Small AI products for decisions that are expensive to get wrong.
And the instruments that prove they work. Each one is kept deliberately narrow, and tested against problems designed to make it fail before anyone relies on it.
CentreCheck
Inspection readiness for licensed Ontario child care centres, checked before the Ministry visit.
no public link yetEvery product here does one job, and the scope is kept narrow on purpose. A tool that tries to do everything cannot be checked properly, and anything handling records this sensitive has to be checkable.
Before anything ships it is run against problems built specifically to make it fail, including ones it is supposed to leave alone, because something that flags everything is no more useful than something that flags nothing. It does not go out until it handles both.
Where something is unfinished, this page says so.
Used to produce your report, and nothing else.
No secondary use, no analytics on the contents, no human review unless you ask for it.
Never used to train a model. Never sold.
Not to train ours, and not passed to anyone else to train theirs.
Deletable at any time.
Ask and it is gone, including from backups on their next cycle. You get written confirmation naming who actioned it and when.
Kathleen Chimirri
Operating as RubyTiger in Waterloo, Ontario. Email reaches the person accountable for the product, not a ticket queue.