What does good AI performance actually look like?

In this episode of, James Corcoran, Head of AI at STAC, joins us to challenge outdated assumptions about performance benchmarking and AI infrastructure strategy. We explore why traditional metrics like latency and throughput no longer tell the whole story and what leaders need to measure instead. From model fidelity and semantic drift to RAG bottlenecks and GPU economics, this episode delivers practical insights for scaling AI systems with confidence. Read the show notes here.