Issue 6, Oct 20, 2026
The 99% Accuracy Trap
Senior AI Engineer interview at Meta, and the interviewer asks:
“Your model gets each token of a tool call right 99% of the time. So why do your 300-token JSON tool calls still fail most of the time? And what would you change in the system, not the model, to fix it?”
Don’t say: “Fine-tune on more tool-call examples”
Why a 99% token accuracy score quietly destroys 95% of your JSON tool calls, and how senior engineers attack the exponent instead of the model.
Share this trapShare on LinkedIn
The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.