The latest news—here are a few quick updates.
There are two things worth watching in AI and on-chain tools. First, ai16z founder Shaw said that 90% of the reverse engineering of all Pump.fun contracts is complete, with work expected to wrap up that evening. He also declared, “Open-source everything, or it will be forced open-source.” Take: Taking apart someone else’s contracts, understanding them, and then making them public is a learning resource for developers and pressure on the platform. Code transparency is becoming a competitive edge. Second, Four.Meme launched a “Binance Agent ID exclusive mode” for AI agents on BNB Chain, supported by Binance Wallet. During the bonding-curve phase, trading is only open to addresses holding an Agent ID. Take: AI agents are starting to get dedicated channels for issuing tokens. The $BNB ecosystem is competing for a place in the AI narrative, but new mechanics bring new risks—understand the rules before jumping in.
AI training strategies are interesting, too. Kimi’s K2.5 uses Agent Swarm to let a lead Agent recruit and direct sub-agents. K2.6 expands this to as many as 300 agents, but its reinforcement learning method, PARL, trains only the “boss”; the sub-agents are frozen. Take: Is teaching only the manager, not the workers, a way to save computing power—or a bet on division of labor? Worth keeping an eye on.
As for market moves and fund flows, according to monitoring by Lookonchain, a smart-money trader opened a long position in 26,974 $ETH after the price fell, worth about $63.88 million. The position is currently up $842,000, and the address has a 100% win rate. It previously made $3.3 million trading NEAR. Take: A good-looking win rate doesn’t mean they’ll get it right next time. Before copy-trading, think about how much of a drawdown you can handle.
On the revenue leaderboard, according to DefiLlama data, Papertrade generated $8.5 million in revenue over the past seven days, ranking fifth among all protocols, behind only Tether, Circle, Pump.fun, and Hyperliquid. Take: A protocol’s ability to generate revenue reflects real demand, but one week of data doesn’t tell us much about the long term.
In traditional markets, Mercuria is suing the 282-year-old Baltic Exchange. The case is set to be heard in the UK High Court. The dispute is about whether the exchange should continue publishing benchmark freight rates for oil tankers traveling through the Strait of Hormuz after the U.S. and Israel went to war with Iran. Take: A single freight-rate index can influence tanker costs and pricing power. Disputes over the rules of traditional markets can also affect commodities and sentiment in risk assets.
Are you more interested in new developments like AI agents issuing tokens, or smart-money activity on $ETH ? Let us know in the comments.
There are two things worth watching in AI and on-chain tools. First, ai16z founder Shaw said that 90% of the reverse engineering of all Pump.fun contracts is complete, with work expected to wrap up that evening. He also declared, “Open-source everything, or it will be forced open-source.” Take: Taking apart someone else’s contracts, understanding them, and then making them public is a learning resource for developers and pressure on the platform. Code transparency is becoming a competitive edge. Second, Four.Meme launched a “Binance Agent ID exclusive mode” for AI agents on BNB Chain, supported by Binance Wallet. During the bonding-curve phase, trading is only open to addresses holding an Agent ID. Take: AI agents are starting to get dedicated channels for issuing tokens. The $BNB ecosystem is competing for a place in the AI narrative, but new mechanics bring new risks—understand the rules before jumping in.
AI training strategies are interesting, too. Kimi’s K2.5 uses Agent Swarm to let a lead Agent recruit and direct sub-agents. K2.6 expands this to as many as 300 agents, but its reinforcement learning method, PARL, trains only the “boss”; the sub-agents are frozen. Take: Is teaching only the manager, not the workers, a way to save computing power—or a bet on division of labor? Worth keeping an eye on.
As for market moves and fund flows, according to monitoring by Lookonchain, a smart-money trader opened a long position in 26,974 $ETH after the price fell, worth about $63.88 million. The position is currently up $842,000, and the address has a 100% win rate. It previously made $3.3 million trading NEAR. Take: A good-looking win rate doesn’t mean they’ll get it right next time. Before copy-trading, think about how much of a drawdown you can handle.
On the revenue leaderboard, according to DefiLlama data, Papertrade generated $8.5 million in revenue over the past seven days, ranking fifth among all protocols, behind only Tether, Circle, Pump.fun, and Hyperliquid. Take: A protocol’s ability to generate revenue reflects real demand, but one week of data doesn’t tell us much about the long term.
In traditional markets, Mercuria is suing the 282-year-old Baltic Exchange. The case is set to be heard in the UK High Court. The dispute is about whether the exchange should continue publishing benchmark freight rates for oil tankers traveling through the Strait of Hormuz after the U.S. and Israel went to war with Iran. Take: A single freight-rate index can influence tanker costs and pricing power. Disputes over the rules of traditional markets can also affect commodities and sentiment in risk assets.
Are you more interested in new developments like AI agents issuing tokens, or smart-money activity on $ETH ? Let us know in the comments.