Currently, when it comes to meme trading, platforms all want to solve one problem: bundling.
That is, an entity uses dozens of wallets to simultaneously buy in during the first few blocks of a token's launch, sweeping up the cheapest batch of chips at the opening in one go. On-chain, it looks like a bunch of new addresses are buying, appearing like natural demand, but in reality they are all controlled by the same person, with a cost basis so low it's almost nothing.
For example, in the discussions over the past few days, analysts found several groups running scam operations on Pons V2, using bundling: after launch, dozens or hundreds of wallets buy in immediately, then dump, ending the battle in 16 seconds, then move on to the next one.
This is also why BUN rose to a new high. Today, against the backdrop of a sluggish on-chain market, Bundle Cat (BUN) hit a new high against the trend, with FDV surpassing $100 million.
Let's talk about this project.
How to solve bundling
Behind $BUN is a launch protocol called Mosh. It introduced a small innovation.
The bundling problem ultimately comes down to "this batch of goods is destined to be sold." Mosh's approach is to crowdfund it, lock it up, and make the act of "selling" structurally disappear.
Step one, crowdfund this bundle. Each launch first opens a fundraising round with a target amount, and only launches once the target is met; the official documentation states there is no deadline, and if not fully raised, it stays open, during which contributors can sell their shares back at the current curve price. After the raise is complete, Mosh takes a one-time 5% launch fee, and the remaining funds are partly used to buy the coins off the Pons curve, with part of the ETH kept in the treasury as the agent's cash. For BUN this time, the team contributed 8 ETH themselves.
Step two, launch and lock-up are completed in the same transaction. The bundled chips are crowdfunded by participants, and at the moment of launch they are locked into a non-custodial contract treasury, including the Mosh team, no one can withdraw them.
Step three, convert contributors' returns from token price to fees. Funders do not get tokens, nor can they redeem their principal. The official risk page states it plainly: after the token launch, funders can neither retrieve their principal nor transfer their shares. What they get is something else: as long as this pool is still trading, they receive a proportional share of the creator fees.
How is the money calculated? The Pons pool charges a 1% base fee per transaction, Pons keeps 30%, and the remaining 70% is returned to the token's creator. For Mosh launches, that means returning it to this bundle. The launcher can also stack an additional creator fee on top; the higher the stack, the faster funders recoup, and the harder the token is to trade. Together these constitute the bundle's fee income, which the contract splits on the spot: Mosh 20%, funders 80%. On Pons, this money is settled in ETH (the paired asset), not paid in the launched token.
According to the project's fee calculations, at 0% creator fee, funders get 56 cents per $100 of trading volume, and recouping requires about 179 times the bundle size in trading volume; at 0.5% creator fee it drops to about 104 times, and at 1% it drops to about 74 times.
So funding a Mosh bundle is like buying a perpetual fee bond: it feeds on trading volume, and you hold no position you can dump. Previous bundlers wanted to pump and exit; Mosh funders, i.e., the new bundlers, want a token that can keep trading for months.
How Mosh solves market making
Micro-cap tokens rely almost entirely on passive AMM liquidity. No one manages inventory, no one supports the price during crashes, no one adjusts spreads. Professional market makers do these things, but they are expensive, hard to onboard, and often a black box; small teams are not even sure whether that "market maker" is supporting the market or eating them.
Mosh's answer is: at the moment the launch ends, the market maker is already on duty with inventory.
The batch of tokens locked in the treasury cannot be withdrawn, but can be bought and sold within contract limits. Sell when there is demand, accumulate ETH, and buy back with ETH when selling pressure comes; Mosh calls this "structural buying." Multiple agents each manage their own inventory and trading budget, can run different strategies, and together they form what the official calls a swarm.
All constraints are written into the contract, not relying on the team's verbal guarantees: each treasury can only trade one pair, with hourly limits on buys and sells; operators can sign and pay gas, but cannot withdraw tokens; pausing an agent will not release inventory to the operator or funders.
The largest wallet on the market has become a counterparty that cannot leave; the cheapest batch of chips at launch has become an inventory that can be used for months.
