SukoonAI

• 

  sukoonai

Inspiration Over 70% of teenagers battling depression, anxiety, and silent emotional struggles never seek help. In our communities and schools, adolescent mental health remains heavily stigmatized. When young people do want to reach out, fear holds them back: fear of being judged by peers, fear of their confessions being reported to parents, and fear of surveillance by school administrators.

We asked ourselves: How can we provide teenagers with a judgment-free, genuinely safe space to heal, while still enabling schools and families to understand and protect their collective well-being?

This question inspired SukoonAI—a compassionate peer-confidant built on the foundational guarantee of absolute mathematical privacy.

What it does SukoonAI is an empathetic mental wellness companion and institutional analytics ecosystem designed specifically for teenagers.

Empathetic AI Confidant ("Dost Ki Tarah"): A friendly, warm peer AI that listens actively to teenagers experiencing emotional distress, panic, loneliness, or burnout without judgment or clinical coldness. Habit+ Positive Transformation Engine: Uses cognitive-behavioral techniques to transform harmful escapist habits into constructive daily quests (journaling, mindful pauses, digital detox challenges, and positive coping routines). Zero-Knowledge Privacy: All student conversations are encrypted on the user's device before leaving it. No one—not teachers, parents, or database administrators—can decrypt or read personal student messages. 4-Tier Institutional Architecture: Student: Personal, encrypted safe space and habit transformation. Teachers & Incharges: Aggregated, 100% anonymized classroom emotional trend heatmaps (enabling educators to identify shared exam stress without singling out or exposing individual students). Parents: General developmental wellness and positive communication mindset guidance. Principals & Leadership: Macro-level campus wellness indices to plan timely institutional counseling and wellness workshops. How we built it We engineered SukoonAI with a modern, privacy-first technical architecture:

Frontend: Built with Next.js / React and responsive CSS to deliver a warm, accessible, and distraction-free mobile-first user experience. Client-Side Cryptography (Zero-Knowledge): Implemented using the Web Crypto API (AES-GCM 256-bit) and asymmetric key derivation. Encryption keys remain solely on the student's local device. Backend & AI Pipeline: Powered by Python / FastAPI orchestrating LLM agents tuned for empathetic dialogue, emotional validation, and de-escalation protocols. Anonymization & Aggregation Layer: Differential privacy algorithms analyze sentiment vectors and emotional themes to generate collective institutional metrics without storing or indexing identifiable personal data. Challenges we ran into Balancing Empathy with Safety Guardrails: Ensuring the AI acts as a comforting peer ("Dost") while strictly detecting self-harm markers and safely providing crisis helpline resources without compromising user trust. Implementing Zero-Knowledge on Web Clients: Managing cryptographic key generation and local persistence without requiring complex technical onboarding from non-technical teenage students. Preventing Re-identification in Small Classrooms: In smaller class sizes, aggregated data can inadvertently reveal who vented. We had to design strict threshold minimums and noise addition to ensure true anonymity in institutional reports. Accomplishments that we're proud of Built a Truly Private HealthTech Product: We didn't just claim privacy in terms of service—we proved it mathematically using client-side Zero-Knowledge encryption. Empathetic Conversational Tone: Perfected a conversational agent that speaks authentically to teenagers without sounding clinical, robotic, or patronizing. Closing the School-Home Gap: Created a system where educational institutions can actively monitor and improve campus climate without turning schools into surveillance states. What we learned Adolescent mental healthcare must prioritize trust and agency over control; teenagers open up only when they know their privacy is untouchable. Implementing cryptographic protocols on client-side web apps requires deep attention to latency and smooth user state handling. Data aggregation for social good requires rigorous ethical boundaries to protect vulnerable demographics from unintended bias or scrutiny. What's next for SukoonAI Voice & Multilingual Audio Journaling: Introducing localized voice interaction in Urdu and regional languages so students who struggle to articulate their feelings in text can speak freely. School Pilot Programs: Partnering with progressive high schools and colleges across Pakistan for real-world beta deployments of the 4-Tier collective dashboard. Offline-First Habit Engine: Developing progressive web app (PWA) offline capabilities so daily quests and grounding breathing exercises remain accessible even without active internet connectivity.

Built With

• css
• database
• fastapi
• html5
• react
• sqlalchemy
• sqlite