Our mission at Kash is to completely redefine how the world prices and distributes beliefs. We are meeting YOU where you already are, starting with social media. To accelerate this journey, we are proud to have partnered with top-tier VCs and raised $2M to bring prediction markets to your feed. Backed by @TheSpartanGroup and @BigBrainVC, with participation from Coinbase Ventures, Polaris Fund, Moonrock Capital, Halo Capital, Kosmos Ventures and Fabric Ventures. [Kash Flash Market: Will at least 3 verified accounts with over 100k followers quote post this announcement within the first 12 hours?]

In 2014, there was a vision: you can have permissionless, decentralized applications that could support finance, social media, ride sharing, governing organizations, crowdfunding, potentially create an entire alternative web, all on the backs of a suite of technologies. Ethereum: the blockchain. The world computer that could give any application its shared memory. Whisper: the data layer. Messages too expensive for a blockchain, that do no need consensus. Swarm: the storage layer. Store files for long-term access. Over the last five years, this core vision has at times become obscured, with various "metas" and "narratives" at various times taking center stage. But the core vision has never died. And in fact, the core technologies behind it are only growing stronger. Ethereum is now proof of stake. Ethereum is now scaling, it is now cheap, and it is on track to get more scalable and cheaper thanks to the power of ZK-EVMs. Thanks to ZK-EVM PeerDAS, the "sharding" vision is effectively being realized. And L2s can give additional and different kinds of gains in speed on top. Whisper is now Waku ( docs.waku.org/ ), and already powers many applications (eg. railway.xyz/, status.app/ just to name two I use). Even outside of Waku, the quality of decentralized messaging has increased. Fileverse (decentralized Google Docs and Sheets alternative: fileverse.io/ ) has seen massive gains in usability over the past year. IPFS is now highly performant and robust as a decentralized way of retrieving files, though IPFS alone does not solve the storage problem. Hence, there is still room to improve there. All of the prerequisites for the original web3 vision are here, in full force, and are continuing to get stronger over the next few years. Hence, it's time to buidl, and buidl decentralized. Fileverse is an excellent example of the right way to do things: * It uses Ethereum and Gnosis Chain for what they are good for: names, accounts and permissioning, document registration * It uses decentralized messaging and file storage to store documents and propagate changes to documents * The application passes the walkaway test: github.com/fileverse/walk-aw… (even if Fileverse disappears, you can still retrieve them and even keep editing them with the open source UI) This is what we mean by "build a hammer that is a tool you buy once and it's yours, not a corposlop AI dishwasher that requires you to register for a google account and charges a subscription fee per month for extra washing modes, and probably spies on you and stops working if you get politically disfavored by a foreign country". If you think this criticism of corposlop is hyperbolic, well turns out, it's literally a concatenation of these three: * mein-mmo.de/en/user-buys-new… * theguardian.com/technology/2… * irishtimes.com/world/us/2025… In 2014, decentralized applications were toys, hundreds of times more difficult to use in web2. In 2026, fileverse is now usable enough that I regularly write documents in it and send them to other people to collaborate. The decentralized renaissance is coming, and you can be part of making it happen.

We just pulled 12 weeks of @Pumpfun user data, and the weekly retention numbers are pretty interesting when you compare them to Web2 norms. Here’s the rough benchmark for Web2 consumer apps: 1. Fintech: Week-1 retention: 10–15% 2. Gaming: Week-1 retention: 7–12% 3. Consumer Social: Week-1 retention: 20% 4. Week-4 retention across all categories: typically 5-10% 5. Week-8 retention: 2-5% ---- Here's Pump: 1. Week-1 retention: 24% 2. Week-4 retention: 12.4% 3. Week-8 retention: 11.4% This is Web2-level retention — sometimes better — for an app that scaled to millions of users and hundreds of millions in revenue at breakneck speed. With no ad spend and a super lean team. --- What about the bots? Web2 bot traffic tends to inflat the *top of the funnel (spam clicks, sign-ups, ad traffic), destroying retention. Onchain bots behave differently, though. - Many bots trade repeatedly (63% of Pump active addresses are recurring) - They show up in returning user metrics - They can be "price agnostic" customers ---- It will be interesting to see if these numbers hold up in a drawn-out bear market. That's why Pump Fun is on @the_defi_report "Watch List" We're providing a full data-driven update on Pump Fun with readers tomorrow If you'd like to have the latest (free) research hit your inbox when it's published, you can sign up below 👇

The three eras of crypto. Proof of work: 2009-2017 Programmability: 2017-2025 Privacy: 2025-203X

Announcing Fortytwo’s Swarm Inference A decentralized AI architecture that outperforms the top frontier models from the biggest labs: > ChatGPT 5 (OpenAI), > Gemini 2.5 Pro (Google), > Claude Opus 4.1 (Anthropic), > Grok 4 (xAI), > DeepSeek R1 (DeepSeek). Thread ↓

Similarly, funds that made fewer early-stage bets saw meaningfully lower failure rates, while high-volume portfolio construction was far more likely to lead to write-offs. Making @BigBrainVC low failure rate amongst portfolio companies despite a large portfolio more impressive.

Legend is the world’s first Trader vs. Trader arena. Trading is already the world's biggest video game - it’s time to make it multiplayer. Today, we’re proud to share that we've raised $1.5M in pre seed funding, from top-tier institutions and angels to build the best competitive-social layer for trading. Become Legend.

We’re committing $5M to push robotics beyond its limits. The BitRobot Grand Challenges Fund backs ambitious teams tackling the hardest problems in robotics, with open challenges and clear benchmarks to measure progress in embodied AI. ↓