AI agents make it remarkably easy to start and generate work. Here is how I would use WIP, work item age, cycle time, and throughput when code review and customer learning cannot keep up.
Most Kanban boards stop at release, which quietly teaches teams that deployment is success. If you want impact, make outcome and impact visible in the flow.
Flow metrics aren't useful because metrics are good. They're useful when important work is slow, blockers show up late, and forecasts keep missing.
Starting numbers for active work and queues, covering solo and multiplayer spec-driven development, pod topology, flow buffers, and replenishment cadence.
The best time to solve a feature adoption problem is before development. Talk about the user's project, look for real pull, and don't build on polite interest.
An AI coaching prompt from my "Product Orientation Through LPM" talk: pressure-test one real initiative, find the real gap, and get one experiment to run.