Tuxedo Wa-Kamen on Nostr: So I can have a repository of "reasonable features" for an application: Spellcheck, ...
So I can have a repository of "reasonable features" for an application: Spellcheck, Load/Save, connecting to a remote API, etc.
By adding some of these features, I can show a user my current interpretation of what they want, which then allows them to make a decision and add it to the specification ("no cloud stuff please", "saving is good", "I want unicorn icons", etc.)
So basically I can "talk"/"write" at an abstraction level that I care about (in contrast to a level that my code editor understands).
And... from a technology view, my approach is almost "boring". I simply add markers that describe features to a spec. So no "text-to-GPT-to-bad-comprehension" pipeline.
I think something like this has been missing in many computer-related applications/tasks:
Comprehensive domain models that allow communication with the machine on a higher abstraction level.
By adding some of these features, I can show a user my current interpretation of what they want, which then allows them to make a decision and add it to the specification ("no cloud stuff please", "saving is good", "I want unicorn icons", etc.)
So basically I can "talk"/"write" at an abstraction level that I care about (in contrast to a level that my code editor understands).
And... from a technology view, my approach is almost "boring". I simply add markers that describe features to a spec. So no "text-to-GPT-to-bad-comprehension" pipeline.
I think something like this has been missing in many computer-related applications/tasks:
Comprehensive domain models that allow communication with the machine on a higher abstraction level.