Unsafe by default
Open assistants can surface content that is simply not meant for a child, and were never designed with age in mind.
Guided AI learning · Ages 6 to 12
Guided AI learning, built for how children learn
Corectio is a parent-managed learning space for children aged 6 to 12. Children can ask questions, practise skills and explore ideas, while adults set learning preferences, daily limits and access.
- Parent-managed profiles and controls
- Safety checks on child messages and AI replies
- English and Canadian French
Corectio
Homework guidance
How do I add 3/4 and 1/6?
Great question. Let's take it one step at a time. What do both fractions need before we can add them?
Guide only
Why a different tool
General chatbots were never built for children
General-purpose assistants are made for adults and are wide open by default. For a six-year-old, that is the wrong starting point.
Endless distraction
Unbounded, open-ended chat pulls children away from learning instead of keeping them focused on it.
Privacy exposure
Nothing stops a child from typing their name, age, or school — the most sensitive data of all.
An answer machine
When homework is done for the child, the learning disappears with it. Answers replace understanding.
Children need a controlled learning loop — not an adult tool with the safety bolted on afterward.
How it works
It guides thinking — it doesn't hand over answers
Our teaching approach is built into every reply: meet the child at their level, then coach them toward the answer instead of replacing the work.
01
Meets the child at their level
Every reply is written at a reading level tuned to the age band a parent sets — clear, warm, and never over their head.
02
Homework mode: hints, not answers
In homework mode the assistant leans on questions, steps, and hints instead of final answers — so the child does the thinking and keeps the learning.
03
Builds step by step
Conversations keep their context, so a tricky problem can be worked through one back-and-forth at a time, the way real learning happens.
04
Checked coming and going
Each child message and each reply is reviewed for safety before it is ever shown — a guardrail on both sides of the conversation.
An honest note: "no final answers" is a strong design goal backed by an output check — we treat it as a soft guarantee, never an absolute promise.
New in the beta
More than one way to ask
Reading and typing should not be the price of curiosity. A child can hear the answer, or show the problem instead of describing it.
Answers read aloud
Any reply can be spoken on request, so a beginning reader hears it while they follow along. Nothing ever plays by itself, and the audio is made for that press and never stored.
Show it instead of typing it
A child can photograph the page in front of them and ask about it — the fastest route from a stuck homework question to a hint. The photo is checked before it is answered and is never written to our disks; the transcript keeps only a note that a photo was attached.
An honest note: the picture check covers less ground today than the text check does. Photos are screened before they are answered, but this is the newer and less proven of the two paths.
Parental controls
Parents set the boundaries — not the child
You build each child's profile. Those settings are enforced on our servers, so nothing a child types can loosen them.
Profile settings are applied server-side on every request. The child's app can never override a safety setting you've chosen.
Age band
Choose 6–8, 9–10, or 11–12 to match reading level, tone, and what the assistant is comfortable discussing.
6–8
9–10
11–12
Allowed subjects
Turn subjects on or off. The assistant works from an allowlist of what you permit — not a blocklist of what you don't.
Homework strictness
Set how firmly homework mode holds back answers, from gentle hints to strict guide-only coaching.
Low
Medium
High
Daily limits
Cap messages and time per day so screen time stays healthy and predictable.
Safety & privacy
Safety checks run on every single request
Protection isn't a feature you switch on — it runs server-side on every message, in layers, and it is the reason this product exists.
Semantic moderation, not keyword matching
Every message is understood by meaning and sorted into allow, redirect, refuse, or crisis. We reject naive keyword blocklists — they wrongly flag words like "skill" and are trivially bypassed.
A warm crisis protocol
If a child is in distress, they never meet a cold "I can't help with that." They get a calm, caring reply that points them toward a trusted adult or helpline — and the moment is recorded in your safety feed.
Privacy considered from the first line
Built for under-13 users in the US and Canada with parental consent first and privacy settled before the first line of product code. Our privacy notice says exactly what is kept and for how long.
Data minimization, private by default
We keep as little as possible, with a clear retention policy, and every child profile starts fully private.
A firm parent-and-child boundary
The profile is set by you, the parent. The assistant never asks a child for their name, age, location, or school.
The rules live on the server
The safety instructions are injected server-side and are never sent to — or editable by — the child's device.
Beta · Pilot schools welcome
From beta to classroom
Guided AI learning for every classroom
Schools are the direction the whole product is built toward, and the foundation is already standing: a school workspace, classes and rosters, and dashboards where a coordinator sees the whole school while a teacher sees only their own class. The same server-side safety that protects one child at home is what holds the line across a classroom.
What works today: school accounts, classes, rosters, and the teacher and coordinator dashboards — all of it demonstrable in this beta. What is still ahead: school licensing, the student-records compliance a district requires, and sign-in through the systems schools already use.
Pedagogy first, always
Guided learning and age-appropriate coaching that fits how teachers already teach — never an answer service dressed up for the classroom.
Auditable by design
Every safety decision is recorded within a clear retention policy, so a school can review, trust, and stand behind the tool.
Built to scale, one class at a time
A teacher sees only the children on their own roster. That narrowing is enforced on our servers by default — it is not a setting somebody has to remember to switch on.
Questions parents ask
Frequently asked questions
Safety is the whole point. Every message a child sends and every reply they receive is checked server-side, in layers, and sorted by meaning into allow, redirect, refuse, or crisis. The safety rules live on our servers and cannot be edited from the child's device.
Because handing over answers removes the learning. In homework mode the assistant uses hints, steps, and questions so your child does the thinking. You control how strict this is per profile. We treat "no final answers" as a strong, output-checked goal — an honest soft guarantee rather than an absolute promise.
Every profile is private by default, we practice data minimization with a clear retention policy, and the assistant never asks a child for personal details like name, age, location, or school. To be plain about one thing: messages, any photo a child attaches, and replies read aloud are sent to OpenAI to be checked and answered — our privacy notice says exactly what that involves.
You do. Parents set the age band, allowed subjects, homework strictness, and daily limits. Those settings are enforced on our servers on every request, so nothing a child types can loosen them.
Distress is handled with care, never a generic refusal. The assistant responds calmly and warmly, points your child toward a trusted adult or helpline, and — depending on your consent settings — flags the moment for you.
We are launching for families in the United States and Canada, in both English and French — so Canadian families can use Corectio in the language of their choice.
Give your child a guided way to learn with AI
Create a free parent account, set up your child's profile, and stay in control from the very first message.
Parent account required · No anonymous access · You set every boundary