BUN live: how that 71% operates in the treasury
BUN is the first trial run of this mechanism, and it was funded by the team itself, so it is a clean base case.
Raised 8 ETH, of which 4.24 ETH bought 714.3 million BUN from the curve, about 71% of supply; the remaining 3.76 ETH stayed in the treasury as the agent's cash. At launch, no creator fee was set, and there was no team share; funders receive fees at the default ratio.
BUN total supply is 1 billion. According to the project page and community statistics from mid-September, it is divided into three parts:
· AI agent treasury, 71.4%, can buy and sell within contract limits, cannot withdraw, cannot transfer to anyone · Pons platform, 8.2%, basically locked forever · Actual circulating supply, 20.4% So BUN's real market cap is only 20% of FDV.
How the money inside the treasury moves
The capital for market making comes from two places: the BUN in the treasury (inventory) and the remaining 3.76 ETH (ammunition). If one side is missing, only one-sided operations are possible: with only tokens, you cannot catch it when it falls; with only money, you have nothing to give when it rises.
That 71% is this treasury inventory, and the daily operations in the treasury go like this:
· When people rush to buy, the agent sells part of the BUN and collects ETH into the treasury;
· When selling pressure comes, it uses the ETH in the treasury to buy the tokens back;
· Profits from buying and selling stay in the treasury, do not flow to the team's pockets, and can be used for the next round of quotes.
Note that it does not throw tokens into the pool as LP. It holds spot and actively quotes and trades within limits: quotes are generated off-chain, and only trigger when the price hits a certain Uniswap tick. What the treasury limits is "how much can be bought and sold per window."
How much can be bought and sold per hour is locked by contract limits; the tokens in the treasury cannot be withdrawn, nor transferred to the team or funders. So if the agent sells more than it buys back, the BUN in the treasury will decrease and ETH will increase; the treasury balance itself is one of the most important indicators to watch.
What the AI decides in the middle
The AI in BUN does only one very specific thing: watching a batch of inventory that cannot run away, deciding when to buy, when to sell, how much to buy and sell, and how wide to set the spread.
Buy and sell BUN within limits; use the ETH from selling to buy back; adjust spreads and direction; provide counterparties to the market; later it also plans to manage Uniswap v4 range liquidity, that is, spread liquidity near the price. These are all things the AI can do in BUN.
The AI is indeed working. The tokens in the BUN treasury are really changing hands, and the balance keeps moving. The team disclosed a set of early data: the agent bought about 1.22 million BUN (spending 2.4 ETH), and sold about 1.26 million BUN (getting back 4.2 ETH).
The math: recouped in the first minute, about 8x in 23 days
We just mentioned the 179x formula; 8 ETH would require about $3.8 million in trading volume to recoup. The team claims that after the opening buy tax (99% decaying to 0) was merged into fees, it covered all 8 ETH in the first minute.
After that, it continued to accumulate through ordinary trading. BUN has generated about $30 million in trading volume, and funders have cumulatively received 64.48 ETH, about $173,000, equivalent to 8.06 times the bundle principal.
At the same time, the team plans to use BUN's fee flow through the Agentic Liquid Fund (ALF) to buy back BUN, connecting platform revenue to the token price. But this is still a plan and not yet live.
Who is the team
Behind Mosh is a company called UV Labs, founded in 2022, 6 people, a US company.
Founder Justin Bebis is also a serial entrepreneur in crypto, early on at Fantom; in 2022 the Fantom Foundation officially published an introduction about him. His resume includes Reaper.Farm (automatic yield compounding), Granary (lending), Reliquary (using financial NFTs to represent positions and incorporating participation time into rewards). He later did Cod3x, also quite well-known, combining AI and DeFi operations into one entry point. Co-founder Sean Kramer was previously a professional esports player; coming to meme trading is like a dimensionality reduction attack.
Currently, the market seems to be buying into BUN's innovation. On the second day after the meme scam factory was exposed, BUN hit a new high against the trend. Will bundling governance + AI market making become standard for later memes?
