What I find interesting about this image is that it doesn't follow the narrative of "everything goes back to the Bitcoin mainnet." It’s more like breaking something down: Bitcoin is at the very bottom, doing the final settlement. GOAT L2 is in the middle, responsible for executing this smoothly. Above that are ClawUp, ERC-8004, x402, AgentKit, etc., which cater to the actual use by individuals, institutions, and agents. This also aligns with the data shared in the previous tweet. In the 9072 experiments from the Bitcoin Policy Institute, AI tends to favor Bitcoin in the "long-term value preservation" scenario, but in the "daily payment" scenario, it is more likely to choose stablecoins. When I saw this image, several scenarios that can already happen came to mind. For example, an agent calls a data interface, and the API directly returns a 402. It pays the money itself, requests to continue, and the task keeps running. Another example is using the same set of tools to create wallets, transfer funds, swap, bridge, and register identities. Once the tasks are completed, the interaction records and reputation can still remain on the chain. You will gradually feel that money shouldn't be squeezed into one layer to operate. Bitcoin does the final settlement, while stablecoins handle high-frequency payments. GOAT L2 connects the execution, identity, payment, and verification in the middle. Looking at it this way, the image is no longer just a concept, but a path that has already started to run: calls can happen, payments can be completed, identities can be left, results can be verified, and finally, value can be anchored back to Bitcoin. If these agents really start to independently call services, complete payments, and handle settlements, then money will naturally start to run in layers. Whoever first smooths out this link, many transactions may directly occur on it afterward.