In previous articles in their series on bringing AI products to market, Justin Norman, Peter Skomoroch, and Mike Loukides outlined the key responsibilities of AI Product Managers. In their latest post, they discuss how AI Product Managers can solve meaningful problems with AI solutions including through the rigorous application of metrics (including guardrail metrics), frequent experimentation, evaluation of ethical considerations, and careful assessment of tradeoffs. They also discuss the stages of the AI development process, from qualitative user research and data exploration through deployment and monitoring, with tips on how to optimize each part. In addition, they provide a few more recommendations on executing the AI product roadmap, including by involving designers and strong technical users, limiting scope, and leveraging rapid prototyping.