Something quiet happened this week that rewrites the whole story we've been telling about AI and creativity. Bluesky dropped Attie, a tool that lets users build their own algorithmic feeds using natural language. Days later, Suno released v5.5, explicitly framing its update around customization over raw generation quality. The pattern is unmistakable: the first wave of AI products promised to do things for you. The second wave promises to do things as you.

Curation as the New Creative Labor

This is not a minor UX pivot. It is a philosophical repositioning. The Stanford study warning about AI sycophancy in personal advice sits directly adjacent to this story. When AI mirrors your preferences back at you without friction, it does not liberate taste. It calcifies it. Attie's bet is that building the feed, rather than just consuming it, reintroduces the necessary resistance. You have to decide what you want before the machine can deliver it. A 2023 paper in Nature Human Behaviour by Bail et al. found that algorithmic personalization reduces exposure to cross-cutting information by up to 40 percent, not because platforms intend it, but because frictionless preference-matching is the path of least computational resistance. Giving users the syntax to describe their own feeds is, in a weird way, more demanding of them.

The Taste Economy and Who Gets to Build It

The deeper implication is economic. Custom feed infrastructure is essentially taste-as-a-service, and taste has always been a form of capital. When Meow Wolf pivots from the experience economy to the transformation economy, they're describing the same phenomenon in physical space: people no longer want content delivered at them, they want to be changed by it. The companies building the rails for personal curation, not the content itself, are the ones with durable defensibility. , which makes the Attie launch look less like a product drop and more like a category staking.