SSSS M M A RRRR TTTTT DDDD EEEEE TTTTT EEEEE CCCC TTTTT OOO RRRR
S MM MM A A R R T D D E T E C T O O R R
SSS M M M AAAAA RRRR T D D EEEE T EEEE C T O O RRRR
S M M A A R R T D D E T E C T O O R R
SSSS M M A A R RR T DDDD EEEEE T EEEEE CCCC T OOO R RR
CONNECTED SIGNALS. CLEAR INCIDENTS. FASTER RESPONSE.
A native Vega TV experience that turns connected-camera events into something you can browse, understand and review from your couch.
Start the judge walkthrough · Technical README · AWS architecture · Emulator download (x86_64) · Fire TV download (ARMv7)
| Where you will test | Download | Account and connection |
|---|---|---|
| Vega emulator on Ubuntu | Emulator app — x86_64 .vpkg |
Production: smartdetector.com, using your free account. |
| Supported physical Vega Fire TV | Fire TV app — ARMv7 .vpkg |
Production: smartdetector.com, using your free account. |
Same app version, different hardware packages—do not interchange them. Both are standalone Release builds; no Docker, localhost server or Metro is needed. Create your free account, launch the appropriate app, then approve its pairing code at smartdetector.com/tv/connect. No paid SmartDetector plan is needed for basic testing; analysis availability depends on configured quotas and connected cameras. The emulator package becomes available after the first successful dual-target release; older releases may have only ARMv7.
I’ve owned smartdetector.com for a few years. Originally, I wanted to build a smarter smoke detector—and I still do. Before this hackathon, though, SmartDetector was just the name and an idea. The SmartDetector project was built entirely during the hackathon.
Ring’s cameras and device APIs gave that idea a practical starting point: real-world activity we could bring into one connected experience. Instead of another isolated dashboard, we wanted one perspective across the devices around us.
Our long-term vision goes beyond cameras: more smart devices, more useful context, and eventually dedicated SmartDetector hardware. This first version starts with the moments your cameras already capture.
We chose the Fire TV track of the Build, Ship, Shape hackathon to build with the new Vega OS and React Native. We wanted checking your home to feel as natural as browsing your favorite shows.
Inspired by Ring and Prime Video, SmartDetector treats cameras like collections and events like episodes. Use your remote to browse a camera, select a moment, and inspect its sampled images and available observations.
Ring supplies the activity. SmartDetector organizes it. Vega brings it to your screen.
UI captured on a physical Vega Fire TV. Camera imagery is replaced with similar AI-generated scenes for privacy; the interface is unchanged.
Watch the browser UI walkthrough · Native capture details
- Connect your cameras. Authorize Ring access and choose the location and cameras you want to share.
- Pair your TV. Scan a QR code, sign in on your phone, and approve the display. No password or API key goes on the TV.
- Browse what happened. Open camera and activity rails, history, and saved-event search using remote-friendly navigation.
- Review the evidence. Step through sampled frames and inspect available detections and AI context.
- Stay in control. Use location-scoped access, revocable read-only displays, data purge and account deletion. Saved event/image retention defaults to 24 hours.
Playwright capture of the pairing interface. The displayed code is a nonfunctional fixture; use the code on your own running app.
The same connected-device dashboard on the web, with similar AI-generated camera scenes substituted for privacy.
View the complete screenshot gallery
The app also includes foreground alerts. An ADM background receiver and actionable system-popup integration are implemented; cross-app popup/image behavior remains a physical-device verification item. Native live video is not implemented: live-view links use the web console. We keep those boundaries explicit rather than presenting planned capabilities as finished features.
The hosted SmartDetector website manages accounts, locations, device connections and incident history. The native TV is a paired, read-only viewer of the same retained evidence.
The backend uses ECS/Fargate, DynamoDB, SQS, optional EventBridge, S3, IAM and CloudFormation. SmartDetector’s own Amazon Bedrock integration summarizes supplied incident observations using the configured model and the installation’s AWS permissions. Configured delivery integrations include SES email and ADM for background TV alerts.
Optional detection add-on: AutoVision is a separately developed external service, outside this hackathon submission. SmartDetector’s integration is free to add and can provide additional visual detection features when configured. External hosting, inference costs and usage limits are separate. AutoVision’s implementation and features are not claimed as SmartDetector hackathon work.
Measurements and AI observations remain distinct. A label is not proof of identity, intent or danger; SmartDetector is not a certified emergency alarm.
We documented the service choices, alternatives and configuration boundaries in the AWS architecture guide.
We are also submitting for the AWS Builder and Open Source mini challenges. The AWS guide explains the backend; the separate SmartDetector MCP / AgentCore integration provides scoped access to SmartDetector’s REST API for agent clients, including a deployment path for AWS AgentCore.
Its five tools are read-only: agents can inspect devices and events, not control cameras or change incidents. Local verification and deployable templates are separate from proof of a running AWS AgentCore invocation. See the technical reference for setup and security boundaries.
Judges: follow the illustrated field guide. It starts at simulator installation, then walks through free SmartDetector signup, TV approval, Ring staging access, and the no-camera developer route. You do not need to buy a SmartDetector plan for the basic walkthrough; contact us if quotas block evaluation.
- Have a Vega Fire TV? Download the latest ARMv7 app and install it using the field guide. Release details and checksum. Until the first public app release is published, the v0.1.18 workflow download remains available with GitHub sign-in.
- Using the simulator? Follow the guide’s x86_64 build instructions; the physical-TV package is not the simulator package.
- No Ring camera? Use the documented localhost Ring Playground path. Sample-stream analysis shares the processing pipeline but is not a production motion-webhook test.
- Building or deploying? Open the technical README, Ubuntu setup, or CloudFormation guide.
A new device platform brings real learning: runtime differences, notifications, device pairing, simulator setup and release packaging. Our developer friction report records the observed issues, workarounds and actionable suggestions, including the successful headless release build.
Next, we want broader device integrations, more useful and carefully evaluated AI, and native mobile companions that work alongside Fire TV. Platform improvements we would welcome include consistent notifications across supported React Native versions, a complete background-delivery sample, and Ring Playground scenarios that deliver signed motion webhooks as well as sample video.
SmartDetector started with a name. This hackathon gave it a working native TV experience—and a foundation for bringing connected devices into one perspective.
Judge field guide · Technical README · AWS choices · Friction report · Privacy policy · Support
First-party code is AGPL-3.0-or-later. Third-party components retain their licenses. Contact support about separate terms for maintainer-owned code.