"Not ready," says your IT department. You cite US civilian agencies:…
When \"not ready\" has never published its criteria, no tool can pass it. What I checked in the public registry, what the ×20 figure leaves out, and an eight-question file to attach to your request.
You asked for an AI tool. Your IT department answered "not ready enough". Three months later the request is asleep in the backlog, and nobody has written down what would wake it. You did not get a refusal, you got a postponement with no date: a brake that publishes no criteria, so no tool can pass it. What you leave with: an eight-question file to attach to your next AI tool request, so "not ready" becomes a list the tool either passes or fails. What it changes (PM, PO, designer, scrum master, Design System lead): a frozen request stops being an opinion and becomes a missing set of criteria you can write. Why now: on 27 October 2026 Anthropic announced that use of Claude in US civilian federal agencies grew roughly 20-fold in a year. That is the vendor's own figure, and I looked at what…