VEA - Violence Ending Agent

Gender-Based Violence Chat Support

Comment

VEA – Violence Ending Agent

Inspiration

The inspiration behind VEA – Violence Ending Agent came from the urgent need to provide discreet, safe, and immediate support for survivors of gender-based violence (GBV). Many survivors face barriers to accessing help due to fear, stigma, or lack of awareness of available resources.

We wanted to create a tool that is accessible, private, and empathetic, allowing survivors to seek guidance at their own pace without judgment. Real-life stories of survivors highlighted the need for immediate support in a safe digital environment, which directly motivated this project.

What it does

VEA is a confidential AI-driven support agent designed to help users in distress. Its core features include:

  • Discreet Interaction – Disguised as a general wellness tool to ensure safety in sensitive environments.
  • Guided Emotional Support – Structured conversations help users share their feelings safely.
  • Crisis Detection – Recognizes crisis keywords (e.g., "I feel hopeless") and prompts professional support.
  • Professional Mode – Provides a structured, counselor-like interaction with empathy and safety-first responses.
  • Support Options – Offers talking through feelings, connecting to hotlines, creating safety plans, or emergency support.
  • Offline Functionality – Scripted fallback ensures support even without internet or AI API access.

VEA balances empathy, accessibility, and safety, acting as a digital companion for those in need.

How we built it

We built VEA using:

  • Frontend: React for a responsive chat interface with file uploads and multiple chat threads.
  • Backend: Node.js + Express to manage chat flow, professional scripts, and crisis detection.
  • AI Integration: Scripted AI responses and optional AI API for conversational support.
  • Crisis Management: Detects keywords to trigger professional support mode.
  • Professional Scripts: Branching decision scripts ensure safe, empathetic guidance.
  • Export Feature: Users can export chat logs as PDFs to track progress or for safe reference.
  • Security & Privacy: All sensitive user data remains confidential and secure.

Challenges we ran into

  • Balancing AI-generated responses with safe, empathetic support.
  • Designing offline fallback scripts for reliable assistance without internet.
  • Maintaining user privacy while allowing professional support.
  • Handling complex, unpredictable conversation flows in the chat.
  • Creating a discreet yet friendly UI for real-world scenarios.

Accomplishments

  • Developed a working prototype of a confidential GBV support agent.
  • Created a professional-mode script providing safe, structured guidance.
  • Implemented crisis detection and branching logic for immediate intervention.
  • Added PDF export functionality for secure chat records.
  • Built a user-friendly and discreet interface suitable for sensitive environments.

What we learned

  • Empathy in technology matters – the tone and wording of automated responses impact trust.
  • Safety-first design is essential in trauma-sensitive applications.
  • Iterative UX testing is crucial to refine conversation flows and accessibility.
  • Collaboration across frontend, backend, AI, and UX teams is key for success.

What's next for VEA

  • Enhanced AI Integration: Smarter context-aware responses while keeping safety.
  • Multi-language Support: Reach a broader audience with diverse language options.
  • Mobile App Deployment: Offline-capable mobile app for easy access.
  • Advanced Safety Features: Location-based emergency services and one-tap hotline contacts.
  • Continuous Script Improvement: Refine professional scripts using anonymized user feedback.

VEA is our commitment to empowering survivors, providing discreet and safe support, and ensuring nobody has to face their struggles alone.

Share this project:

Updates

Submission history