Safeguarding-grade AI for schools and parents. We watch the surfaces enterprise filters can't reach — and surface signal, not noise — to the adults responsible for a child.
Enterprise web-filtering vendors built the safeguarding stack of the 2010s — block-lists, on-network scans, weekly summaries that nobody reads. They were aimed at the school's pipes.
But the modern child's risk surface isn't the school's pipes. It's WhatsApp at the bus stop. Discord at 11pm. AI companions in a private tab. The places enterprise filters were never going to see.
Child Protect Online is the safeguarding layer for those surfaces — with parent consent, with the school's DSL in the loop, and with a model that's tuned for concern, not for blocking.
What the DSL sees on a Monday morning is a short list of real concerns with the context they need to act — not a 400-event dump from a keyword scanner.
School-issued devices, MIS, behaviour log, attendance, pastoral notes. We plug in via documented APIs — no agent install.
With parental consent, we extend the surface to home devices the school filter can't see. Parents stay in control of their child's view.
Ranked by severity with full context, sources, and the suggested KCSIE-compliant next step. Five rows, not five hundred.
CPOMS, MyConcern, or our own log — the audit trail is automatic. Friday-afternoon write-ups disappear.
Parents see concerns, not chat logs. We don't show you what your child said — we show you what we'd want a thoughtful adult to know.
When something serious lands, you and the DSL see the same concern in the same words. The 9pm panic email becomes a Tuesday-morning conversation.
The model takes account of age, neurodivergence, family situation. We don't flag your 15-year-old for the same things we flag at 9.
Parents pull consent and the child's data is gone — from us, from the school, from the model. Verified by audit.
Safeguarding earns trust by being boring about privacy. The promises below are written into the contract every school signs and every parent agrees to. If we break them, you can leave with the data deleted.
Ever. Not in aggregate. Not in private fine-tunes. Our suppliers are contracted to the same line.
The DSL sees a concern, a severity, and the minimum context needed to act. Not your child's diary.
No third-party trackers. No retargeting pixels. Schools pay us. Children are not the product.
Signals expire on a schedule the school sets. Closed concerns auto-delete unless flagged for KCSIE retention.
For the first time in eight years as a DSL, my Monday inbox isn't a horror show. It's five rows, and I can actually do something about each one.
It found something my filter wouldn't have — a Discord pattern at 11pm — and the way it surfaced it gave us time to act calmly, not in crisis.
Every concern flagged includes the relevant KCSIE paragraph and a suggested next step in the language of the statutory guidance.
DPIA template provided. Parental consent flow per the ICO's children-and-the-GDPR guidance. Right-to-erasure honoured in 30 days.
Designed as a monitoring tool — not a replacement for the school's filtering provider. Sits alongside Smoothwall, Securly, etc.
We're running a guided pilot through the 2026–27 academic year. Six schools, free for the year, in exchange for a real safeguarding partnership and a quarterly review with the founders.