What Caught My Eye goes out every Thursday night. It's a curation newsletter -- a shortlist of what was worth paying attention to that week. The problem wasn't finding good material. It was deciding which good material actually belonged in the edition.
By Thursday, there would be thirty saved links. The edition could hold five or six. The decision process was slow, second-guessing, and mostly happened in a text editor with no structure. Everything looked equally promising at the moment you saved it. Nothing looked equally promising at the moment you had to cut.
The first version of what became Tastoria was a script that scored articles against each other. It didn't make the decision. It made the decision easier to see. Which pieces held up? Which lost something when you put them next to each other? Which ones had taken, which had stopped short?
That was useful enough to keep running. Then useful enough to improve. Then useful enough to wonder whether it should be a product. The question was: is this Thursday night problem specific to one curator, or is every independent curator dealing with some version of it?
Every one we asked had the same Thursday night. Different beats, different voices, same structural problem: a shortlist that's too long, a decision that's too hard, and no tool designed for the editorial act itself.
That's the product. Not a newsletter platform. Not an AI writing tool. An editorial decision layer. The thing that lives between saving a link and clicking send.