SmartGuard AI

• 

  Image of the code

Inspiration

Home security notifications often flood users with generic "Motion Detected" alerts without providing actual context on what is happening. We wanted to make front door security smarter, less intrusive, and immediately actionable using AI.

What it does

SmartGuard AI captures Ring front-door camera and sensor events in real time and sends the context through Amazon Bedrock. The generative AI model evaluates the metadata (e.g., visitor duration, package delivery context, behavior patterns) to instantly generate:

1. A 1-sentence plain-language safety summary.
2. A severity rating (Low, Medium, High).
3. Recommended homeowner actions (e.g., "Check live stream", "Use two-way talk").

How we built it

• Ring Integration: Configured a Private Ring App using Ring API endpoints and event payloads to capture camera motion triggers.
• AWS Bedrock Engine: Utilized Python and boto3 (bedrock-runtime) to send prompt-engineered event parameters to Amazon Bedrock for rapid AI security evaluation.
• Language & Runtime: Built with Python 3 inside a virtual environment for fast execution.

Challenges we ran into

Setting up local Python dependencies on macOS required isolating package execution inside virtual environments (venv) to comply with modern system constraints. In addition, prompt engineering for Bedrock was refined to ensure security outputs remained concise and concise for quick readability.

Accomplishments that we're proud of

Successfully building a complete end-to-end bridge between Ring device events and AWS Generative AI services without requiring paid hardware.

What's next for SmartGuard AI

• Connecting live video stream frames directly into multimodal vision models in AWS Bedrock.
• Adding automated SMS / Push notifications via AWS SNS.

Built With

• amazon
• amazon-web-services
• api
• bedrock
• boto3
• gen-ai
• gitthub
• python
• rest-api
• ring