Detective ROC is roc.up's built-in AI reviewer. It reads what your workers write, surfaces the risks a busy manager would never catch by scanning a list, and drafts the incident reports that used to eat an afternoon — with a human in the loop on every decision.
Six capabilities, all reading the same data your workers already produce — shift notes, handovers, and clinical logs. No new forms, no double entry.
Three near-falls in nine days across three different workers. A mood sliding across a fortnight of shifts. Sleep or appetite changing across weeks. Detective ROC surfaces the pattern on the dashboard before the next shift, not at the next monthly review.
A shift note that reads like an incident — a fall, a medication refusal, aggression, an abscondment — is flagged for review, with the exact sentence that describes it. One click to escalate to a formal incident report.
From a flagged note, Detective ROC drafts a structured incident report — what, when, where, who, response, actions taken. Your team reviews, edits and signs off. What used to take an hour of typing takes ten minutes of reviewing.
Rushed or incomplete handovers get flagged. Vague or copy-pasted notes are caught. Your audit trail stays substantive — something an NDIS Commission auditor can actually read, not "Client had a good day" fifty times.
Every signal is ranked low, medium, high, or critical, with a clinical-relevance check. A "refused a snack" note doesn't spawn the same alert as a missed medication escalation. Managers see what matters first.
Every output is a suggestion your team reviews. Detective ROC never files an incident, closes a report or changes a record on its own. Your workspace content is never used to train third-party AI models.
What Detective ROC actually catches — with the copy your team sees on the dashboard.
Three different workers noted a near-fall on the way to the bathroom across nine shifts. No incident report was filed — each worker thought their shift's event was a one-off. Detective ROC surfaces the pattern with the three source notes attached, so the coordinator can update the risk assessment and schedule an OT review.
The note describes a medication refusal followed by physical aggression, but was filed as a routine progress note. Detective ROC flags it, drafts the incident report, and the shift lead has a signed submission for the Commission within the required window.
The medication check field was skipped. Detective ROC flags the gap, the incoming worker sees the flag before starting the shift, and a targeted refresher is suggested to the same worker after two more skips.
Not every flag is bad news. Detective ROC picks up a three-week improvement in sleep quality after a routine change and calls it out — evidence the coordinator can attach to the participant's next plan review.
Detective ROC runs on Anthropic's Claude models — Sonnet for per-note scoring, Opus for trend analysis and incident drafting. A keyword-only mode is always available as a fallback and cannot be disabled.
14-day free trial. No credit card. See what your team's notes actually say.