AIdeas: Whispurr — A Safe AI Mental Health Space for Teenagers
Shh... 🤫 Looking for a secret hideout away from the world? A warm cup of tea and a listening cat await you.
App category
Social Impact
Team Members
Team LATTE (Listen, Analyze, Talk, Triage, Escalate)
- Heejin Yu
- Taerin Ki (@xaerinoo)
- Jeonghwan Park (@ohmyparkjh)
Our vision
"What if teenagers had a place where they could whisper their feelings safely?"
Many teenagers carry heavy emotions but have nowhere safe to talk. They hesitate to speak to parents, and fear being judged by friends. Also, visiting a therapist often requires parental consent or social exposure. So their feelings stay hidden. That is why we created Whispurr.
The name combines:
- Whisper — sharing something quietly and privately
- Purr — the comforting sound a cat makes when it feels safe
Whispurr is designed as a secret attic café run by a cat barista.Instead of entering a serious therapy room, teenagers enter a warm hidden space where they can talk freely.
The goal is simple:
Turn silent emotional struggles into safe conversations.
Whispurr is an AI-powered mental health platform designed specifically for teenagers aged 13–19. The platform provides a private emotional refuge where teenagers can talk about their feelings without fear of stigma, judgment, or exposure.
Together, Whispurr represents a safe place where teenagers can quietly share their feelings and receive comfort.
Whispurr is designed as a secret attic café run by a cat barista.
Instead of entering a clinical “therapy room,” teenagers enter a cozy hidden space where they can talk freely.
Many teenagers struggle with:
- academic pressure
- social anxiety
- loneliness
- family conflict
Yet they often cannot seek help due to stigma, parental consent barriers, or fear of being judged.
Whispurr aims to become the first safe step toward emotional support.
The platform focuses on three core ideas:
- Privacy-first experience
- Teenagers can safely use the app with features such as disguised notifications and secure access.
- Long-term emotional insight
- The app analyzes emotional patterns and generates reflective summaries called “Cat’s Receipt.
- Hybrid support system
- If serious distress signals appear, the system can suggest connecting with professional counselors, supported by AI-generated summaries of the user's emotional history.
Why this matters
Teen mental health is becoming a global crisis.
According to the World Health Organization (WHO):
- 1 in 7 adolescents experience mental health disorders
- After COVID-19, 1 in 4 teenagers show symptoms of depression
- Youth anxiety disorders increased 52% between 1990–2021
However, teenagers often struggle to access help.
In South Korea especially, barriers include:
- parental consent requirements for psychiatric visits
- intense academic pressure
- persistent mental health stigma
Many teenagers therefore face emotional challenges in silence.
Existing AI mental health tools provide useful features but still have several limitations:
- designed primarily for general users or adults
- limited privacy protections for teenagers
- weak integration with real-world counseling
- limited cultural adaptation
Whispurr aims to address these issues by creating a teenager-first mental health experience.
How I built this
Whispurr was built using a privacy-first cloud architecture on AWS, designed to support conversational AI, emotional analysis, and secure data handling while remaining lightweight enough for an MVP.
The system integrates Amazon Bedrock for AI reasoning, a secure VPC-based backend, and encrypted storage for sensitive mental health data.
The overall architecture is shown below.
System Architecture Overview
The Whispurr platform consists of four major components:
- Mobile Client (User Interface)
- Application Backend
- AI Conversation Engine
- Secure Data Storage
Users interact with the mobile app, which communicates with the backend hosted inside an AWS VPC.
The backend then processes requests, communicates with the AI model, and stores relevant conversation data securely.
Client Layer — Mobile Application
Teenagers interact with Whispurr through a mobile application interface designed to feel like a quiet conversation in a cozy attic café.
The app supports:
- text-based conversations with the AI companion
- optional voice interaction
- emotional journaling
- viewing emotional insight reports
Backend Layer — Web Application
The backend service runs inside an AWS VPC, providing a secure environment for handling user requests and AI responses.
The backend performs several responsibilities:
- managing user sessions
- processing conversation requests
- generating emotional insight reports
- coordinating AI analysis workflows
AI Conversation Engine — Amazon Bedrock
The conversational intelligence of Whispurr is powered by Amazon Bedrock.
Bedrock allows us to generate:
- empathetic conversational responses
- emotional reflection prompts
- structured psychological summaries
The AI does not attempt to diagnose users.
Instead, it focuses on supportive dialogue and emotional pattern recognition.
Example AI behavior includes:
- validating user emotions
- asking reflective questions
- identifying recurring emotional themes
This helps transform casual conversations into structured emotional insights.
Secure Data Layer — Amazon RDS
Sensitive data such as conversation history and emotional insights must be handled carefully.
Whispurr stores structured data in Amazon RDS located in a private subnet, ensuring that the database is not directly exposed to the public internet.
This architecture ensures:
- secure storage of emotional conversation data
- controlled access through backend services only
- separation between application logic and data storage
Additional encryption layers can be applied to protect user data and maintain privacy.
Development Environment
The development workflow integrates Kiro, which assists with code development and AI-assisted programming.
Kiro helps accelerate:
- backend development
- AI prompt design
- infrastructure experimentation
This allows the team to rapidly prototype new features while maintaining a structured development process.
This architecture was designed with three key goals:
Demo
What I learned
AI for teenagers requires ethical guardrails, not just technical capability
While researching youth mental health and AI usage, we discovered that designing AI systems for teenagers requires strict ethical considerations.
The American Psychological Association (APA) published a 2025 Health Advisory on AI and Youth Well-Being, warning that teenagers may struggle to distinguish between simulated empathy from AI and real human relationships.
Because of this, youth-focused AI systems must include safeguards such as:
- age-appropriate default settings
- transparent disclosure that the system is AI
- strong privacy protections
- bias and misinformation prevention
- features that encourage real human relationships
This insight shifted our design approach.
Instead of building a powerful AI companion, we focused on building a safe AI companion.
Emotional AI systems must prioritize trust and privacy
Mental health conversations involve extremely sensitive information.
Through this project, we learned that privacy architecture is not optional — it is a core product feature.
This influenced several architectural decisions:
- storing sensitive data in private subnets
- minimizing long-term storage of raw conversations
- sharing only summarized insights with counselors instead of full transcripts
Designing the system this way helped us balance useful AI analysis with strong privacy protection.
AI should guide users toward human support, not replace it
At first, we thought the main value of the system would come from the AI conversation itself.
However, while studying existing mental health platforms, we realized that the most effective tools are those that connect digital support with real-world care.
This insight led us to design a hybrid support model, where AI acts as:
- an emotional first responder
- a reflection tool for users
- a bridge to professional counseling when needed
The goal is not to replace therapists, but to lower the barrier to seeking help.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article