1.Survivor Resource Navigator
Survivors, vulnerable workers, migrants, and people seeking help struggle to find safe, relevant services across shelter, legal aid, healthcare, transportation, language support, and hotlines. Build a multilingual assistant that helps users find safe and relevant services based on their needs, location, language, and situation.
Start here — Search a small curated resource directory and return matched services.
Push further — Add multilingual support, eligibility filtering, location-aware recommendations, source citations, and safety-sensitive language.
Helps survivors, caseworkers, hotline teams, NGOs, migrant support groups · Uses RAG, multilingual chatbot, resource matching, eligibility filtering, citation-based answers · Demo: a user asks "I need legal help and safe housing" and gets relevant resources, eligibility notes, and safe next steps.
2.Caseworker Copilot
Caseworkers and frontline staff manage complex notes, timelines, referrals, appointments, safety plans, and follow-ups with limited time and outdated tools. Build an AI assistant that helps caseworkers summarize notes, organize timelines, track referrals, generate checklists, and reduce administrative burden.
Start here — Upload case notes and generate a summary plus next-step checklist.
Push further — Add referral tracking, risk flags, privacy controls, role-based outputs, and structured case timelines.
Helps social workers, shelter teams, nonprofit case managers, legal aid teams, outreach workers · Uses summarization, structured extraction, timeline generation, task generation, workflow automation · Demo: messy case notes becoming a clear summary, timeline, referral list, and next-step action plan.
3.Survivor-Safe AI Communication Guardrails
AI chatbots, companions, and support agents increasingly interact with survivors, vulnerable workers, and youth. Poorly designed responses can cause harm — asking unsafe questions, giving risky advice, sounding judgmental, overstepping as a therapist, or failing to escalate urgent danger. Build a guardrail system or evaluator that helps survivor-facing AI systems communicate with dignity, privacy, caution, and trauma-informed care.
Start here — A tool where users paste a survivor message and an AI response. The tool labels the response as safe, risky, or unsafe, explains why, and suggests a safer version.
Push further — A guardrail layer between user and AI that detects high-risk conversations, checks outputs for harm, rewrites unsafe responses, protects sensitive details, and routes urgent situations to human or emergency support.
Helps survivors, survivor-serving organizations, hotline teams, AI developers, youth-support platforms · Uses LLM safety guardrails, response evaluation, trauma-informed rewriting, crisis-escalation detection · Demo: a risky chatbot response to a survivor disclosure, an explanation of what's unsafe, and a safer, trauma-informed rewrite.
4.Hospitality & Transportation Awareness Assistant
Hotel staff, rideshare drivers, transportation workers, venue staff, and volunteers may notice concerning situations but not know how to respond safely. Poorly handled responses can put vulnerable people at greater risk. Build a mobile-friendly training or guidance tool that helps frontline public-facing workers recognize concerns and respond safely, respectfully, and without vigilantism.
Start here — A short mobile training flow with common scenarios and safe response guidance.
Push further — Role-specific training for hotels, rideshare, restaurants, venues, transit staff, and volunteers, with multilingual support and local resource handoffs.
Helps hotels, rideshare drivers, transit workers, venue staff, restaurant teams, volunteers, tourism partners · Uses scenario-based training, multilingual microlearning, safety decision tree, role-specific guidance · Demo: a user answers scenario questions and gets safe, practical guidance on what to do and where to refer concerns.
Not feeling these? Combine ideas, adapt to your local context, or propose a completely new solution within this track. A focused prototype that works beats an ambitious idea that stays abstract.
Ready to build? Register now · Join our Discord · Submissions close July 9, 2026.
← Back to all tracks