FocusedFind

Inspiration

FocusedFind was born out of the desire to streamline and enhance the search experience for users by providing more relevant suggestions based on keywords and context. The inspiration came from the frustrations of traditional search engines where results can be overwhelming and disconnected from what users actually intend to find. We wanted to build a tool that could more accurately predict a user’s search intent and assist them by providing tailored autofill suggestions, making their search process smoother and faster.

What it does

FocusedFind is an intelligent autofill suggestion tool that analyzes the context of a user's search query and presents them with highly relevant, context-aware search suggestions. By utilizing Chrome's Buildt-in Prompt API, it predicts what a user might be looking for based on a set of keywords extracted from the user's workflow that help describe their intent. The product aims to replace the default search suggestions with more relevant options.

How we built it

FocusedFind is a chrome extension built using a combination of modern web technologies and artificial intelligence tools. The core functionality relies on querying Chrome's Built-in Prompt API to generate relevant search suggestions based on the user's input and associated keywords extracted from the user's workflow from other browser tabs. These suggestions are then dynamically injected into the search bar.

Challenges we ran into

FocusedFind was initially designed to be a tool that filters search results based on the extracted keywords. However, due to most search results being from a specific context due to how the search engine functions along with disruptions by SEO, this approach was unsuccessful as certain searches caused no results to be displayed. To overcome this issue, FocusedFind pivoted into a search suggestion tool.

Accomplishments that we're proud of

Context-Aware Suggestions: We were able to create a system that can generate search suggestions that feel natural and tailored to the user's specific query by extracting the context of the user's workflow and basing suggestions off of it. Seamless Integration: The tool works seamlessly within the existing search bar interface of the browser, replacing the default suggestions with relevant ones in real-time. Real-Time Predictions: We successfully implemented a system where the suggestions are generated and displayed in real-time as the user types, providing immediate feedback and enhancing the search experience. Scalable Architecture: As the model and computation runs locally, the extension is designed to scale well, meaning that it can handle an increasing number of users without significant performance degradation. This allows us to expand the tool’s functionality in the future without major refactoring.

What we learned and What's next for FocusedFind

Reducing Time Taken to Generate Suggestions: While we've made progress, there’s still room for further optimization. We're exploring ways to reduce latency even more by considering alternative methods of storing and retrieving keyword data, as well as experimenting with generating predictions. We're also looking into caching suggestions to speed up the process for repeated queries.

Enhancing Keyword Extraction: As we continue to improve our system, one of the next steps is refining our keyword extraction algorithm to handle more complex and varied content types. We plan to implement machine learning techniques to better understand context and improve the precision of the extracted keywords, ensuring that we get the most relevant input for generating search suggestions.

Bug Fixes and Stability Improvements: As with any project, bug fixes and stability improvements are always ongoing. We’ll continue to address edge cases where the extension might not perform as expected. Improving error handling and ensuring the tool works smoothly across a wide variety of websites will be a priority moving forward.

Built With

• chrome
• javascript
• promptapi