The safest first AI space for curious kids.
Witbot lets kids safely ask, learn, imagine, and practice using AI — with parent-controlled boundaries, age-aware answers, redacted review, and privacy-first defaults.
Adult AI was not designed for childhood.
Your child is going to meet AI. The question is whether their first encounter is a giant open chat box — or a space built for them.
Too open-ended
Adult chat tools can answer too broadly for a child — long, abstract, or off-topic when a kid wanted a quick, age-appropriate explanation.
Too opaque
Parents often can't understand what happened, why a response was given, or what the model chose to skip — even when something feels off.
Too hard to govern
Screen-time limits and content filters can cap how long, but they can't shape how the AI talks to your child or when to redirect.
Witbot turns curiosity into a supervised learning loop.
Four moves. Once you set them, your child can explore — and you stay in the loop without reading every word.
Choose safe modes
Safe Curiosity, Story Builder, Reading Helper, Math Practice, Science Explorer — pick what fits your child.
Pair a child device
QR code or short pair code. No child email, no child password, no child profile to manage.
Every turn is checked
Input policy, model response, output scan — before any text appears on screen or anything is read aloud.
Review what matters
Redacted transcripts, safety events, reason codes, retention controls — without surveilling every word.
Safety is not just a prompt. It is the product system.
Five layers run on every turn. Each one is independent — if a single layer flags something, the answer doesn't reach your child until it's resolved.
Before your child hears an answer
Witbot checks whether the question is safe, whether the response fits your child's age, and whether anything should be redirected to a parent. Layers run in order — and any flag halts the turn before it speaks.
Set boundaries without becoming a prompt engineer.
Switch modes for your child. No prompt editing, no model wrangling, no JSON — just a preset that fits the moment.
Safe Curiosity
Broad, safe exploration. Witbot answers a wide range of questions in age-appropriate ways, and redirects sensitive topics to a parent.
Story Builder
Collaborative storytelling. Witbot writes one paragraph at a time, then asks what happens next — no scary or unsafe directions.
Reading Helper
Vocabulary, comprehension, and read-along. Tuned for emerging readers and stronger readers with adjustable difficulty.
Math Practice
Guided practice — never an answer dump. Witbot walks through one step at a time and checks understanding before moving on.
Science Explorer
Safe explanations and parent-approved experiments. Witbot will suggest activities only from a vetted family-safe list.
Off
Witbot is fully paused on this child's device. No mic, no responses, no input — until you turn it back on.
Teach Witbot how your family handles important topics.
Guided topic controls — not raw prompt editing. See tested examples before anything goes live.
Simple for kids. Governed by parents.
For a 6-year-old, Witbot looks like one mic and one mode. The complexity lives where it belongs — in your dashboard.
Transparency without turning curiosity into surveillance.
You see safety events and topics. You don't have to read every sentence your child says to feel safe.
Not ChatGPT for kids. A family trust layer for child–AI interaction.
Witbot isn't a kid-themed wrapper on an adult chatbot. The parent governance and safety layers are the product.
| Capability | Adult AI tools | Basic parental controls | Witbot |
|---|---|---|---|
| Age-aware voice and reading support | — | Limited | |
| Parent policy controls | — | Limited | |
| Output checked before speech | — | — | |
| Redacted history & reason codes | — | — | |
| No child email or password | — | Limited | |
| Curated learning modes | — | — | |
| Retention, export, and delete controls | Limited | — |
Built for privacy-conscious families testing safe AI habits.
We're inviting families into a small, careful private beta. Real parent feedback shapes the next presets, safety layers, and family-lens topics.
What parents ask first.
Straight answers about safety, privacy, age fit, and what Witbot is and is not.
Witbot is designed for children ages 6 to 12. Safety layers, vocabulary, reading levels, and modes are tuned to this window — not a generic chatbot stretched to fit any age.
No. Children never create an email, password, or profile. You set up the family on your account and pair a child device with a QR or short pair code. You can revoke access from any device at any time.
You see a redacted history by default — questions, topics, safety events, and reason codes — without scrolling through every raw sentence. You can open a full transcript for any specific event. Witbot is designed to give you signal, not surveillance.
No raw audio is stored by default. Only the recognized transcript is saved, and you can turn transcript storage off as well. Witbot also does not train models on your child's content.
Math Practice mode is built for guided practice — Witbot walks through one step at a time and checks understanding rather than dumping a final answer. Other modes will redirect homework-style asks back to guided practice or ask your child to talk to you.
The input is flagged before a response is generated, the response (if any) is checked again before it reaches speech or text, and the event is logged with a reason code on your dashboard. The most common outcome is a kind redirect to a grown-up — never a scolding, and never a silent failure.
No. Witbot is explicitly not a friend, therapist, or always-on emotional support bot. It is a sandbox for safe curiosity, creative play, reading, and structured practice — not a relationship product.
Yes. One-click delete actually deletes — transcripts, safety event metadata, and any stored audio. Export is also one click. The default retention window is 30 days; you can shorten or extend it.
Five independent safety layers run on every turn: a parent policy preset, private-info detection and topic routing, age-aware response generation, an output check before text or speech, and a redacted history with reason codes. A normal chatbot is one of these layers at best.
Witbot is currently in a private beta for a small group of families. Join the waitlist and we'll invite you when there is room and a child profile that matches yours.
Give your child a safer first step into AI.
Join the early access list for families who want safe curiosity, parent control, and privacy-first AI from day one.
No child accounts · No ads · You can leave any time.