this experiment from @OptimismGov is fascinating and fits perfectly with the concept of "protocol copilots" for DAO governance. i can already see there are three major tasks that these protocol copilots can take care of: > agentic governance, where copilots can make dynamic or subjective decisions, e.g., dynamic adjustment of protocol fees, even drafting proposals that humans then vote on (akin to decision markets from @robinhanson et al.). > agentic grant administration, where copilots can evaluate grant applications and adjudicate allocations (as in RPGF). > agentic treasury management, where copilots can actively manage DAO treasuries by deploying across DeFi, more on this here using x402 and ap2: but one thing is clear: these copilots must be verifiable. since these copilots are being run by operators, the protocol users and the protocol itself need guarantees that the copilot's outputs are honest and not manipulated by the operator.