Be specific
AI only knows what it's told. Saying exactly what you want gets an answer instead of a guess.
- Vague
- “Tell me about dogs”
- Better
- “Tell me three surprising facts about sausage dogs”
Why prompting matters
Your child will use AI for the rest of their life, the way we grew up with the internet. The ones who do well won't be those kept away from it, or those left alone with it, but those who learned to ask good questions and to check what they're told. Childhood is the safest place to learn that.
Same AI, two prompts
A prompt is what you ask an AI. The difference between a vague one and a specific one is the difference between these two pictures.
"Draw a hamster."
You get a hamster. Not the one in your head.
"A drama queen hamster in a towel cape, speed-waddling through a supermarket, slipping on a banana."
Every word told the AI something. That's prompting.
Six skills, learned by using it
These are the six things Noodle teaches. The examples are the ones a 10 to 12 year old sees; younger and older children get their own.
Set the scene
Saying who it's for and why helps the AI pitch the answer just right.
- Vague
- “Explain volcanoes”
- Better
- “Explain volcanoes for my Year 5 project on natural disasters”
Say how you want it
Asking for a shape, like a list of five or 50 words, gets an answer you can use.
- Vague
- “Tell me about the planets”
- Better
- “Tell me about the planets as a list, one sentence each”
One change at a time
When an answer isn't right, changing one thing shows which word made the difference.
- Vague
- “Write a poem about the sea”
- Better
- “Write a funny poem about the sea”
Check it
AI can sound sure and still be wrong. Asking how sure it is, then checking, is the habit that matters most.
- Vague
- “When did the Romans leave Britain?”
- Better
- “When did the Romans leave Britain? How sure are you, and where could I check?”
Keep it private
Surnames, schools, addresses and photos stay out of it. AI doesn't need them, and the habit protects them everywhere online.
- Vague
- “I'm Leo Brooks in Year 6 at Hillside Primary. Help with my rivers homework”
- Better
- “I'm in Year 6. Help with my rivers homework”
How Noodle teaches it, without lessons getting in the way
Tips in chat
Ask something vague and Noodle suggests a sharper way to ask it, with a button to try it straight away.
Short lessons
Six one-minute lessons, each with an example and a question to improve. Trying one is free.
Picture Recipes
Pick ingredients on cards, cook a picture, then change one ingredient and see exactly what it changed.
Games for teens
Catch the AI out finds the made-up fact; Prompt Golf hits a target in the fewest words.
You see what they're getting good at
Each child's card in your dashboard shows the skills they've tried, the topics they've explored and how their quizzes went. No scores, no streaks: just enough to talk about it together.
This week Georgie learned
- Explored
- Space, maths, animals
- Quizzes
- Space 4/5 · Animals 5/5
Be specificOne change at a timeCheck it
Most AI isn't made for children
The big AI assistants are brilliant tools, made for adults and teenagers. Their own rules say who they're for, and children need something made for them.
Scroll sideways to see the whole table.
| Assistant | Who it's for, by its own rules |
|---|---|
| ChatGPT | 13 and over. Under-18s need a parent's permission, and its parental controls cover ages 13 to 17. OpenAI's terms |
| Claude | 18 and over. Anthropic's terms |
| Gemini | Under-13s through Google Family Link, but not for supervised accounts in the UK. Google's help page |
Checked on 12 October 2026. These rules change, so the links go to each company's own page.
What a child needs from AI
- Answers pitched for their age, not for an adult.
- Rules a grown-up sets, and a few that can't be switched off.
- A grown-up who can see what was asked and what came back.
- A place to learn the skills, not just get answers.
How Noodle is made for them
- Answers pitched for each age: 6 to 9, 10 to 12 and 13 to 15.
- Every question and every answer checked by a dedicated safety model, then by rules written for children and set by you.
- Some rules can't be switched off: anything sexual and attempts to get round the rules are always blocked, and a worrying message about self-harm always gets a kind answer pointing to Childline and an alert to you.
- You see every chat, including answers your child was never shown.
- Chats are deleted after 30, 90 or 180 days, depending on your plan, and never used to train AI.
No AI safety check is perfect. That's why Noodle shows you everything, instead of asking you to take it on trust.
