Prida

Inspiration

Every time we apply for a job, fill out a form, or respond to a detailed question online, we end up retyping the same information again and again. Resumes, bios, medical histories, or even simple sign-ups, each site asks for the same details in slightly different ways. The process is slow, error-prone, and exhausting.

While existing autofill tools can handle basic fields like names or addresses, they fail when the question becomes subjective like “Which project challenged you the most?” or “Describe your approach to teamwork.” These require context, tone, and understanding.

That’s where PrIDA (Private Data Agent) was born, a Chrome extension that combines on-device AI and personal context to fill, rewrite, and proofread intelligently while keeping all your data private.

What it does

PrIDA brings the power of Chrome’s built-in Gemini Nano AI directly into your browser to make online writing and form-filling smarter, faster, and more personal.

It understands both what the page is asking and who you are by drawing from your saved profiles. Whether you’re applying for a job, filling a medical form, or posting on social media, PrIDA adapts its answers to your chosen context.

Here’s how it helps:

AI-Powered Auto-Fill: Recognizes input labels and page context, then suggests accurate answers based on your selected profile.
Text Rewriting: Improves your writing clarity and tone using the Gemini Nano Rewriter API.
Proofreading: Fixes grammar, punctuation, and spelling mistakes instantly — all offline.
Profile Management: Lets you create separate profiles (e.g., Job, Health, Social) so you stay organized and in control.
Privacy-First Architecture: All processing happens locally via Chrome’s on-device AI. No cloud storage. No data leaks.

In short, PrIDA makes online interaction smarter without compromising privacy or performance.

How we built it

PrIDA was built using React, TypeScript, and Tailwind CSS, with Vite powering the development workflow. The extension’s logic revolves around three main systems:

Auto-Fill Engine: A content script scans pages for all editable fields including rich text editors and injects a small AI icon next to them. Once clicked, it extracts the label and nearby text, retrieves relevant context from the user’s selected profile, and generates an answer using Chrome’s on-device LanguageModel API.

Rewriting & Proofreading System: Right-click menus allow users to rewrite or proofread selected text. Both features rely on Chrome’s Gemini Nano Rewriter and Proofreader APIs, enabling real-time results without sending any data online.

Local Profile Storage: User data and profiles are stored securely in IndexedDB. Each website domain can be linked to a preferred profile for example, a “Job Application” profile for LinkedIn and a “Social Media” profile for Twitter.

Everything from prompt creation to data management happens locally, ensuring both privacy and speed.

Challenges we ran into

Building PrIDA meant balancing technical precision with absolute privacy. The biggest challenge was supporting every kind of input field, especially modern rich text editors that don’t follow standard HTML patterns. To solve this, I implemented Mutation Observers that dynamically detect changes and inject the icon reliably on complex platforms like LinkedIn or Gmail.

Accomplishments that we're proud of

PrIDA showcases how on-device AI can power intelligent browser experiences:

Full integration with Chrome’s Gemini Nano AI APIs (Writer, Rewriter, Proofreader)
Zero network dependency, all requests are processed locally
Dynamic DOM detection for real-time field injection
IndexedDB-based local storage for secure profile handling
Context-aware prompting that blends page content and personal data

By merging privacy, usability, and intelligence, PrIDA demonstrates that the next generation of browser tools can be both personal and private, all without the cloud.

What we learned

I learned how to build a fully private, on-device AI system using Chrome’s built-in Gemini Nano APIs. Developing PrIDA showed me that local AI can match the usefulness of cloud models while keeping user data completely secure. I learned to handle complex form structures, dynamically detect inputs on modern websites, and design an intuitive profile system that makes personal AI assistance both fast and privacy-preserving.

What's next for Prida

PrIDA redefines how people interact with the web. It’s not just about saving time, it’s about giving users control over their personal data while still benefiting from AI assistance. By running entirely on-device, PrIDA proves that powerful, context-aware automation doesn’t need to compromise privacy.

In the future, I plan to expand PrIDA with: