Could AI Track Government Policy Compliance Instead of Writing New Policies?

A thought experiment published on DEV Community proposes using AI not to generate government policies, but to monitor whether existing ones are actually being implemented. The author argues that governments already possess vast amounts of laws, schemes, budgets, audit reports, and citizen grievances — the real challenge is making sense of it all. The proposed framework would use AI agents, retrieval-augmented generation, and knowledge graphs to map departmental obligations against real-world evidence and audit findings. The goal would be to produce evidence-backed compliance intelligence rather than subjective policy opinions. The author describes it as an early-stage idea and is seeking feedback on architecture and any similar existing projects.