Our agents successfully sniped 70% of the Bundle Cat supply. This supply is locked forever, and will be used to make the Bundle Cat market indefinitely. Thank you all for participating! CA on Pons / Robinhood: 0x07EBB29a38Fbcb41563817e5E19f2ceC619C90D2

UV now has 27 live agents managing over $20,000,000. It's literally day 1. Check out mosh.trade today!

Most people looking at @BundleCatAI / @Moshdottrade are focused on one thing: the Liquidity Swarm. But I think there's a more interesting experiment hiding underneath Mosh's market-making infrastructure. What happens when a memecoin's trading fees start financing an autonomous investment fund? That's the idea behind Agentic Liquid Funds (ALF), and it could become an important part of Mosh's broader economic model. Let me explain. --- ➠ First, follow the money. Remember how Mosh works? Instead of allowing bundlers to accumulate a massive token allocation and dump it whenever they please, Mosh commits that inventory to vaults managed by AI agents. For $BUN, roughly 71.4% of supply sits in the Liquidity Swarm. The agents use that inventory to buy, sell and manage liquidity around the existing AMM. Funders sacrifice access to their original capital in exchange for trading-fee income. But here's what makes BUN different. The team funded BUN's opening bundle themselves. According to @justinbebis, the initial raise was 8 ETH, with 4 ETH used to purchase the opening bundle. The team reported recovering its funding through fees within the first minute and the team claim they've now earned approximately 10× their bundle investment in fees. Instead of holding a large, freely withdrawable token allocation, the team receives income linked to BUN's trading activity. The longer the market stays active, the more fees the bundle can potentially generate. And that brings us to ALF. --- ➠ The second engine: Agentic Liquid Funds Mosh doesn't intend to let all that fee income sit idle. The team's proposed next step is to use eligible bundle revenue to fund a different class of financial agents. Think of it as giving an AI trading desk its own investment budget, financed by the trading fees generated through Mosh. The intended mechanism is straightforward: BUN trading → bundle fees → ALF capital → autonomous trading → ecosystem investment There are now two distinct engines. - The Liquidity Swarm manages a token's market using inventory committed during its launch. - ALF would manage capital generated from fee income, potentially trading BUN and other Mosh-aligned assets. Team has described the idea as extending buyback tokenomics with an active trader attached. An ALF introduces discretion through an automated strategy. It could theoretically accumulate during heavy selling, preserve $ETH when conditions are unfavorable, or deploy capital across several eligible tokens. And unlike burned tokens, assets purchased by a fund can potentially be sold again. ALF is not automatically a buyback-and-burn mechanism. It's an attempt to make fee-generated capital productive. -- ➠ The overlooked BUN connection Creators launching on Mosh may be able to whitelist BUN holders for bundle funding. That gives BUN a role beyond being the first Liquidity Swarm token: potential access to selected launches How it could work: - Hold BUN → qualify for selected bundles - Fund launch inventory (held in agent vaults) - Receive a claim on trading fees Details aren’t final. A whitelist doesn’t guarantee allocation, or profit. Still, it positions BUN as an 'access asset' inside Mosh’s funding system. Together with ALF, Mosh is testing two BUN-linked paths: - BUN holders may get access to future bundle funding - Team-controlled fees may fund agents that can buy BUN (and other ecosystem assets) At scale, the pitch is a very interesting flywheel: More launches → more trading → more fees → more ALF capital → more ecosystem investment. Personally, if Mosh can demonstrate that both operate sustainably, it could have something more substantial than an AI-powered memecoin launch mechanism. But until the capital flows and investment results are verifiable, ALF remains an intriguing extension of the original experiment rather than proven token value accrual. NFA. DYOR.

Want to know how our latest tech works? Justin wrote an article diving deep into every aspect - give it a read!

Justin spoke with the MCG team for over 30 minutes last week about how our new product Mosh will solve on-chain market structure. Very informative!

