How Should AI Agents Deal with Recommendation, Attribution, and Profitability Issues?

We have always regarded these agents as tools for completing tasks. Book this one. Compare that one. Summarize various options. Find the most suitable supplier. It's that simple. But once an agent starts influencing the actual purchase, installation, subscription or trust of the user - it is no longer just an assistant. It has become part of the information dissemination. This raises an unsettling question: How should funds flow at this level? The old network is filled with advertisements, affiliate links, search engine optimization, attribution pixels, review websites, and all those chaotic mechanisms with commercial intentions. The agent environment is different. Recommendation information may appear in the conversation content, summaries, or presented in a neutral "reasoning" form. This makes trust more fragile rather than more solid. So what should the rules be? Why can suggestions be made transparently? When an agent creates real business value, should developers also be compensated? Where is the boundary between useful suggestions and paid promotion? How should responsibility be determined when users have never clicked on traditional links? Is this an advertising network, an affiliate network, or a brand new protocol layer? The brief version is: "Let's convert agent recommendations into revenue." The precise version is: "How can we do this while not turning agents into 'small advertising machines' that only politely sell products?" I'm curious about how other builders view this issue.