Niblet.ai — AI-Powered Meal & Calorie Tracker

Imagine tracking every meal simply by chatting with your AI friend—no buttons, no forms. Niblet.ai transforms calorie logging into a conversation, making healthy choices feel as effortless as texting a buddy.

Project Overview

Niblet.ai (formerly Tally) is a mobile-first web application designed around a chat interface powered by OpenAI’s Assistant API. Users send a quick message—“I had oatmeal and berries”—and Nibble breaks it down into calories and macros, logs it automatically, and offers motivational feedback. Everything happens in one smooth flow.

Problem Statement

Traditional calorie trackers overload users with manual inputs, drop-downs, and confirmation dialogs. This friction leads to abandoned logs and incomplete data. We needed an intuitive, feature-rich solution that feels natural, conversational, and lightning-fast.

How We Solved It

1. Conversational Interface
   We built a single chat window where users type or speak their meals. Nibble parses the input, logs nutrition, and responds with a friendly breakdown—no extra taps required.

2. AI-Powered Parsing & Logging
   Leveraging OpenAI Assistant API and context memory, the app recognizes foods, quantities, and even past preferences. It auto-logs entries, only prompting when clarification is essential.

3. Mobile-First Design
   The UI, crafted with React and Material UI, adapts seamlessly to any screen. Swipe gestures toggle between Today’s Meals and Progress Chart, keeping primary tasks within thumb’s reach.

4. Seamless Authentication
   Implemented NextAuth with Google Sign‑In for frictionless access. Secure session management with server‑side NextAuth callbacks and role-based admin permissions.

5. Real-Time Data & Persistence
   Firebase Authentication and Firestore power instant syncing and persistent conversation history—even across sessions and page navigations.

Key Features

• 💬 Chat-Driven Logging: One-line messages log meals; AI handles the rest.
• 📈 Unified Chart View: Swipe to view weight trends and calorie vs. consumption graphs with macro filters.
• 🎙️ Voice Interaction: Optional voice‑to‑text input for hands‑free logging.
• 🔒 Secure Auth: NextAuth with Google for quick signup/sign‑in; admin panel for managing AI personalities and prompts.
• ⚡ Fast & Lightweight: Client-side logic in Next.js + TypeScript; serverless functions for AI calls; optimized bundling for sub-2s page loads.

Technical Stack

• Frontend: Next.js, React, TypeScript, Material UI, Chat UI components.
• AI Integration: OpenAI Assistant API with persistent memory for each user thread.
• Authentication: NextAuth (Google OAuth) for client & server sessions.
• Database: Firebase Firestore for real-time chat history and user settings.
• Deployment: Vercel for frontend; Heroku for secure serverless AI proxy and backend functions.

Architecture Snapshot

flowchart TB
  A[User Chat Input] --> B[Next.js API Route]
  B --> C[AI Proxy on Heroku]
  C --> D[OpenAI Assistant]
  D --> C
  C --> B
  B --> E[Firestore]
  E --> B
  B --> F[Frontend Renderer]

Impact & Results

• Engagement: 3× increase in daily logs vs. traditional form‑based trackers.
• Retention: 60% of users return daily after one week.
• Satisfaction: 4.8★ average feedback rating for ease of use and speed.

Next Steps

• 🔌 Flex Fields: Add dynamic fields so the AI can capture custom entries (e.g., water intake, meal ratings).
• 📲 PWA + Mobile App: Package as a Progressive Web App with offline support and push notifications.
• 🤖 Advanced AI Coaching: Introduce multiple AI personalities (Best Friend, Professional Coach, Tough Love) selectable from chat.

Try Niblet.ai

Dive into a smarter way to track meals:
Launch Niblet.ai

Experience meal & calorie tracking—as simple as sending a text.