Inspiration

In a flood, earthquake or fire, the first reports are chaos. They arrive as voice notes, texts, photos and calls, often in mixed languages like Hinglish or Tanglish, and the same incident gets reported dozens of times. Control rooms lose critical minutes sorting duplicates, guessing locations and deciding what to handle first. We wanted a system that absorbs the noise so humans can focus on decisions.

What it does

ResQFlow is an AI emergency-response command center that turns messy reports into clear, prioritized, verified incidents.

  • Takes in anything: text, voice and image reports, in multilingual and code-mixed language.
  • Understands each report: AI agents extract the location, type of need, severity cues and supporting evidence.
  • Removes duplicates: many reports about one event are merged into a single incident, with every source report linked as evidence.
  • Triages deterministically: priority comes from transparent, rule-based scoring, not an LLM's guess. The same input always gives the same priority, and every score can be explained.
  • Keeps a human in control: an operator reviews each incident and chooses Verify or Reject. No task is created or dispatched without human approval.
  • Shows the situation live: a live report feed, a priority triage queue, and a map of incidents, shelters, nearby services and volunteers.
  • Simulation mode: a flood scenario generates realistic incoming reports so the full flow can be demonstrated end to end.

Who it serves: people in distress (request help, track requests), volunteers (register, receive tasks), operators and admins (monitor, triage, dispatch) and the public (nearby services, shelters, emergency info).

How we built it

  • The app is built on Base44, with a dark "Tactical Command Center" interface.
  • A Leaflet + OpenStreetMap map needs no API key and no sign-in, so it works anywhere.
  • AI agents handle extraction, language understanding and deduplication, while scoring and priority stay rule-based and explainable.
  • The app has a full sidebar: Home, Dashboard, Request Help, Live Map, Nearby Services, Shelters, Volunteer Register, My Requests, Volunteer Dashboard, Admin, Notifications, About and Emergency Info.

Challenges we ran into

  • Handling code-mixed, multilingual reports of uneven quality.
  • Deciding what AI should and should not do. We use AI for understanding and rules for priority, so decisions stay explainable.
  • Merging duplicate reports without losing the evidence behind each one.
  • Delivering a working demo within a 30-hour hackathon.

Accomplishments that we're proud of

  • A complete flow from a messy report to an approved task, with a human decision at the key step.
  • Explainable triage that an operator can trust and check.
  • A free map with no keys or sign-ins, so deployment is simple.

What we learned

In emergencies, trust matters as much as speed. AI works best when it organizes information and leaves the final decision to a person.

What's next for ResQFlow: AI Command Center for Emergency Response

  • Real voice and image processing at scale
  • Integration with official helplines and SMS or WhatsApp channels
  • Offline-friendly intake for areas with poor connectivity
  • Support for more regional languages
  • Real-time volunteer routing and shelter capacity tracking

Built With

Share this project:

Updates

Submission history