creating a guest pass for ensured entry .
every incidents noted timeline
automatic incident triger
AI incident investigator

Inspiration

Residential communities need reliable ways to monitor visitor access and investigate suspicious activity. Security guards may need to check visitor passes, review camera events, and compare gate records while responding quickly to potential incidents. We created VIGILYX to explore how AI can help connect these pieces of information and make security investigations faster and clearer.

What it does

VIGILYX is an AI-assisted residential security dashboard designed to support visitor verification and incident investigation. It includes a guest pass register, a security event timeline, rule-based automatic incident triage, and an AI incident investigator powered by a locally running language model.

The prototype uses simulated Ring camera motion events and gate-entry signals to demonstrate how security teams could identify situations requiring further review. The system provides recommended next steps while keeping human guards responsible for verification and decisions.

How we built it

We built VIGILYX using React, Vite, and JavaScript for the dashboard interface. Ollama runs the Qwen 2.5 3B language model locally to generate incident investigation summaries without requiring a paid cloud AI API. Rule-based logic evaluates the available demo records and highlights potential concerns.

Challenges we ran into

One challenge was designing useful incident alerts without treating a camera motion event as proof of unauthorized entry. We also worked to integrate local AI with the dashboard and make the interface understandable for security personnel. Keeping the prototype accessible without paid AI services was another important consideration.

Accomplishments that we're proud of

We created a working interactive dashboard that combines visitor pass management, simulated security events, automatic incident triage, and local AI-assisted investigation in one interface. We are proud that the prototype demonstrates a practical security workflow while emphasizing human oversight.

What we learned

We learned how to build a React dashboard, integrate a locally hosted language model, and combine rule-based logic with AI-generated explanations. We also learned that security AI must communicate uncertainty clearly and should support—not replace—human judgment.

What's next for VIGILYX

Our next steps include exploring real, authorized Ring integrations, connecting supported gate-controller or sensor data, improving event correlation, adding audit logs, and testing the system with realistic scenarios. Any real-world deployment would require appropriate permissions, security testing, and human-controlled access policies.