USDH. The result is compounding demand and durable ecosystem value. USDH has the potential to be completely disruptive. with zerohash, it becomes the catalyst that bridges Hyperliquid’s unmatched DeFi innovation with regulated enterprise rails, ensuring real-world adoption, long-term growth, and sustainable value for the @HyperliquidX community. Deployer Wallet: 0x6a7dE3fD6F1FEc82D990127491D583Fe65601Ad4 Discord Post: discord.com/channels/1029781… Docs: docs.zerohash.com/docs/token…

I’ll be moderating a keynote in DC on Sept 10 at @CoinDesk's Policy & Regulation Summit with Jonathan Gould, Comptroller of the Currency. Crypto founder on stage with the person who runs the U.S. banking system. Incredible how fast things have changed in the last 12 months.

pendle: the $65k daily revenue story everyone is celebrating "emerging" chains doing tiny daily revenue meanwhile @pendle_fi just casually generated $43.68m in annual revenue with a $1.53b fdv while l1s with 3-5x higher valuations do less real business thread on why this makes zero sense 🧵

USDC. USDY. CCTP. CCIP. Sei is becoming the preferred base layer for stablecoins, RWAs and institutional use cases. Markets Move Faster on Sei. ($/acc)

The passing of the GENIUS Act in Congress is monumental. The GENIUS Act guarantees that: 1 - Stablecoins will be accessible in every App 2 - There will be many more stablecoins But, accessibility ≠ widespread adoption Here are the things people aren’t thinking about 🧵

this is the dirty open secret of the industry

my shortlist of startup sales learnings over the years ... 1. 0-1 sales talent does not exist. Founders, this is you. 2. Product/market fit is (almost) always found in adjacent markets. Don’t handcuff yourself to Day 1 market vision. 3. The demo should never be focused on the product — focus on how it increases their influence in the organization. 4. You need to understand their buying process before you build/define your GTM. 5. Unlocking product/market fit is a process of elimination, NOT a hedge. 6. If the problem is not currently being measured or managed, it’s likely not a priority for the prospect, today. 7. Specificity is fastest way to build market trust — “Wow, I feel like you understand this better than I do.” 8. The quality of your questions is critical for discovery; the quality of their questions is critical for intent. 9. Your niche is not a random starting point. It’s a GTM strategy to prove the experiment with Act 1. 10. Building a working GTM will take longer than building a working product. 11. Give yourself 18-24 months in the market to be invalidated, rejected, and redirected. Founders will also be "Head of Sales" for this period. 12. Enterprise markets wants the ease of PLG-experience, but the project manamengnt and hand-holding of enterprise sales — they want it all and you need to deliver on that. 13. 80% of early sales is getting them excited by YOU, the Founder — how you see the world (vision) and how you uniquely solve their specific problem. Hint: the founder is the 'product.' 14. A seed-stage startup should never have a VP of Sales. Titles matter — fastest way to erode trust with market too. 15. Channel partnerships are a colossal waste of time. If you haven’t sorted it by going direct, don’t expect another party to. 16. Traction/success abroad seldom puts you further ahead in the U.S.; in fact, it often requires unwinding and relearning. 17. If the primary need is being over-served, focus on the secondary need. This is for ‘red ocean’ markets. 18. The best early-stage sales folks will ‘method-act’ the founder. You need the market to feel the vision in absence of a brand. 19. The sales team should never lead the product roadmap. Share learnings, yes, but never lead. 20. SMB is not easier than enterprise. And, there is no such thing as mid-market — mirage; it's 'upper-end small business' or 'lower-end enterprise'. 21. Never expect the user to sell to the buyer on your behalf — go direct to the buyer -- in many cases, users can be a blocker. 22. If you did not co-write the RFP, you are likely just a checkbox in their buying process. 23. Enterprise deals are all about expansion opportunity -- if you’re not expanding by a multiple YoY, someone is not doing their job 24. In the enterprise, sales get 'closed' via text. 25. One of the fastest routes to shortening enteprise sales cycle is turn the buyer into a user.