What
Being able tor retrieve the usage limits.
Alternative: we infer how much % the current session consumed from the amount of tokens/dollars consumed.
Context
With autopilot, the idea is that I make a prompt and let the AI run for many hours at a time — I only review a big chunk of work at the end.
That's why I think we'll eventually (maybe even already now?) need to be careful about token consumption.
Especially now that we found anti-lazy pills.
Idea: the user can set the maximum consumption per prompt (e.g. 10% of weekly limit).
See also
What
Being able tor retrieve the usage limits.
Alternative: we infer how much % the current session consumed from the amount of tokens/dollars consumed.
Context
With autopilot, the idea is that I make a prompt and let the AI run for many hours at a time — I only review a big chunk of work at the end.
That's why I think we'll eventually (maybe even already now?) need to be careful about token consumption.
Especially now that we found anti-lazy pills.
Idea: the user can set the maximum consumption per prompt (e.g. 10% of weekly limit).
See also