Rita
02/07 · the timekeeperdecay detectionRita tracks decay — what aged, what expired, what still claims to be current.
- a clock on every record
- ranks decay by retrieval
- 36-month memory
Rita doesn't ask whether a page is wrong. She asks when it was last true. Content doesn't fail loudly — it drifts: the plan gets renamed, the figure gets superseded, the product gets retired, and the page keeps saying what it said the day it was published. Rita puts a clock on every record.
Her sharpest instrument is retrieval weighting. A stale page nobody visits is housekeeping; a stale page in your main navigation is what your AI agents quote first. Rita ranks decay by how likely a page is to be retrieved, which is why her findings read like risk, not tidiness.
Sample findings
"47 pages in your main navigation haven’t been updated in over a year. These are the pages your AI agents are most likely to retrieve."
"The oldest page in your main navigation was last touched in 2021. It’s still the third result for your own product name."
"Your pricing page references a plan you renamed 14 months ago."
"62% of your blog predates your current positioning."
Method
- 01Age evidenceLast-modified headers, sitemap dates, visible bylines, copyright lines — cross-checked, not trusted singly.
- 02Dated-claim extractionYears, versions, plan and product names — checked against the estate’s own current pages.
- 03Decay rate per sectionHow fast each section ages relative to how often it changes.
- 04Retrieval weightingNavigation position and inbound links decide which stale pages matter first.