
The discourse around trimming AI costs has focused on prompts. Lesser attention has been paid to the data the tools send back, even though that's where most tokens are spent. While developers focus on creating prompt libraries, counting…
View original source — Hacker Noon ↗



