Where AI earns its place in an app

Almost every app idea we hear today includes AI. Sometimes it is the whole point. Sometimes it is there because it feels like it should be. Across our own five apps we have learned one test that separates the two: AI should remove real work from the person using the app, and leave them in charge of the result.

OwnSites: AI does the first draft

OwnSites turns a one-sentence description into a complete website. Writing every page from nothing is exactly the kind of work most small businesses never get round to. But the owner stays in charge: they review every page and change any of it by asking for the change.

ProfVault: AI reads the form so you don't have to

Job applications ask the same questions in a hundred different layouts. ProfVault reads each form, works out what every field is asking for, and fills it from the documents you have stored. The tedious part disappears. You still review what it filled in before you submit.

Homeroom: AI writes, then gets checked twice

Homeroom writes new reading passages for kids so they never run out of practice. Every generated passage has to pass two checks before a family sees it. A set of rules checks the length, the questions and the answer choices. Then a second AI model answers the questions without the answer key, and any question it gets wrong is treated as unclear and the passage is thrown away. Finally, a parent approves the passage before their child reads it.

Brandreel: AI drafts the film, you direct it

Brandreel writes a script and lays out the scenes of a brand film from a short brief. Then you change any scene in an editor with a live preview before anything is rendered.

What we leave out

We do not add AI where a simple rule does the job better. Homeroom's worksheets, for example, are generated by rules, not by a language model, because rules are predictable, instant and free to run.

If your idea has AI in it, we will help you find the part where it earns its place, and build the checks that keep it honest.