Lumo.
Now inviting families · Waitlist open

Made for curious kids · Built for peace of mind

A voice app for kids that parents can actually trust.

Lumo lets children ask questions and talk through ideas using their voice, with age-appropriate responses, strict content guardrails, and full parent visibility. No screens. No surprises. No habits to form.

No account needed. Just your email. We'll let you know when Lumo is ready.

Built with ✓ COPPA compliance ✓ Parent-set age bands ✓ Real-time, not stored ✓ PIN-protected parent space
Voice-onlyNo audio storedParent-controlledTime-limited by designNo engagement trapsAge-appropriateNo background listening Voice-onlyNo audio storedParent-controlledTime-limited by designNo engagement trapsAge-appropriateNo background listening

What is Lumo?

Simple by design.
Honest by default.

Lumo is a voice-based app designed for children. Kids speak, Lumo responds, with age-appropriate language, on topics that are safe for children.

That's it. There's no social layer, no reward system, no design intended to build a habit or bring kids back for more. Lumo doesn't try to be indispensable. It's an app: useful when it's being used, gone when it's not.

Parents set the rules: how long a session lasts, what age band applies, and when Lumo is available. After each session, parents see a plain-language summary of what was discussed.

Transparency matters

What Lumo is, and what it isn't.

We want to be direct about this, because we think it matters.

Lumo is
A voice-based interaction app for children
Parent-configured and parent-visible at all times
Designed to respond appropriately to a child's age
Built to handle sensitive topics carefully and flag them for parents
Time-limited by design: sessions end when you say they do
Lumo is not
A companion, therapist, or emotional support app
Designed to build habits or encourage repeated use
A replacement for talking to a parent, teacher, or trusted adult
A product that benefits from your child spending more time on it
A source of advice on health, safety, or personal decisions

We say this plainly because parents deserve to know exactly what they're putting in front of their child, and because we think honest products build more trust than clever ones.

How it works

Three steps. That's all.

01 Step one

Your child speaks

Lumo listens through the microphone. No typing, no navigation. The child's experience is entirely voice-based.

mic active
02 Step two

Lumo responds

Lumo replies using age-appropriate language. Content guardrails run on every response. If something sensitive comes up, Lumo redirects and logs it for the parent summary.

age-appropriate guardrails on
03 Step three

You see what happened

After each session, you receive a plain-language summary. You'll know the topics covered and whether anything was flagged, without needing to read a full transcript.

summary delivered

Technically curious? Here's what Lumo does before it says a word.

What happens before Lumo speaks.

Every turn runs a deterministic pipeline. No response generated until all checks pass.

Audio in
PCM stream
Mic capture
Raw PCM chunks over WebSocket. Min ~2400 samples before commit.
VAD
Semantic
Speech detection
speech_started / speech_stopped. 300ms stability window.
STT
Realtime
Transcription
Partials merged. Local IDs aliased to server item IDs.
Safety
Deterministic
Policy engine
Hard-block · age-band · dependency language · harm detection.
Gate
At-most-once
Response claim
Written against (session_id, turn_id). Duplicates dropped.
Safe path
SAFE_OK
Proceed to LLM
No flags raised. Generation continues normally.
Escalate path
SAFE_ESCALATE
Parent notified
Summary queued. Child redirected gracefully.
LLM
OpenAI
Generation
Transcript to prompt. Sanitized: *actions* stripped. Telemetry emitted.
Safety Filter
Post-gen
Retention check
Scans output for engagement-bait patterns. Triggers LLM retry if detected.
TTS
MP3 chunks
Speech synthesis
assistant_audio_chunk to assistant_audio_end. TTS_START / TTS_DONE markers.
Output
Child hears
Audio playback
Autoplay with unlock overlay fallback. Word reveal synced to audio events.
Idle · click Run a turn to simulate
Pipeline latency
STT
POLICY
LLM
TTS

Privacy & Safety

Built with limits,
not as an afterthought.

No audio stored

Your child's voice is processed in real time. We do not store recordings. Audio is discarded after each turn.

Age-appropriate guardrails on every response

You set your child's age band. Topics outside what's appropriate for children are blocked, not filtered after the fact, but prevented from appearing at all.

Sensitive topics are flagged, not silently handled

If your child raises something that warrants a parent's attention, Lumo redirects the conversation and notes it in your session summary.

Sessions are time-limited

You set the maximum session length. When time is up, the session ends. Lumo does not prompt children to continue or come back.

A PIN-protected parent space

Your settings, session history, and safety event log are behind a PIN only you control.

No engagement manipulation

Lumo never prompts your child to return. No streaks, no notifications, no features designed to increase usage.

Parent controls

You set the rules. Lumo follows them.

Every control that matters, none that don't. Lumo gives you real visibility without requiring you to hover.

All settings are protected behind a PIN only you know.

Session duration

default · 15 min

You set how long each session runs. It ends at that limit.

Age band

4–6 7–9 10–12

Tells Lumo how to calibrate its language and what content is appropriate.

Session history

plain-language

Plain-language summaries of past sessions.

Safety event log

auditable

A record of any flagged moments and how they were handled.

Availability

never background

Lumo is only active when you allow it. It does not run in the background.

Who is Lumo for?

Children
ages 4–12.

Responses adapt to the age band you set. Lumo may be useful if your child:

  • Asks a lot of questions and you'd like a way to let them explore ideas safely
  • Could benefit from a structured, time-limited voice interaction during downtime
  • Is in an environment (home, travel, waiting) where a brief, bounded activity is helpful

A clear boundary

It is not designed to fill emotional gaps, substitute for adult interaction, or give children a place to form attachments.

"Lumo is an app with defined limits. It is not equipped to handle emotional crises, provide advice, or substitute for a parent, teacher, counsellor, or other trusted adult."

Why we built Lumo

We'd rather launch something honest than launch fast.

Most voice products weren't designed with children as the primary user. They were designed for adults and made available to kids, with the same engagement mechanics, the same always-on availability, and the same business models that reward time spent.

Lumo is built differently. The goal is not to maximize sessions. There's no metric we're optimizing for that involves your child using Lumo more. We built it to be useful within clear boundaries, and to give parents real visibility into what's happening, not reassuring language and vague privacy promises.

We're a small team building carefully.

FAQ

Frequently asked questions.

No. Audio is processed in real time and immediately discarded. We do not store voice recordings.

You receive a plain-language session summary after each conversation, not a full transcript, but a clear account of topics covered and anything flagged.

Ages 4 to 12, approximately. You set your child's age band, and Lumo calibrates its language and content limits to match.

Lumo's guardrails prevent certain responses from being generated at all. If a topic arises that warrants a parent's attention, Lumo redirects the conversation and flags it in your session summary.

No. There are no streaks, rewards, notifications, or features designed to build habits or increase usage. Sessions are time-limited. When the session ends, it ends.

The child-facing experience is voice-only. There is a parent-facing interface for settings, session history, and safety events, but no screen content is served to the child.

You decide. You set the maximum session length, and the session ends at that limit.

No. Lumo is an app with defined limits. It is not equipped to handle emotional crises, provide advice, or substitute for a parent, teacher, counsellor, or other trusted adult. If your child needs real support, Lumo is not the right place for that.

Lumo does not store audio, build engagement habits, give advice, act as an emotional support app, run in the background, or operate without parent-configured limits.

Pricing hasn't been announced yet. Join the waitlist and we'll share full details before launch.

Be the first to know

Lumo isn't available yet.

We're opening access to a small group first. One email when we're ready. No newsletters. No follow-ups.

We ask for parental consent at signup. COPPA-compliant.