Today on MCG $BUN | @BundleCatAI w/ @justinbebis Bundle Cat is the mascot for MoshTrade, a project that helps launch tokens with built-in market making agents Highlights: 00:56 - Justin's background 05:39 - UV Labs involvement 08:36 - What Mosh does 09:45 - Problem it solves 10:38 - Vault mechanics 12:17 - Agents trade Uni v4 18:10 - Tokenomics 20:29 - Deflationary mechanic 21:04 - Structural bid 22:45 - moshtrade launches 23:44 - Principal amount back in a minute 24:20 - Public launch soon 26:24 - Circulating vs FDV clarity

Today on MCG, presented by @MetaDAOProject 12:30 - $DOT @Usedotai w/@Stagedhappen 1:00 - $WEBSITE @notawebsite_rh w/@Metinzam2 1:30 - $BUN @BundleCatAI @uv w/@justinbebis 2:00 - Trader radio, follow us on @fomo @ChillTRD - fomo.family/r/ChillTRD @mynt_josh - fomo.family/r/mynt_josh

Yesterday I mentioned a project that might evolve into a new building block for token launches. Meet @BundleCatAI / $BUN. On the surface, BUN could pass for “just another memecoin.” In reality, it’s the first live test for something much larger running underneath: Mosh Trade and its Liquidity Swarm. Most small launches replay the same script: Launch → early wallets load up → hype tops out → they dump → volume dries up → liquidity thins → everyone moves on. And it’s not merely a price issue. Most micro-caps lean almost entirely on passive AMM liquidity. There’s no one actively working the book, managing inventory, tightening/widening spreads, shifting bids and asks, reacting to changing conditions. Yes, professional market makers do that, but they’re fucking pricey, hard to onboard, and often a black box (there's a reason why WM is so infamous). Simply put, smaller teams or even solo devs may not even be sure if the “MM” is supporting the market or simply trading around them. Mosh’s goal is to automate that whole role. Instead of paying a traditional market maker, a launch raises capital to fund a large opening bundle. That bundle buys a meaningful chunk of the token supply. Normally, one entity holding that much would scream “run.” Mosh inverts the logic. The bundled inventory is deposited into agent-controlled vaults, effectively the balance sheet for a swarm of AI market-making agents. So the pipeline becomes: Crowdfunded capital → large token bundle → agent vaults → AI Liquidity Swarm → adaptive bids + asks The launchpad and AMM continue to serve as the launch infrastructure and baseline liquidity. The swarm operates around it, buying and selling dynamically as conditions shift. In theory, the agents can manage inventory, tune spreads, react to order flow, and provide liquidity in places a passive AMM simply won’t. That’s also why the first POC of Mosh, BUN’s distribution looks so strange. About 71.4% of BUN supply sits across three swarm wallets. For almost any other token, that would be a nightmare. Here, the concentration is the feature. One key nuance, though: That 71.4% isn’t burned or taken out of circulation. It’s more accurate to think of it as swarm-managed inventory. Agents can sell BUN into demand, build up $ETH, then recycle that ETH to bid BUN when sell pressure hits. Over time, the intended loop is: trade → earn fees/PnL → recycle capital → deepen liquidity → support more volume → earn more fees And this is where Mosh becomes more interesting than simply "AI market making." The capital used to create the bundle is meant to remain productive indefinitely. Funders provide the initial capital and, instead of simply withdrawing their principal later, gradually earn it back through trading fees. Mosh's model estimates breakeven at roughly 100× cumulative trading volume relative to the original funding, with everything above that becoming return (this subjects to change as the official docs updated with latest figures from performance) So Mosh is effectively combining crowdfunded market making + permanent inventory + autonomous trading agents. And $BUN is the first live proof-of-concept (kinda similar position with $AI from LONG). One important clarification, though: Mosh is not really anti-bundler. @ponsdotfamily V2 already has aggressive anti-snipe mechanics, including a launch tax that starts near 99% and decays over the opening window, alongside launch-and-buy mechanics and whitelisted launchers. Mosh takes a different approach. It does not try to eliminate the bundle. It tries to make the bundle non-extractive. Instead of the biggest early wallet eventually becoming the biggest seller, Mosh wants that inventory to become the token's permanent market maker. Mosh turns the bundle from exit liquidity into AI-driven liquidity infrastructure. Currently, they plan to refine the agentic swarm parameters and launch the liquidity swarm publicly after.

