ROI, quality metrics, cost models, and executive reporting. Ten pieces from a weekly archive that now runs backward from June 20, 2026.
Eric KimRecent legislative proposals argue that if the public funds the foundational research for frontier AI, the public deserves an economic stake in the resulting commercial models.
Eric KimAI can make your team write code and draft emails twice as fast, but if the output quality drops, the speed is an illusion. Learn how to track the true velocity of AI workflows.
David VanceExecutive AI dashboards are often cluttered with vanity metrics. Discover the core KPIs that actually indicate whether an AI investment is generating business value.
Tack MediaCalculating Return on Investment for AI is notoriously difficult due to hidden costs and soft benefits. Here is a framework for honest, rigorous AI ROI measurement.
Eric KimStop measuring AI costs by tokens. The only metric that matters for enterprise accounting is the Cost Per Useful Action (CPUA).
Eric KimThe excitement of a launch demo rarely translates to sustained usage. Learn how to track deep adoption metrics that reveal whether an AI tool has actually integrated into daily work.
Julian ThorneModels fail, and handling those failures costs money. Incorporating exception rates into your ROI calculations provides a realistic picture of automated workflows.
Tack MediaSubjective scoring systems for AI quality are useless for engineering. Learn how to implement objective, verifiable metrics to compare model performance.
Single Grain AI TeamDeploying an AI model is cheap; keeping it running is expensive. Uncover the hidden maintenance costs that derail enterprise AI budgets.
David VanceYou cannot prove an AI tool is faster if you don't know how slow the manual process was. Discover the critical baseline metrics teams must capture before launch.