Your privacy-first, multimodal, on-device AI communication copilot, built for the Google Chrome Built-in AI Challenge.

Verve AI Assistant is a powerful web application that pushes the boundaries of what's possible with on-device AI. It leverages the entire suite of Chrome's built-in AI APIs, including the advanced multimodal capabilities of the Prompt API, to provide a comprehensive set of tools for writing, editing, translating, and analyzing content in various formats—all without your data ever leaving your machine.

Verve AI Assistant is more than a tool; it's a complete, intelligent workspace.

Interface	Architecture

• 🧠 The Brain: Advanced Prompt API

  The core of the assistant, allowing direct interaction with the Gemini Nano model. It's fully featured:

  • Text Analysis: Perform complex tasks like creating tables or extracting information.
  • Creative Control: Fine-tune the AI's creativity with Temperature and Top-K sliders.
  • Conversation Memory: The AI remembers the context of your conversation for follow-up questions. Includes a "Reset Session" button for full control.
  • 🎤 Voice-to-Text (Audio): Transcribe spoken words directly into the text field using your microphone.
  • 🖼️ Image Analysis (Image): Upload an image and ask the AI to describe or analyze it. The implementation uses a robust, multi-format approach to ensure compatibility.
  • Advanced Controls: All operations support real-time streaming and can be canceled with a "Stop" button.

• ✍️ Truly Interactive Proofreader (Proofreader API)

  Go beyond simple correction. Verve AI provides a professional editing experience:

  • Highlights Errors: Uses the CSS Highlight API to underline spelling, grammar, and punctuation errors directly in the input field without slowing down the app.
  • Color-Coded Legend: A dynamic legend explains what each color-coded underline means.
  • Interactive Corrections: Uses the Popover API to show suggestions when you hover over or click on an error, allowing you to accept fixes instantly.

• 🛠️ Full Writing Assistance Suite

  A complete set of tools for any writing task, each with advanced controls like streaming, cancellation, and shared context.

  • Intelligent Rewriter (Rewriter API): Refine your drafts by adjusting tone, length, and output format (plain text/markdown).
  • Advanced Writer (Writer API): Generate high-quality text from a simple prompt with full control over tone, length, and format.

• 🌐 Universal Translator

  • Real-time Detection (Language Detector API): The assistant automatically detects the source language as you type, displaying it for confirmation.
  • Multi-language Translation (Translator API): Translate text between multiple supported languages with a clean, simple UI.

• 📄 Smart Summarizer (Summarizer API)

  Distill long texts into clear, concise summaries. Choose the type (tldr, key-points, headline), control the length and format, and see the result generated in real-time.

• ✨ Thoughtful User Experience

  • API Chaining: A dedicated ↩️ button allows you to instantly copy the result back into the input field for multi-step workflows (e.g., Summarize -> Rewrite -> Translate).
  • Context Awareness: The UI always shows you which function was used last, so you never lose track of your work.

• Core: HTML5, CSS3, Vanilla JavaScript (ES6 Modules), Pico.css
• Architecture: Fully modularized codebase with clear separation of concerns for each API handler.
• AI: Google Chrome Built-in AI APIs
  • Prompt API (Text, Audio, Image)
  • Language Detector API
  • Translator API
  • Summarizer API
  • Writer API
  • Rewriter API
  • Proofreader API
• Advanced Web Technologies:
  • contenteditable for the interactive input field.
  • CSS Highlight API for high-performance error highlighting.
  • Popover API for interactive correction suggestions.
  • MediaRecorder API for capturing audio from the microphone.
  • AbortController for cancellable AI operations.
  • CustomEvent for cross-module communication.

To run this project locally, you need a compatible version of Google Chrome Canary with the necessary experimental flags enabled.

1. Open Chrome Canary and navigate to chrome://flags.
2. Search for and Enable all flags related to "Built-in AI", "Prompt API", "Rewriter API", etc.
3. Relaunch the browser.

This project requires Origin Trial tokens to function. You will need to register for them and place them in the <head> of the index.html file.

1. Clone the repository:

   git clone https://github.com/vero-code/verve-ai-assistant.git

2. Navigate to the project directory:

   cd verve-ai-assistant

3. Since the project uses ES Modules, you need to serve it from a local server. The easiest way is using Python's built-in server:

   # For Python 3
   python -m http.server

4. Open your browser and go to http://localhost:8000.

This project is licensed under the MIT License. See the LICENSE file for details.