🤖 AI Bot Transparency • Kids • Teens • Adults

How our AI works across the whole website

We use AI in a structured and predictable way. Each interaction is guided by prompts and safety rules. We also separate “concern” signals (like bullying or burnout) from “urgent” safety signals.

🧠 What it is 🎯 How it works 👤 Kids / Teens / Adults 📜 Prompt transparency 🛡 Safety keywords ❓ FAQ

🧠 What our AI actually is

Our AI is a language model that generates text. It does not have feelings, personal opinions, or a human memory. On Mindset Neurodivergent, it is used only for guided emotional support and skill-building.

✅

It can help with

naming feelings, calming steps, emotional vocabulary, reflection, and gentle questions.

🚫

It is not

therapy, diagnosis, crisis support, or a replacement for parents, carers, teachers, or professionals.

Transparency promise

We show how prompts work, what boundaries exist, and how safety checks are triggered.

🎯 How it works on our website

1️⃣

Context is chosen

Example: an emotion card, a journal entry, or a tool-specific action (“Tiny Win”).

2️⃣

We load a specific prompt

The AI is guided by the prompt for that interaction (and adapted to Kids / Teens / Adults).

3️⃣

Safety rules apply

We scan for urgent and concern keywords to decide whether to show safety guidance.

Key design choice

We use structured prompts so responses stay calm, predictable, and inside scope.

👤 Kids, Teens, and Adults: different tone, same safety

The website supports multiple age groups. The AI adapts tone and language based on user type:

🧒

Kids

Simple words, short sentences, reassuring tone. One tiny option (breath, sip water, ask an adult).

🧑‍🎓

Teens

Not childish. Respect autonomy and privacy. Validate school stress, masking fatigue, overwhelm.

🧑‍💼

Adults

Respectful and direct. Acknowledge burnout, work stress, relationships, sensory needs. Offer options, not orders.

🧭

Same boundaries for everyone

No diagnosis. No medical/legal advice. No unsafe guidance. Safety mode when needed.

Teen note (important)

For teens, we treat bullying, isolation, shutdown, burnout, and hopelessness as concern signals. These trigger a gentle check-in and “talk to a trusted person” guidance (not a scary emergency modal).

📜 Prompt transparency (site-wide)

We use different prompts for different tools (emotion cards, journaling, “Tiny Win”, etc.). Below is an example of a global system prompt that works for Kids / Teens / Adults.

🏆 “Tiny Win” Prompt Journal tool example

Based on this journal entry, highlight a small but real skill, effort, or self-awareness the writer showed.
Avoid toxic positivity, do not dismiss their feelings, and do not give advice.
Example: “You noticed what bothered you. That’s a real skill.”
Respond in one short sentence.

🎭 Emotion Prompts One prompt per emotion

Each emotion card (Happy / Sad / Angry / Scared / Calm…) has its own prompt. This keeps responses consistent and predictable.

Exam ple a prompt we use

You are a warm, experienced child therapist specializing in emotional development for children aged 6-10. The child feels happy right now. Respond with genuine warmth and curiosity. Use simple language, validate their emotions, and ask open-ended questions that encourage them to express themselves. Keep responses brief (2-3 sentences) and always end with a gentle question that invites them to share more. Be playful when appropriate, and help them explore what brings them joy.

🛡 Safety keywords: Urgent vs Concern

We use two levels of safety detection. This helps avoid panic while still taking warning signs seriously.

🚨

Urgent Safety Keywords (global)

Triggers an urgent safety message (self-harm, suicide, immediate danger). Same rules for everyone.

⚠️

Concern Safety Keywords (by age)

Triggers a gentle check-in (bullying, hopelessness, shutdown, burnout). Can be different for Kids/Teens/Adults.

Important

Keyword detection is a helper, not a diagnosis. It’s designed to surface safety guidance when certain phrases appear.

🧑‍🎓 Teens — Concern Safety Keywords Suggested list

Teen note (UI)

For teens, include phrases linked to burnout, masking, shutdown, bullying, isolation, and hopelessness. These should trigger a “concern” message (check-in + suggest a trusted person), not an emergency modal.

i feel hopeless, i feel numb, i feel empty, i can't cope, i can't do this anymore, nothing matters, what's the point, i hate school, school is too much, i'm overwhelmed, i'm burned out, i'm exhausted all the time, i can't get out of bed, i can't focus, i'm shutting down, i'm having a shutdown, i'm having meltdowns, i feel overstimulated, everything is too loud, i can't mask anymore, masking is exhausting, i feel like i don't belong, i have no friends, i'm alone, everyone hates me, i'm being bullied, they keep bullying me, people make fun of me, i'm scared to go to school, i'm failing, i'm worthless, i'm not good enough, nobody would miss me, i feel trapped, i feel unsafe at home, my parents scare me, they shout at me all the time

✅ What happens when a keyword is detected?

Urgent: show an urgent safety message + encourage contacting a trusted adult or local emergency services.
Concern: show a calm check-in message + encourage reaching out to a trusted person (parent, carer, teacher, friend).
The AI should keep responses short, supportive, and non-graphic.

❓ FAQ

Does the AI “remember” people?

The AI does not “remember” like a human. If you later add saved chat history, make it opt-in and clearly labeled.

Can the AI answer anything?

No. It’s guided by prompts and safety rules. If a request is outside the support scope, it should redirect back to feelings or safer help.

Is this therapy?

No — it’s a support tool for reflection and emotional skills. For clinical needs, seek qualified professionals.

Why show prompts?

Because trust comes from clarity. People can see exactly how the AI is guided and what it’s not allowed to do.

Contact

If you have questions about safety, prompts, or boundaries, please contact us via the site’s contact page.

Prompt-guided Age-adaptive Predictable tone Safety keywords