The problem nobody in the sector argues about
Ask any centre director what eats their week and you get the same answer: documentation. Not the children — the paperwork about the children. Registration forms typed in by hand, one child at a time. Observations scribbled on paper during a shift and transcribed later, if they survive the day. Weekly program plans written from scratch, every week, usually at home.
And underneath all of it, the quiet dread of a quality assessment arriving while your documentation is three weeks behind.
What we built
- Registration parsing. A registration PDF goes in and a complete child profile comes out — name, date of birth, home language, schedule, allergies, dietary requirements, and everything a parent wrote about their child's interests, needs and character.
- Observation capture. An educator logs what they saw in seconds, on the floor, on a phone. No transcription step.
- Developmental mapping. Each observation maps automatically to developmental domains rather than sitting in a notebook as prose.
- Research-grounded recommendation. The system suggests what activities come next based on early-learning research and the ELECT framework — not on whatever the educator happened to remember that week.
- Plans and parent summaries generate from the observations already captured.
- AQI readiness tracked continuously, so compliance is a state you're in rather than a scramble you do.
Why this isn't a generic AI wrapper
The hard part was never the text generation. It's that a recommendation for a three-year-old with limited verbal communication and a sensory sensitivity has to be defensible to a supervisor and a regulator. The domain logic — developmental domains, ELECT alignment, what actually follows from a given observation — is the product. The AI is the interface to it.
It's saving us 10–15 hours a week and makes onboarding stupidly easy. We used to spend so much time manually adding each child's interests, their needs, their character — now it just gives it to us instantly, and no notes from parents ever get lost.
The observation notes are incredibly helpful for having a holistic view of each child, making sure they're being seen and taken care of. And the AQI assessment means we're always compliant — one of the scariest parts of this job, the part that makes us really anxious, is taken care of.
The pattern underneath
Root Skills and Welend — a mortgage brokerage system we built in a completely unrelated industry — independently arrived at the same core mechanic: a human uploads a document, and the system parses it straight into a structured record. A registration form becomes a child profile. A mortgage application becomes a deal with underwriting attached.
That's the thing worth noticing if you're reading this because your own business has a document problem. The specifics differ enormously. The shape of the fix doesn't.
See it running
Root Skills is a live product with its own site and its own customers: rootskills.ca. It's not a demo built for this page.
Claim discipline: the 10–15 hours a week is reported by Humbercrest Nursery School and published with permission. It's one centre's experience, not an average across all seven. The seven paying licensed centres figure is exact.