Join our founder @justinbebis to talk about Mosh and the future of agentic trading.

This is potentially game-changing anti-bundler tech for the Robinhood (RH) eco Mosh (@uv) demonstrated its agentic market making tech with their own token "BUN" - 71.4% of supply was held inside three AI market-making vaults - 20.4% in initial liquidity - 8.2% locked Instead of bundlers cornering a thin float and dumping on everyone else, the agents hold enough inventory to trade against aggressive early demand The result after 16 hours: 1,185 holders, the largest public wallet owns just 0.5% & the biggest linked cluster holds 0.73%... essentially no obvious bundled supply There's a huge amount of opportunity here (if executed well) to leverage the distribution Mosh could become the anti-bundler layer for @ponsdotfamily , @longdotxyz or other RH launchpads, with launches paying fees used to buy + burn BUN If a portion of the fees are distributed from each BUN assisted token launch then this could go even more vertical than the platform it was birthed on Very early and the contracts still need verifying, but if the integration to broader launchpads materializes, this could send @MEADGod I'd recommend checking this one out & figuring out a way to integrate it into Pons... it'll improve the UX + it's already an ecosystem project King making legitimate innovative apps that come out of the Pons ecosystem will drive more value accrual to Pons imo

Today we're declaring war on farmers, scammers, and bundlers. Mosh is our new Defi protocol built on Pons that fixes lopsided risk/reward between bundlers and traders by pairing every token with an AI market maker. Watch this video to see how it works, and check out the first ever Mosh token - Bundle Cat solana:Em6ajM2xwou6y2YkfqAxsr5zMjmLepbJUtBaEzCypump: 0x07EBB29a38Fbcb41563817e5E19f2ceC619C90D2 Full platform soon! Thanks everyone for your participation.

Our team has been working around the clock to redefine how AI and Finance will intersect. LLMs combined with smart contracts allow us to facilitate financial engineering at an enormous scale. We've developed a novel HFT architecture and brand new harness to facilitate this.

A lot of VCs spent the first half of this year telling me AI wasn't going to make it into trading systems. Meanwhile finance is the #1 AI buyer on the planet and Jane Street & Citadel are buying low latency chips right off the line. 🤷‍♂️

Fun game, try this out :)

We're opening up a fun little slice of our internal benchmarking toolset. Point your agent at it and see how it scores against an expert - it's free and takes minutes!

A critical blocker for agentic finance is maturation of the infrastructure layer. You need the right abstractions and the right packaging for speed+scale. It's why we're focused on infra for @uv and also why we're excited about @CambrianNetwork's launch today. Congrats!

'Open Weights' is now a marketing term, not a technical one. OSS companies love changing the definition of 'Open' and AI is no different. MiniMax published one of the best video/audio models but it's banned in dozens of countries. Learn more:

Super fun to have you in the Apollo program. Keep building! Check out UV labs for anyone that hasn’t.

Our founder @justinbebis had an incredible day with @StanfordSBA, @0G_labs, and a host of incredible founders and investors last week in San Francisco for the Apollo graduation. Huge thanks to everyone that made it possible!

AI models have a hammer, and they think everything is a nail. Rigorous post-training takes this to the extreme - RLVR allows models to funnel a wide breadth of inputs to high-quality outputs with greater reliability. The cost is a little bit (or a lot) of over-fitting. Model architectures, inference services, and training techniques are evolving around this reality, but there's one fundamental truth we can't ignore: The AI industry needs to specialize. Everyone is building coding models and expecting every other industry to use them effectively. Our team is building a specialized inference endpoint for finance & developing training techniques using our massive corpus of agentic trading data. Interested? Reach out!