Many technical tools for organization require users to think of the schema up front.

Airtable is a brilliant app, but it still requires users to think about the schema when they create their airtable.

One of the reasons they've invested so much in templates and guides for many verticals.

But at the beginning even knowledgeable users don't necessarily know the schema they want to use.

They want to sketch ideas, jot down data, collect disparate things to then organize later.

Computers by default demand precision; they demand schemas.

Tools built in the computer often force the human to work on the computer's terms.

Tools like spreadsheets are built in the computer but are almost infinitely flexible, allowing you to jot down any unstructured data.

But they take time and effort to wrangle into a structure later.

And it's always easy to fall back off the structure.

They're the nosql of UI-first databases.

But that lack of structure will bite you later if you try to do anything scaled.

LLMs can do all kinds of fuzzy structured things.

For example, take a picture of the books on your bookshelf and ask for a JSON representation, most LLMs today will do a great job!

With this new magical duct tape, can we make tools that allow humans to act like humans, but with the benefits of scaling that only computers can offer?