Glimpse

Glimpse (prev. OneLook) is an intelligent SaaS platform designed to declutter your inbox by transforming individual newsletter subscriptions into a single, comprehensive daily digest. By leveraging Google Gemini 3, Glimpse reduces complex emails into concise summaries, allowing users to stay informed while saving time.

Usage

Link to the web app: https://glimpse-digest.vercel.app/

1. Sign Up/Login: Access the frontend and log in. (Restricted right now since app isn't published through Google)
2. Connect Gmail: Use the Google OAuth flow to login.
3. Dashboard: View your daily digests (generated by Google Gemini 3) and sorted newsletters.
4. Audio Experience: Click the "Play" button on your digest to generate and listen to an audio summary (powered by ElevenLabs / Gemini 1.5 TTS).
5. Email Notifications: Check your inbox daily (maybe in spam / junk folder !!) for a summary of your subscriptions.

Since the login is restricted right now, you can use the demo button to view the experience on a pre logged in account.

Features

Core Experience

• Automated Ingestion: Seamlessly integrates with Gmail to identify and collect newsletter emails.
• Intelligent Summarization: Utilizes Google Gemini 3 (with custom thinking levels) to generate high-quality, concise summaries of long-form content.
• Daily Digest: Aggregates all processed newsletters into a single, easy-to-read daily report.
• Analytics & Tracking: Monitors token usage, model performance, and ingestion metrics.

Advanced Capabilities

• Audio Digests: Listen to your newsletters on the go. Glimpse converts your daily summary into an audio briefing using ElevenLabs (for premium realistic voices) or Gemini TTS.
• Email Updates: Receive a beautifully formatted summary directly in your inbox every morning, ensuring you never miss a beat even if you don't log in to the dashboard.

Architecture

The system follows a modern asynchronous architecture to handle high-volume email processing and AI tasks efficiently.

graph TD
    User((User))
    Web[Frontend: Next.js]
    API[Backend API: FastAPI]
    Worker[Celery Workers]
    Redis[(Task Queue: Redis)]
    DB[(Database: PostgreSQL)]
    LLM[AI Layer: Gemini]
    S3[Object Storage: S3/R2]
    TTS[TTS: ElevenLabs/Gemini]

    User <--> Web
    Web <--> API
    API <--> DB
    API --> Redis
    Redis <--> Worker
    Worker --> DB
    Worker <--> LLM
    Worker --> S3
    Worker --> TTS

Loading

AI & Background Processing

Glimpse uses Celery to orchestrate complex AI workflows, ensuring the API remains responsive.

1. Parallel Summarization: When a digest is triggered, summarize_emails_batch processes multiple emails concurrently to reduce latency.
2. Context Synthesis: A final summarize_digest_overall task aggregates individual summaries into a cohesive narrative.
3. Token Tracking: Every interaction with Gemini is logged with input/output token counts to monitor cost and performance.

Tech Stack

• Backend: Python 3.11+, FastAPI, Celery, SQLAlchemy
• Frontend: Next.js 14, TypeScript, Tailwind CSS
• Database: PostgreSQL (Supabase/Local)
• Task Queue: Redis
• AI/LLM: Google Gemini 3
• Audio: ElevenLabs / Gemini 1.5 TTS
• Infrastructure: Docker, Google Cloud (OAuth, Gmail API)

Getting Started

Follow these instructions to set up Glimpse on your local machine.

Prerequisites

• Python 3.11+
• Node.js 18+
• PostgreSQL (Local or cloud instance)
• Redis (Local or cloud instance)
• Google Cloud Credentials:
  • OAuth 2.0 Client ID
  • API Key for Gemini
• AWS Credentials (for S3 storage)
• ElevenLabs API Key (Optional, for high-quality audio)

Backend Setup

1. Navigate to the backend directory:

   cd backend

2. Install dependencies: We use uv for fast dependency management.

   pip install uv
   uv sync

3. Configure Environment: Create a .env file in the backend/ directory with the following variables:

   DATABASE_URL=<url to postgresql database>
   SECRET_KEY = <secret key here> (for Fernet encryption)
   SYSTEM_EMAIL_REFRESH_TOKEN=<refresh token for system email> (system email for sending email notifs)
   
   GOOGLE_CLIENT_ID = <google client id of cloud project>
   GOOGLE_CLIENT_SECRET = <client secret of cloud project>
   GOOGLE_REDIRECT_URI = http://127.0.0.1:8000/auth/google/callback
   GOOGLE_API_KEY = <api key of cloud project>
   
   ELEVENLABS_API_KEY = <elevenlabs api key here>
   
   AWS_SECRET_ACCESS_KEY=<aws secret access key here>
   AWS_ACCESS_KEY_ID=<aws access key id here>
   AWS_REGION=<aws region here>
   S3_BUCKET_NAME=<s3 bucket name here>

4. Run the Services: You will need to run the web server and the Celery workers. (or use poe to run these tasks, check pyproject.toml)

   Terminal 1: Web Server

   uv run uvicorn main:app --reload

   Terminal 2: Celery Worker

   uv run celery -A celery_app worker --loglevel=info

   Terminal 3: Celery Beat (Scheduler)

   uv run celery -A celery_app beat --loglevel=info

   Optional: Use redis-server to run a local Redis instance.

Frontend Setup

1. Navigate to the frontend directory:

   cd frontend

2. Install dependencies:

   npm install

3. Configure Environment: Create a .env.local file in the frontend/ directory:

   NEXT_PUBLIC_BACKEND_URL=http://localhost:8000

4. Start the Development Server:

   npm run dev

   The application will be available at http://localhost:3000.

Project Structure

glimpse/
├── backend/
│   ├── app/                 # Main application logic
│   │   ├── ai/              # AI summarization & TTS logic
│   │   ├── api/             # API dependencies
│   │   ├── core/            # Config and settings
│   │   ├── db/              # Database models and migrations
│   │   ├── routes/          # API endpoints
│   │   └── tasks/           # Celery background tasks (Sync/Digest/Audio)
│   ├── main.py              # Application entry point
│   ├── celery_app.py        # Celery configuration
│   ├── Dockerfile           # Backend container definition
│   └── pyproject.toml       # Python dependencies
├── frontend/
│   ├── app/                 # Next.js App Router pages
│   ├── components/          # Reusable UI components
│   └── lib/                 # Utilities and API clients
└── README.md                # Project documentation

Deployment

Frontend

The frontend is a standard Next.js application and is optimized for Vercel.

1. Push your code to a Git repository.
2. Import the project into Vercel.
3. Set the NEXT_PUBLIC_BACKEND_URL environment variable in the Vercel dashboard.

Backend & Infrastructure

The backend is containerized and stateless, making it easy to deploy on any cloud provider.

1. Compute: Deploy the Docker container to Google Cloud Run.
2. Database: Use a managed PostgreSQL instance (we use Supabase).
3. Task Queue: Use Redis Cloud (or a managed Redis instance) for the Celery broker. This is critical for production reliability.
4. Configuration: Ensure all environment variables (DATABASE_URL, GOOGLE_API_KEY, ELEVENLABS_API_KEY, etc.) are set in your production environment.