Inspiration
I was inspired to build a platform that helps people discreetly get support in uncomfortable or unsafe public situations by making it easier to reach trusted contacts when speaking openly is not an option. Street harassment is a widespread issue, not just a feeling of unease: in a national U.S. study by Stop Street Harassment, 65% of women reported experiencing street harassment in their lifetime, and 57% reported verbal harassment. The same research found that among men, 25% reported experiencing street harassment, showing that this issue affects multiple groups, but women experience it at much higher rates. Fear of harassment is also common across genders: a 2024 survey found that 44% of women and 21% of men feel unsafe when walking in the dark, while 51% of women said being followed by a stranger was one of their top fears.
Coming from a woman-centric household where the majority of my family members are women, I became more aware of the ways safety concerns shape everyday behavior in public spaces. This perspective helped motivate my interest in building a platform that could provide discreet, immediate support during uncomfortable or potentially unsafe situations. Rather than placing the burden on individuals to explain themselves openly in moments of stress, I wanted to create a system that could quickly connect them to trusted contacts and facilitate a safer exit from the situation.
What it does
Urgentic is a mobile-first web app for discreet safety communication when someone needs help without drawing attention. In situations that feel unsafe or awkward, the user can quickly alert trusted contacts they’ve already added; the app places outbound voice calls through Vapi and Eleven Labs so each contact hears context, options to respond, and a prompt to share where they are. The reporter’s approximate location can come from device GPS (reverse-geocoded for a readable address) or fall back gracefully when location isn’t available. For multi-contact alerts, everyone shares one session so the user sees live call status, each contact’s response and self-reported location, and once calls finish, plain-language “what to do next” guidance (AI-assisted with a rule-based fallback). The session can also surface a nearby public place (e.g. police, hospital, library) on a map preview with walking directions, clearly labeled as an approximate suggestion.
from home → choose contacts → call in progress → results and guidance
How we built it
Backend: Next.js API routes on Node, Supabase (PostgreSQL) for contacts and alerts, service-role access for server-side writes
Frontend: Next.js 14, React 18, TypeScript, Tailwind CSS, shadcn-style UI
APIs & services used:
Vapi - Outbound phone calls, assistant configuration, server webhooks for end-of-call data and structured outputs, uses ElevenLabs for Natural text-to-speech / voice for the calling agent, configured on the assistant in the Vapi dashboard
OpenAI - Post-call session guidance; Vapi agents can also use OpenAI (or another provider) for LLM behavior as configured in Vapi
Mapbox - Geocoding, Search Box for nearby public places, Static Images for map previews; walking directions open in Google Maps
Challenges we ran into
Spent a significant amount off time trying to implement a feature to automatically send text message updates on locations, however to get a verified number to send texts takes a whole process to be approved
Accomplishments that we're proud of
Concurrent calling of voice agents is a cool feature that allows the user to receive help from multiple people at once.
This is important because think of other alternatives:
An SOS call might not be appropriate because certain situations might not need to escalate to the ambulance or police immediately
Calling or texting a friend could possibly aggravate or incite something within the aggresor and escalate the situation more. Furthermore you'd have to wait linearly for each call/text.
Urgentic is the fine line between not wanting to have first responders/law enforcement and not wanting to be seen contacting someone.
What's next for Urgentic
Implementing this into an actual mobile app.
Built With
- mapbox
- nextjs
- openai
- react
- supabase
- typescript
- vapi
Log in or sign up for Devpost to join the conversation.