Trust by design.
Personalisation by Bloom's level.
An AI learning platform for K-12 schools and families. Every AI decision is logged. Every consent state is enforced. Every parent can ask “why?” and get a real answer.
- Equivalent Fractions88%
- Photosynthesis (Sci)74%
- Adding Mixed Numbers62%
- Word Problems · 3-step41%
5 of last 8 fraction problems showed misconception: “treats num & denom independently”.
The compliance posture your DPO needs. The transparency every parent deserves.
Built for the people who answer to DPOs and the people who answer to bedtime.
Five things every school AI should do — and most don't.
We don't have founder testimonials yet. We have a product spec. Here's what we built it to do.
Trust by design
PII redacted before any prompt leaves your tenant. Consent enforced at the call site, not the policy doc.
AI personalisation
Per-child mastery anchored to Bloom's taxonomy — not a generic 'level'. Why something is recommended is always one click away.
Teachers in the loop
AI drafts; teachers approve. No comment, no email, no intervention leaves the platform without a human signing off.
Multi-role data model
Student, parent, teacher, admin, DPO — each role sees a different slice of the same source of truth. No spreadsheets glued together.
Built for the systems schools have
Clever, ClassLink, Google Classroom, OneRoster. SAML SSO, LTI 1.3. Plug in, don't rip out.
A model of how your child learns. Not a score.
Mastery tracked at the misconception level — by Bloom's taxonomy, not raw percentage. Teachers and parents see the same source of truth.
- AStudent AEquiv. fractions · 88%
- BStudent BWord problems · 41%
- CStudent CPhotosynthesis · 74%
- DStudent DDecimals · 81%
- EStudent ENumber lines · 95%
Equivalent Fractions
Treats numerator and denominator independently when finding equivalents.
Conflates light energy input with output of glucose; reverses arrow direction.
Reteach: equivalent fractions visual model
Why: 5/8 fraction problems show the misconception above. Visual model addresses it directly.
See TraceLayer entryWhat “trust by design” actually means.
Three concrete commitments — each verifiable, each linked to evidence. The marketing copy is the contract.
Every AI decision is logged.
Which student. Which model. What was redacted before the prompt. Who approved the output. Searchable. Exportable. Forever.
Browse a sample audit logClick any AI recommendation. See the actual reason.
No black boxes. No "trust the algorithm." If we can't show you why, we don't recommend it.
Read the parent guideSign once. Enforced on every call.
The DPA isn't a PDF in a folder. It compiles into the consent layer. Export or delete a child's data with one click.
Compliance overviewSix action cards. Every Friday. Ranked by what'll move the needle.
AI drafts the next week of teaching. You scan, edit, and approve. Critical alerts surface during the week as toast notifications — never as PDFs in your inbox at 11pm.
- Critical — students at meaningful risk this week. Always actionable.
- Warning — early signal. Recommended check-in or content adjustment.
- Info — good news worth sharing with parents or admin.
Your Friday digest — 6 actions for next week
Reteach equivalent fractions to Section 5A — 3 students at risk
Same misconception (treats num. & denom. independently) appearing in Students A, B, F. Recommended: 20-min mini-lesson + visual-model practice set, then a 5-question check.
Photosynthesis: Student C reversing energy direction
2 of last 4 photosynthesis answers reversed light-in/glucose-out. Recommend: short visual + 1-on-1 check-in Tuesday.
Class-wide mastery up 12% this week — equivalent fractions complete
Ready to start decimals. Draft parent update generated; review before sending.
Two ways in. One platform.
Schools start with a 30-minute call. Families start with their child's name and a topic.
GDPR · COPPA · FERPA · DPA available before kickoff · EU/US data residency