Thesis: AI is a new way of building startups

The thesis

I think AI will spawn a new way of building applications.


At Rows.com, we built a Charts solution using an open-source Charts library. We integrated it by integrating the library deep with Rows platform: schema, persistence, FE app framework. We surely kept appropriate structures and abstractions for our technology process at the time, which was “designers and engineers build the simplest UX possible for the user, which is a polished UI, and maintain it”. For our solution to cover the most common chart types, it took us many many months and several engineers.

The problem is that the technology process above did not include the possibility that a prompt could translate fuzzy to syntax (human text to code). If that UX becomes possible, then you can offer a new, faster experience. You have a generic html component, and when the user asks to do a Chart of XY dataset, with such and such property, you ask your ML to render that as a Chart of library X. And “that’s it”. Adding new Charts is faster, and you can always later push the syntax into a generic visual UI too.

Key benefit