Placement that searches, not guesses
Around fifteen minutes of adaptive questions narrowing in on the exact edge of what your child can do. Not a grade-level quiz — a search for the first skill they're not yet solid on.
Fifteen minutes of placement finds the exact skill your child is ready for — not their grade level, the actual skill. Then it teaches from there, adapting every session, and tells you honestly what it sees.
No card to place. Placement takes about 15 minutes.
Placement result
Maya, age 6
Ready for: blending CVC words
Not a grade level. The first skill she isn't solid on yet, and the one thing standing behind it.
312 skills
A mapped progression from letter sounds to multiplication, each with its prerequisites
Every session is transcribed and readable by you, in full, forever.
LittleAI is a tutor, not a diagnosis. If we notice a persistent pattern, we describe what we saw and suggest talking to a teacher — we never name a condition.
You cannot answer that from a report card, and the apps aren't helping — they hand out stars for effort and tell you nothing.
Grade level is not a skill level
A child can be a confident reader who never learned to blend consonant clusters, or fluent in addition and lost on place value. “First grade” averages over the specific gap that is actually holding them up.
Apps optimise for time spent
Streaks, coins, confetti. Engagement metrics are the product's goal, and they're not the same as your child learning to read. A child can spend forty minutes and learn nothing.
Nobody tells you the truth
Twenty-eight children and one teacher means honest, specific feedback about your child is rare. So you're left guessing between “they're doing fine” and a worry you can't name.
It finds the real starting point, teaches one skill at a time, and shows you what it saw.
Around fifteen minutes of adaptive questions narrowing in on the exact edge of what your child can do. Not a grade-level quiz — a search for the first skill they're not yet solid on.
312 skills with their prerequisites mapped. Blending comes after letter sounds; place value comes before regrouping. Nothing is taught before what it depends on.
A model of what your child actually knows, updated after every answer, that distinguishes a lucky guess from real understanding — and lets mastery decay if a skill goes untouched for weeks.
Voice-first, warm, and never gives the answer away. It asks what your child tried, finds the misconception underneath the wrong answer, and works from there.
Complete transcripts of every session, readable by you, forever. Not a summary — the actual conversation. Nothing your child says to LittleAI is hidden from you.
What they mastered, what they struggled with, what to do about it. If a struggle persists across weeks, we say so and suggest talking to their teacher.
Place, practise, report. About fifteen minutes a day.
Your child talks with the tutor while it works out exactly where they are. You'll see the result as a specific skill, not a grade — “ready for blending CVC words”, not “level 2”.
Ten to fifteen minutes, voice-first, one skill at a time. It picks what to work on from what they've mastered and what's next in the graph — never a random worksheet.
The full transcript after every session, and a weekly summary that tells you the truth. Pause, adjust the daily limit, or export and delete everything at any time.
“We take safety seriously” is not a mechanism. Here is what the system does, each of which is code with a test behind it in our repository.
Output passes a filter for personal-information requests, off-topic drift, unsafe content and anything that breaks character. If the check cannot run, the response is blocked — it fails closed, never open.
If your child asks whether it's real, it says it's a computer program that helps with reading and math — warmly, and then moves on. It is never a friend, and it never says it misses them.
Anything else — family, feelings, the news, buying things — is gently redirected. There is no general-purpose conversation, no matter how the question is phrased.
If your child says anything suggesting harm or danger, the session ends and you are contacted right away with the exact transcript. The tutor does not attempt to counsel them.
Full transcripts of every session, forever. No private mode, no summary-only view. What your child said, and what the tutor said back.
Your child's conversations are never used to train models — not by us, and contractually not by our model provider either.
We designed the safety controls before the tutoring, and the repository has a SAFETY.md mapping each commitment on this page to the code that enforces it and the test that proves it. If we can't point at the test, we don't put the claim here.
Two subjects, mapped end to end. Every skill lists what has to come before it, so nothing is taught out of order.
116 skills · Letter names → multisyllable decoding
196 skills · Counting → multiplication and fractions
Placement can start your child anywhere in either strand — the two are tracked independently, because most children are further along in one than the other.
Placement is free and complete. You'll know where your child stands before you decide anything.
$0
Find out where they actually are.
$19/mo
Daily tutoring in reading and math.
$29/mo
Up to four children.
Custom
Classrooms and districts.
Cancel in one click. Export everything, delete everything — your data leaves with you.
Fifteen minutes, no card, and a real answer at the end of it — a specific skill, not a grade level.
Start the free placementAges 4–8. Reading and math. Cancel or delete everything any time.