AIAssist+

This is a working extensions whoch is been pinned in chrome
this is options.html which allows you to put in API keys or use it in demo mode
it uses different tones to summarize text
it uses different tones to rewrite text
This shows the history of the particualr webpage
It shows different personas the aiassist can use
it shows different operations ai assist can perform

Inspiration

Every day, we drown in open tabs — articles to read, reports to summarize, and notes to take.
Switching between tools for translation, rewriting, and research kills focus.
So I thought: what if the browser itself could think with us?
That idea became AIAssist+ — an AI-powered Chrome extension that turns every webpage into an intelligent, interactive workspace.

What it does

AIAssist+ transforms your browsing experience by bringing Gemini Nano-powered AI directly inside Chrome.
It can:

Summarize articles or selected text instantly.
Rewrite or proofread any content on the page.
Translate in one click, without leaving the tab.
Remember webpages you visit using a Page Memory system — a built-in knowledge layer for your browsing.
Let you choose Personas like “Analyst,” “Journalist,” or “Creative Writer” to shape the AI’s tone and style.

All of this runs through Chrome’s Prompt API and on-device Gemini Nano, ensuring speed and privacy.

How I built it

The project is built as a Manifest V3 Chrome Extension with:

A background service worker to handle AI prompts and API streaming.
A content script that injects the floating AI UI and captures page data.
A popup dashboard for direct interaction with Gemini Nano and saved memory.
An options page to toggle between demo and live API modes.
Chrome APIs like chrome.storage, chrome.contextMenus, and chrome.runtime to connect everything smoothly.

I used simulated outputs for Demo Mode (for offline judging), but the architecture supports real-time Gemini Nano inference via the built-in Chrome AI endpoints.

Challenges I ran into

Understanding and integrating Chrome’s new AI APIs (Prompt, Summarizer, Writer, etc.) was tricky — documentation was sparse.
Managing streaming AI responses inside a Manifest V3 environment took trial and error.
Implementing Page Memory persistence required creative use of chrome.storage.local and messaging between service workers and content scripts.
The hardest part was balancing user privacy with AI intelligence — so all demo data runs locally.

Accomplishments that I'm proud of

Built a fully functional AI Chrome Extension from scratch — complete with popup UI, options page, and background worker — all in Manifest V3.
Integrated Chrome’s built-in AI stack (Gemini Nano + Prompt API) to perform summarization, rewriting, and translation directly inside the browser.
Designed an offline Demo Mode
Created a unique Page Memory system that captures and recalls webpage insights — essentially giving Chrome a short-term memory.
Developed a streaming AI UI that simulates real-time token output, creating an interactive, ChatGPT-like experience right in the popup.
Designed a clean, responsive interface using only native web tech — no frameworks, no libraries, just pure JavaScript, HTML, and Chrome APIs.
Learned to manage complex background–popup–contentScript communication, which most extensions struggle with.
And most importantly — proved that AI doesn’t need to live in the cloud. It can live inside your browser.

What I learned

This project taught me how browser-based AI can work fully client-side, without external dependencies.
I learned about:

The architecture of Chrome’s on-device AI (Gemini Nano).
Advanced service worker communication in Manifest V3.
Designing intuitive, streaming UIs for in-browser AI experiences.
Most importantly, I learned how to make AI useful, not just impressive.