Alfredo

Alfredo - Voice-Controlled Nutrition Assistant

Comment

Alfredo – A Voice-First AI Nutrition Assistant

About the Project

Alfredo is a browser-based, fully voice-enabled nutrition assistant that lets users manage their meals, pantry, and health goals completely hands-free. With a simple wake word — “Hey Alfredo” — users can log meals, analyze their nutrition, track pantry inventory, and receive smart meal suggestions using just their voice.

The app uses real-time speech recognition and text-to-speech (TTS) inside the browser — no external services like ElevenLabs. All backend data, user profiles, pantry logs, and subscription tiers are handled via Supabase. The UI is sleek and minimal, using a black-and-white theme with subtle shadows for depth and occasional accent colors for clarity.

Core Features

Voice Assistant Integration

  • Wake-word detection: “Hey Alfredo”
  • Browser-based voice recognition with fallback strategies
  • Handles multi-speaker scenarios and background noise
  • Natural voice feedback using TTS with adjustable pitch, rate, and tone

Natural Language Processing

  • Intent recognition for commands like “log two eggs” or “analyze this meal”
  • Context-aware conversation handling with error recovery
  • Custom vocabulary for nutrition-based tasks

Voice Response System

  • In-browser text-to-speech with conversational pacing
  • Prioritized response queueing
  • Dynamic tone adjustments for clarity and user feedback

Nutrition Features

Meal Logging

  • Log meals via voice or image
  • Structured data model for food, portions, and combinations
  • Real-time nutrition engine for calories, macros, and micros
  • Support for verification of unusual entries (e.g. “12 scoops of mayo”)

Smart Analysis & Suggestions

  • Personalized goals based on user profile
  • Nutrition pattern tracking over time
  • AI-based meal suggestions
  • Generate reports and graphs on progress

Pantry Management

  • Tracks inventory based on logged meals and added items
  • Reminds users when items run low
  • Suggests groceries based on planned meals
  • Auto-generates shopping lists

Technical Implementation

Backend (Supabase)

  • Real-time data sync using Supabase subscriptions
  • Row-Level Security (RLS) for per-user data protection
  • Edge Functions for subscription checks and analytics
  • API routes for voice commands, meal logs, and pantry updates

Subscription Management

  • Custom subscription tiers with feature matrices
  • Optional Stripe integration for payments
  • Edge Functions to handle status validation and tier enforcement

Error Handling

  • Full error logging system with fallback messaging
  • Offline sync queue for data logging
  • Graceful degradation for feature unavailability

UI/UX Design

  • Minimal black-and-white interface with box shadows for depth
  • Accessible navigation and voice status indicators
  • Responsive design across all device sizes
  • Visual and audio feedback for recognition states
  • Premium feature indicators for upselling

Testing and Deployment

Testing

  • Unit tests for meal logging, NLP, pantry updates, and subscriptions
  • Integration tests for voice recognition accuracy, real-time sync, and meal analysis
  • Offline support verification and sync tests

Deployment

  • Managed environments: dev, staging, and production
  • Supabase environment separation + secure API key management
  • Feature flags and monitoring for rollout
  • Voice interactions respond within 2 seconds max

Tech Stack

  • Frontend: HTML, CSS (Tailwind), JavaScript
  • Voice: Browser-based Speech Recognition & TTS APIs
  • Backend: Supabase (Auth, Database, Storage, Edge Functions)
  • Design: Black-and-white minimal UI with light accent colors

What We Learned

We learned how to implement real-time voice interfaces inside the browser, manage intent with natural language, and sync everything to Supabase in a secure, scalable way. Creating a voice-first UI without external APIs was a challenge — especially balancing recognition accuracy with natural flow — but the result is a fully working assistant that feels intuitive, helpful, and delightfully minimal.

Built With

  • ai
  • bolt
  • supabase
Share this project:

Updates