Audra Carpenter
AI with AUDRA Podcast
Same Model, Same Task, 36-Point Performance Gap
0:00
-1:59

Same Model, Same Task, 36-Point Performance Gap

Most teams are spending money to solve a problem that better prompting would fix for free

Everyone's debating Claude vs GPT, which tier, which API.

The research says that's the wrong conversation.

A 36-point performance gap came from prompting strategy and workflow design, not the model.

Optimize the harness before you upgrade anything.

Discussion about this episode

User's avatar

Ready for more?