From Tokenmaxxing to Trusted Throughput

Official Schedule Context

AI adoption is accelerating, but for many engineering organizations, token consumption is now

significant enough to demand real economic discipline. Drawing on Ironclad’s experience scaling AI

across engineering, Mingsheng Hong will introduce the concept of trusted throughput: the rate at

which teams convert AI usage into reviewed, validated, maintainable, and safely deployed customer

value. He will share a practical framework for measuring AI cost and return, identifying bottlenecks

in code review, CI, and merge workflows, and improving ROI through better guardrails, engineering

practices, build-versus-buy decisions, and token optimization. Attendees will leave with a clearer

way to evaluate AI efficiency—not by minimizing usage or rewarding tokenmaxxing, but by maximizing

trusted customer value per dollar of AI spend and unit of human attention.

No official session recording transcript was found by exact title match on the AI Engineer YouTube channel during this run.

Pending transcript synthesis when an official recording or confirmed matching video is available.