I've built fast and killed projects when they didn't make sense to maintain. The "10% coding, 90% everything else" reality is something I keep coming back to.
Hosting costs killed my first app. The temptation to just keep building because it's cheap now, that's the trap.
Really, the Spotify trajectory is what happens when "can we?" replaces "should we?"
I'm curious - it seems to me that one of the benefits of AI is the ability to make the implicit explicit. With that said, why can't practitioners use the existing AI tools to answer the question of what should be built (while respecting the user) without necessarily sacrificing on the explosion of product development velocity. You mentioned "When building is cheap, restraint becomes the most valuable thing a team can practise" - I guess I'm trying to understand from whose perspective does this ring true? With the cost of product development reduced, prototyping velocity increased, and the ability of these LLMs to process more and more context - what's stopping teams from being able to deploy agents that have the ability to predict the best feature to build given a set of constraints?
Angèle, the restraint argument resonates so much!
I've built fast and killed projects when they didn't make sense to maintain. The "10% coding, 90% everything else" reality is something I keep coming back to.
Hosting costs killed my first app. The temptation to just keep building because it's cheap now, that's the trap.
Really, the Spotify trajectory is what happens when "can we?" replaces "should we?"
100% every time I open Spotify, something new pops up
it is like in Chinese all-you-can-eat - more stuff is not always better.
I love comparaison with the French cook J Robuchon !!
I'm curious - it seems to me that one of the benefits of AI is the ability to make the implicit explicit. With that said, why can't practitioners use the existing AI tools to answer the question of what should be built (while respecting the user) without necessarily sacrificing on the explosion of product development velocity. You mentioned "When building is cheap, restraint becomes the most valuable thing a team can practise" - I guess I'm trying to understand from whose perspective does this ring true? With the cost of product development reduced, prototyping velocity increased, and the ability of these LLMs to process more and more context - what's stopping teams from being able to deploy agents that have the ability to predict the best feature to build given a set of constraints?