Inspiration
In a world where women continue to face harassment, legal hurdles, and emotional stress, access to timely and trustworthy help can be life-changing. Many don’t know where to turn in emergencies or whom to trust for legal and mental support — especially in rural or underserved areas.
DAYA was born out of the need for a non-judgmental, always-available, AI-powered companion that can offer women quick support, legal helplines, emotional comfort, and practical advice — whenever they need it most.
What it does
DAYA is a conversational AI agent built to assist women in real-time with: Emergency detection & helpline response (based on location and intent) Legal aid awareness and access to trusted government/legal services Emotional support during distress Friendly, mental health–oriented conversation Context-aware chat with memory of recent messages All conversations are conducted in a warm, respectful tone — like chatting with a supportive friend.
How I built it
Agent framework: Fetch.ai uAgents LLM engine: ASI:One LLM (asi1-mini) Platform: Agentverse (deployed and testable) Backend logic: Python, uAgents, OpenAI API wrapper for ASI Memory: Conversation history stored in a short-term buffer for contextual responses Location-aware emergency routing: Keywords + state detection trigger helpline lookups from a curated database
Challenges I ran into
Designing an intent-aware emergency detection system that feels natural yet urgent Mapping user messages to their correct state (India has 28+ states and many variations in phrasing) Creating a memory-aware assistant without external state persistence Ensuring sensitive topics like abuse or harassment are handled responsibly and empathetically
Accomplishments that I am proud of
Built and deployed a working conversational safety agent on Agentverse from scratch in just days Successfully integrated ASI:One’s LLM to provide coherent, emotionally aware support Created a flexible system for triggering regional emergency helplines dynamically Kept the entire stack lightweight and agent-ready using decentralized uAgents te
What I learned
How to create and deploy AI agents using Fetch.ai's uAgents framework How to combine emotional intelligence with rule-based triggers in an LLM-driven agent The importance of empathetic design in user-facing conversational tools How LLMs can augment public service applications with contextual awareness
What's next for DAYA
Add native language support (Hindi, Tamil, etc.) for wider accessibility Integrate voice + IVR support for low-literacy or visually impaired users Offer legal form-filling help and status tracking for schemes like Domestic Violence Act support or legal aid Include real-time GPS-based emergency routing and one-click SOS Launch a community version for women to connect, share, and support each other anonymously
Log in or sign up for Devpost to join the conversation.