Operational efficiency is a design problem, not a tooling problem
Most teams don't lack tools. They lack a shared model of the work.
Every team I have worked with reaches for the same reflex when things slow down: add a tool. A new tracker, a new dashboard, a new integration. Six months later the friction is still there, now spread across one more surface nobody fully owns.
The reason is that tools encode decisions about how work should flow, and most teams never made those decisions explicitly. They inherited a shape from whatever software they adopted first, and then spent years contorting the actual work to fit it.
What I have started calling operational design is the practice of making that model explicit before you reach for software. What is a unit of work here? When is it done? Who holds its state, and where does that state live when nobody is looking at it? These are design questions, and they have to be answered in your own language before any tool can help.
pmstate came out of exactly this frustration. It is not a project manager; it is a way to treat the state of work as a first-class, inspectable thing — something you can snapshot, diff, and reason about — instead of a side effect of whichever app happens to be open. Once the model is real, the tooling almost falls out of it.
None of this gets easier with AI in the loop. It gets more urgent. An agent that can act on your behalf is only as good as the model of the work it is acting on. Get the operational design right and AI compounds it. Get it wrong and you have automated the confusion.