FamilyFlow AI

Less scrolling. More living together. Parent-approved adventures across family screens, voice, and AWS.

Comment

FamilyFlow AI — Less scrolling. More living together.

FamilyFlow AI is a parent-approved family adventure coordination platform connecting a web dashboard, television-friendly interface, Alexa voice skill, and AWS services. Families browse age-banded activities, children request adventures, parents approve requests, and parents verify completion before non-purchasable rewards are awarded.

Customer-facing landing page: https://familyflow-ecru.vercel.app/ Open-source code and testing instructions: https://github.com/LightLLM/FamilyFlow-AI

Inspiration

Families have abundant screen entertainment but often struggle to coordinate meaningful shared activities. FamilyFlow uses the living-room screen and voice interface as starting points for offline family adventures.

What it does

  • Curated age-banded family adventure library
  • Next.js parent dashboard for household setup, planning, approvals, and activity history
  • TV-friendly 10-foot browser shell with device pairing and activity requests
  • Alexa custom skill, plus an Alexa+ MCP preview integration
  • Parent-controlled approval and completion verification; points awarded once
  • Optional Amazon Bedrock suggestions with curated fallback and mandatory parent review
  • Privacy-oriented design: age bands instead of birth dates, no unrelated Alexa or Fire TV activity collection

How we built it

A pnpm/TypeScript monorepo combines Next.js, Tailwind, Alexa Skills Kit, an MCP server, AWS API Gateway and Node.js Lambda, Cognito JWT parent authentication, device pairing tokens, DynamoDB, CDK, CloudWatch, and AWS Budgets. Reusable domain packages enforce approvals, idempotent rewards, and audit sanitization.

Challenges

Separating parent credentials from TV device sessions, keeping approval gates consistent across interfaces, and coordinating voice, web, and TV workflows. Full Fire TV/Vega device validation and Alexa OAuth linking remain outstanding.

Accomplishments

  • Deployed AWS sandbox API and verified /health endpoint
  • 28/28 unit tests and 17/17 automated E2E tests passing as documented in the repository README
  • Alexa skill created; launch utterance tested successfully in Developer Console
  • Working browser-based TV shell and parent dashboard code
  • Documented architecture, security, deployment, testing, and known limitations
  • Customer-facing landing page linked above

What we learned

Cross-device family experiences need device-specific interfaces and strong authorization boundaries. AI can suggest an activity, but parents must decide whether to approve it.

What's next

Complete Alexa account linking and MCP demo validation, run Fire TV/Vega simulator/device testing, validate Bedrock model access, and complete privacy/security review before onboarding real families.

Testing

See README for setup. Run pnpm install, build shared/domain packages, bash scripts/run-tests.sh, and bash scripts/e2e-fire-tv-alexa.sh. A browser TV shell is not a native Fire TV/Vega application; device testing is not claimed complete.

Built With

  • alexa-skills-kit
  • amazon-bedrock
  • amazon-cognito
  • api-gateway
  • aws-cdk
  • aws-lambda
  • cloudwatch
  • dynamodb
  • model-context-protocol
  • next.js
  • pnpm
  • tailwind-css
  • typescript
Share this project:

Updates

Submission history