Bookmark

Bookmark extension
Search your URL with natural language
Get your context automatically to generate embeddings

Inspiration

Back in 2021, during a one-on-one with my manager, I was excited to share an insightful article on blockchain that I’d bookmarked. The meeting was flowing, and I knew this link would help me back up my ideas. But when I started looking for it, it turned into a frustrating search. I went through category after category in my bookmarks, trying different keywords and scrolling endlessly. Half an hour passed, and I still couldn’t find it. By then, the call was over, and the moment—and my excitement—had passed.

This experience stuck with me, and I started thinking about how easily this could happen in crucial meetings with clients or partners. There are times when referencing the right content could make or break a conversation, and I realized I needed a way to access bookmarked links based on the context I remembered, not just keywords or URLs. This would allow me to focus on the conversation without worrying about searching for the link in the middle of an important moment.

Determined to find a solution, I dove into researching technology that could help, eventually learning about Generative AI and Retrieval-Augmented Generation (RAG). With these tools, I set out to create an application that could not only store bookmarks but also understand and retrieve them based on context. After building, testing, and refining, I finally launched it as a Chrome extension. Now, I can search my saved URLs simply by describing the topic or context I remember.

With this new tool, I’m able to instantly bring up the exact resources I need, even if the URL itself is a distant memory. This GenAI-powered bookmark extension has turned into an essential tool, helping me keep conversations seamless, informed, and efficient.

What it does

I can search your bookmarks with Natural human language with the context of the bookmark and you do not need to linearly search your bookmark on chrome

How we built it

I have built it using RAG where I am using llama embedding model which help to retreive the embeddings which I am keeping with MongoDB vector database.

Challenges we ran into

Finding the right embedding model , right dimensions and learning JS.

Accomplishments that we're proud of

I am using it everyday which makes me feel proud of building it.

What we learned

I have learnt the right use of GenAI , RAG and mongoDB robustness with search

What's next for Bookmark

I am working on its UI and also working on optimising the number of steps copying the URL and create its embedding in the vector database.