Three Great Scenes And No Bad Ones

The tech industry is buzzing after SpaceX announced its finalized acquisition of Cursor. The $60B price shattered the prior record for an acquisition of a venture-backed company and a lot of focus is now on explaining the company's meteoric growth. There is no single reason why Cursor has been as successful as it has been, but I will offer an under-appreciated factor in its early success: a product which adhered to a winning formula.

The Formula

Howard Hawks described a good movie as "three good scenes and no bad scenes." I think that many great early-stage products are similar: they have a few great features and nothing frustrating. That's it. A few early products which I personally loved spring to mind:

• Arc: vertical tabs, Cmd+Shift+C, split view; it was built on Chromium and there were no major regressions as compared to Chrome.
• Granola: notes, one-click share, ask anything; it didn't have a bot enter each meeting with me (I was not previously using a notetaker partly for this reason).
• and, of course, Cursor: Cmd+K, Cmd+L, tab; forked VSCode and invoked APIs of the same models I was otherwise using.

The list could go on and on, but I'd prefer to offer three good examples and no bad ones. In the case of Cursor, even the previous products which the team was working on followed this formula:

Why It Works

I suspect that Hawks had a ruthlessly pragmatic view of the role that movies play in peoples' lives: they are there to (1) entertain for 90 minutes and (2) create memorable moments which can be shared. And there is already a kind of formula in typical character development and plot arc. It's unlikely that you'll have a viewer perched on the edge of their seat for the full duration of the movie, and that's actually quite alright; it's surprisingly likely that that viewer might not even make it through the movie to get to the good bits.

The same is true of catchy products. They don't need to solve every one of their customers' problems, they just need to clearly and memorably solve a few and introduce no regressions to how they work or live.