Zeno Rocha, co-founder of the email delivery service Resend, describes a shift in how developer tools compete for adoption. Agents, not humans, now read documentation first and decide which service to recommend or call. Resend has begun writing and structuring its documentation with that audience in mind, treating the agent as the first reader and the human developer as the second.
This surfaces a design challenge that sits outside the product itself. The question is no longer whether a human finds your docs clear; the question is whether an agent can extract enough signal from them to make a confident recommendation. Rocha calls this answer engine optimization, a parallel to search engine optimization but aimed at language model selection rather than search ranking. The implication for product teams is that legibility to an agent and legibility to a human are different targets, and optimizing for one can work against the other.
For designers and product leads, the tension is in trust and attribution. A human developer who follows an agent's recommendation may have no idea why that service was chosen. The selection happened upstream, out of view. That changes how a product earns and keeps trust, and it raises questions about what confidence the agent should signal when it makes a recommendation, and whether the human can see the basis for it.