Every week it reads your own records and points at what is slipping: a site where hazard reports have stopped, a procedure nobody has looked at, a wave of certificates about to expire. With the numbers, and how sure it is.
Incidents at one site are running above its own normal.
Plant 2 · last six months
Next: look at the six months together with the site's open hazards before the next leadership review.
1Enough history to learn fromNot yet2Proven on months it had not seenWaiting3Switched on by youWaitingA forecast appears only after all three. Until then, nothing is called a forecast.
The same records always give the same findings. What changes week to week is your data, and what you decided last week.
Hazards, incidents, actions, checklists, training, documents, surveys and reviews. Nothing extra to enter.
What it saw, how it compares, and how much data that rests on. Thin data is said to be thin.
One click makes an action or a plan item. Dismiss it with a reason and it will not nag you about the same thing next week.
A short weekly note in the app, a plain-words summary, and the top findings carried into your quarterly review.
Six areas. A few of the questions in each, below. There are many more, and all of them are answered from records you already keep.
Most "predictive" safety tools ask you to take their word for it. This one shows its working.
It learns from your own history, is tested on months it has not seen, and is switched on by you. Until then it is not called a forecast.
A finding that rests on a handful of records says so, and ranks lower than one that rests on hundreds.
You see reasons, like an expired certificate or an overdue action, behind a separate permission. Never a number on a person.
Act, acknowledge or dismiss with a reason. The decision is kept and the finding is not raised again as if it were new.
A walkthrough runs the insights on a sample business and shows how a finding becomes an action. Your own records start building the day you begin.