A portfolio assumes a person arrives, clicks around, and decides what to make of you. My bet is that, increasingly, the first visitor will be an agent doing that groundwork for someone else. What should a personal site become when the first question comes from a system?
I think people may become callable collaborators in AI workflows: something like a plugin, but with judgment, taste, and a history that can't be reduced to an endpoint. The clumsy version turns a person into a service menu. The interesting version helps an agent recognize why this particular human matters to this particular problem.
This site is my first experiment. A person can follow the projects and stories at their own pace. An agent can read a public map of the same work, ask a grounded question, and bring back a useful route into the site. The pages and the interface should lead to the same person, not two different versions of me.
That is why there is an assistant beside the pages and a public MCP interface behind them. I want an agent to understand the elevator pitch for each project, the connections among them, and the kinds of questions I like to work on. A job title alone can't show that. The trail from trading automation to market research to strange Moon stories might.
An agent might decide that a focused review, a conversation, or a deeper collaboration would help. It should be able to find the relevant path and prepare a thoughtful handoff. It should not invent my opinion, send someone's details without consent, or claim a booking happened because it found a link.
This is a hypothesis, not a finished model of the future. The point is to make my work legible to new kinds of visitors while leaving the part that makes it mine—the questions, connections, and human judgment—fully in view.