Each of these starts from a finding, not a feature list. The interface exists to make one specific thing visible. Everything runs on synthetic data structured to match the real public source it is built on.
Quarterly filings show 3.59 hours per resident day against a 3.48 minimum. Compliant on paper. But 130 of 351 individual days fell short, and 77 percent of those were weekends, which are only 28 percent of days. Averaging hides the pattern. CMS surveys sample individual days.
Open the dashboard →A straight period-over-period comparison reports 65 percent revenue growth. Adjusting for a fourteen-day tracking outage in the prior period and a one-off promotional spike brings it to 35 percent. Both numbers come from the same dataset. Only one is defensible.
Open the dashboard →These use synthetic data, labeled as such inside each dashboard. The structure, field names, quirks, and failure modes are taken from the real sources: submission gaps, partial reporting periods, and inconsistent identifiers are present because they are present in the originals. Thresholds and definitions are real.
Working with someone else's live data is a different engagement. These exist to show what the output looks like and how the reasoning is handled, not to stand in for it.