Trying to validate an idea around AI Agent Optimisation for websites

Given the rise of agents on the web, it seems like websites would need a framework for having them interact/transact on their websites securely, without flagging them as malicious bots

Would love to get feedback on possible ways to solve this problem from folks here