SaSe /sæs/ - A Web-Based AI Tutor for Safe Sex Education

A privacy-first AI tutor that makes sensitive education safe and accessible through a discreet, real-world usable chat experience.

Comment

🌟 SaSe /sæs/ A Web-Based AI Learning Platform for Safe and Private Sex Education

Inspiration

As a Vietnamese girl, growing up, I (Nhu Y) always felt a natural sense of avoidance when it came to sexual education topics. It was treated as something inappropriate or even shameful to talk about, even though I did not fully understand why.

Later, through proper education, I realized that sexual health is not something to be ashamed of — it is natural, scientific, and essential for personal well-being. This made me question:

Why do so many people, especially teenagers and women, feel embarrassed about something so fundamental?

Looking deeper, I found that this discomfort is shaped by cultural and historical influences. In Vietnam, this can even be reflected in literature such as “Làm đĩ” (1936) by Vũ Trọng Phụng, which highlights social attitudes toward sexuality and gender norms.

Speaking Number

The consequences of inadequate sex education are serious and measurable.

🇻🇳 Vietnam

  • 47 cases of child sexual abuse (2023), 36 cases (2024), and over 54 cases (2025) reported in Thanh Hóa
  • 250,000–300,000 abortions per year (Ministry of Health)
  • Rate of first sexual intercourse before age 14 doubled between 2013–2019

Source: link

🌍 Global Context

  • Sex education is often incomplete or avoided
  • Teenagers rely heavily on the internet for information
  • Cultural stigma prevents open discussion

Source: link

Our Insight

The problem is not just lack of information —
but lack of safe and private access to that information.

What SaSe does

SaSe Chatbot is a privacy-first AI learning platform that enables teenagers to safely access sensitive education through an intelligent and discreet conversational experience.

The system provides:

  • Structured, age-appropriate conversations with SaSe — a neutral AI tutor
  • A completely anonymous and judgment-free environment (no sign-up required)
  • A unique disguise mode, allowing users to switch into a normal chatbot via:
    • Double-click logo
    • Ctrl + Shift + H
    • Auto-switch after 30 seconds inactivity

What SaSe helps

  • Enables learning in low-privacy environments
  • Encourages open questioning without fear
  • Improves access to reliable sexual education

Our Target Users

Teenagers aged 13–19, especially those facing barriers to accessing sensitive information.

Early Teens (13–15)

  • Curious but shy
  • Limited formal education
  • Easily misinformed

Late Teens (16–19) ⭐ Main Target

  • More exposure to relationships
  • Actively search online
  • Still embarrassed asking adults

Core Focus

Ages 15–18 — highest need + highest digital usage

How we built SaSe

Tech Stack

  • Frontend: TypeScript (React / Next.js)
  • Backend: Node.js + Express
  • AI: LLM API + Prompt Engineering

Safety System

$$ \text{Response} = \text{Explanation} + \text{Why It Matters} + \text{Advice} $$

Core Backend workflow - Guardrail

User Input
->
Normalize
-> Keyword Scan
-> Rule-Based Safety
->
Risk Scoring
-> AI Guardrail
-> Allow / Block / Redirect
->
LLM
 -> Output Guardrail

Core backend flow

Dual Mode System

  • Education Mode → Sexual education
  • Disguise Mode → Math / study chatbot

What we learned

  • AI is about control, not just intelligence
  • UX can solve social problems
  • Privacy is contextual, not just technical

Challenges

1. AI Safety

Balancing helpful vs safe responses

2. Disguise Mode UX

Ensuring realistic switching

3. Time Constraint

Delivering impact in 36 hours

🏁 What we're proud of

We didn’t just build a chatbot —
we removed the barriers that prevent people from learning.

What’s next

  • Localization for Vietnam & Asia
  • Partnerships with schools & NGOs
  • Stronger safety systems

Impact

Short-term

  • Safer access to education
  • Reduced hesitation

Long-term

  • Better awareness
  • Reduced misinformation
  • Cultural shift toward openness

A future where sensitive learning is safe, private, and stigma-free

Built With

  • ai-based-moderation
  • and-a-custom-built-multi-layer-guardrail-system-combining-normalization
  • express.js
  • keyword/regex-filtering
  • local-storage
  • node.js
  • openai-api-(llm)
  • prompt-engineering
  • react-(next.js)
  • restful-apis
  • risk-scoring
  • rule-based-safety
  • typescript
Share this project:

Updates