Build the experience, not the screen

Context

Two things I built for myself started from the same frustration.

The first was a job search tool. Job platforms each decide what I see, in what order, and why. I wanted opportunities from several platforms in one place, with my own rules for what deserves attention. The tool matches each ad against profiles I write about what I want next, not against my past. What counts as interesting is a setting I own.

The second was Email. Each mail provider shows me mail the way it thinks everyone should see it. In AlanOS, every account arrives in one mailbox, sorted into Inbox and Other by my rules, and I move a sender between them whenever I like. The providers still deliver the mail. They no longer decide how I read it.

Neither tool is clever. In both, someone other than the provider decides the experience.

Screens are not the point

A screen is only the visible part. The experience is everything underneath: what gets my attention first, what is left out, what happens when something new arrives, what one click does.

A provider designs one experience for millions of people, and it has to serve the provider's goals as well as the reader's. Reasonable, but it is still someone else's answer to "what matters to you?"

I think of it as building blocks. A finished set gets you one model, designed by someone else, assembled by the instructions. The blocks let you build what you actually need. For a long time building your own was too expensive, so we all bought the set.

What changed is the cost

Models have made it fast and reliable to assemble an experience from parts. What used to be a project no longer has to justify a product team.

Cheap also means replaceable: if my way of reading mail stops fitting, I change it and lose nothing.

Own the data, then shape the view

That is the condition for everything else. Email works because the mail lives in my own domain first, and the interface is one view of it. I can build another tomorrow without asking anyone.

A professional network's feed, built on my own copy of my own connections, could show me what I care about instead of what keeps me scrolling. My finances already work this way, with categories designed for my questions rather than a bank's taxonomy.

The model I find most attractive is a product that ships its data layer as a first-class product and treats its own interface as one option among many. The provider keeps doing what it is good at: storage, delivery, the network itself. The person gets an experience shaped around their own attention.

The catches I can see

This is a working theory, not yet tested outside my own tools. These are the objections I take seriously.

• Access is the provider's choice. Many services let data out through exports or APIs. Many restrict it, and that can change without notice.
• Incentives point the other way. A business that earns from attention has little reason to let people design that attention away.
• An experience that acts needs limits. Reading is low risk; sending, paying or deleting is not. In AlanOS, accounts connect read-only first, Email never deletes mail, and nothing leaves without my yes.
• Most people do not want to design anything. They should not have to. A person describes what matters and a model assembles the rest, the way I describe the job I want instead of building a search engine.

What I want readers to take from this

I can craft my own experiences on top of my own data, cheaply enough to change them whenever they stop fitting. I think other people should be able to as well, eventually without writing any code.

Ask it of any tool you use every day: who decided what you see first? If the answer is not you, changing that may be cheaper than you think.