cherry

InspirationI’ve often found myself losing valuable information while browsing, that one perfect StackOverflow answer, a recruiter’s profile, or a useful article buried in my tabs. Bookmarks and browser history just weren’t enough.

I wanted a tool that could remember what I read, summarize it, and help me recall it later, all while keeping my privacy intact. That’s what inspired me to create Cherry, an AI memory companion for Chrome that acts as a second brain for the web.

What it does Cherry is a lightweight, privacy-first AI Chrome extension that remembers everything you read online.

It automatically summarizes, organizes, and lets you ask questions about your browsing history using natural language.

How we built it Cherry was built using a modern, privacy-centered tech stack:

Chrome Extension APIs, to access page content, tabs, and browsing history Gemini Nano (Summarizer + Writer APIs), for on-device AI summarization and rewriting IndexedDB, for storing and managing local semantic memory React + TailwindCSS, for a clean, minimal interface in the popup and chat overlay Local Embedding Search, to match user questions with stored summaries semantically

Challenges we ran into Challenges We Ran Into

Integrating Gemini Nano within Chrome’s resource limits Designing a semantic memory model that supports natural recall Maintaining fast performance while processing and summarizing pages Creating a smooth, chat-style overlay that feels natural in Chrome Ensuring data privacy while enabling meaningful AI interactions

Accomplishments that we're proud of Accomplishments That We're Proud Of

Built a fully functional AI memory extension for Chrome Achieved natural language recall of browsing history Designed an elegant and minimal UI that blends into the browser Ensured 100% data privacy, all AI runs locally Created a genuinely useful tool that makes browsing more intelligent

What we learned What We Learned

How to integrate local AI models like Gemini Nano within browser environments How to design semantic memory systems using embeddings and local databases The power of privacy-first AI design How thoughtful UI/UX decisions improve AI interactions The value of combining AI, memory, and usability into one seamless experience

What's next for cherry Smarter memory engine with vector embeddings for deeper recall

Optional encrypted multi-device sync Smart context actions (draft notes, emails, follow-ups) Gemini 2 integration for reasoning and proactive assistance Open-sourcing Cherry for the developer community. Cherry’s mission is to become the AI memory layer for the web,helping people remember what truly matters while keeping their privacy protected

Built With

• chrome
• css
• extension
• indexeddb
• indexeddb/chrome
• node.js
• plasmo
• pnpm
• react-18
• rewriter
• rowser
• typescript
• via
• writer