Ask AI

Inspiration

I'm a heavy user of both ChatGPT and Google Gemini, but there are two main issues I've encountered:

I often have to switch between tabs to search for and open a query, and when I have 30+ tabs open, it becomes difficult to find the right one. If I close a tab, the previous query and its result disappear, and I have to ask again. Sometimes the response is different, so I end up reading everything from the beginning to fully understand the solution the AI provides.

What it does

This extension will assist developers in quickly finding the precise information they need from our platform's collection of books. Instead of manually browsing through pages or searching, they can simply ask the AI model, which will provide relevant answers based on the content available.

How we built it

I’ll be using React.js to build the extension, with Node.js and Socket.io on the backend to handle real-time communication. There will be no login or registration required—just a straightforward socket-based system to receive real-time, chunked responses from the Google Gemini AI model.

Challenges we ran into

I want to save the entire chat history for each user, so that even if they close the extension, they can still revisit previous responses or ask the same question again at any time.

Accomplishments that we're proud of

I’ve successfully integrated real-time content search using the socket system, allowing users to receive responses in chunks while using the extension, rather than having to wait for the full response to be generated on the backend and then sent to the frontend.

What we learned

How does Google Gemini operate efficiently, and how can we leverage its powerful APIs across various applications?

What's next for Ask AI

I will store all user chats (queries and answers) so that users can easily refer back to previous conversations. This way, they can search their chat history for the same question/query, reducing the need for additional API calls.

Built With

• gemini
• google
• javascript
• node.js