Devs Are Making Claude Talk Like a Caveman to Cut Costs—And It Works
Source: Decrypt

A Reddit post claiming 75% output token savings sparked 400 comments and multiple GitHub repos dedicated to making AI grunt its way to efficiency.
Source: Decrypt
