LightScan
An AI-powered Chrome extension that helps college students navigate digital research material, summarize their findings, and reduce their research and study time.
Background
As a college student, you've probably encountered countless articles and websites that are filled with dense, lengthy content. While this information may be valuable, it can be overwhelming and time-consuming to sift through all the details to extract the main points.
College student’s also experience a phenomenon called “research anxiety” where the research process invokes feelings of stress, apprehension, and uncertainty.
How do we tackle the time and emotional roadblocks student’s experience during their research process ? LightScan, an AI-powered Chrome extension that assists college students in navigating lengthy digital research materials, minimizing the time they spend on research and studying.
Research
To validate our assumption that college students find it both time-consuming and emotionally taxing to extract and consolidate information from digital sources, we employed two research methods : user surveys and usability interviews.
We initially conducted a user survey of 17 students with varying majors and classifications. Our user base wanted to identify specific facts and figures, take notes, and summarize key points . Most of our users relied on skimming, the control F function, or reading the full source material to try to understand the concept. All of their processes had limitations and were time draining.75% of our users didn’t have a dedicated tool to optimize this process .
Landing on the Solution
Initially, we identified a market gap where existing AI systems, like Chat Gpt, lacked the ability to upload PDFs and interact with them in real-time, hindering college students' ability to extract information effectively.
Our plan was to develop a web application that allowed students to upload PDFs of up to 1000 pages. The user could then use an AI-powered chatbot to extract key points and summaries from the document.
We found that most of our users preferred websites and online eBooks instead of PDF documents.We also faced a setback when we lost one of our developers during our 5th sprint. Implementing this feature would have overburdened our remaining developer, jeopardizing our targeted ship date.
We decided to pivot and focus on a solution that served our users' needs in real time, while accommodating our new team structure.
Solution
LightScan is an AI-powered Chrome extension that helps college students navigate digital research material , reducing their research and study time.
We are launching LightScan with three key features
- Summarization & Elaboration - Students are able to analyze the web page/digital article they’re using for research, enabling them to identify, summarize, and gain a deeper understanding of the source material at an accelerated rate .
- Bookmark - Students are able to store and revisit all prompts and answers in chronological order .
- Edit, Revise, Copy - Students are able to edit and convert the generated response into their own words, avoiding plagiarism and AI detection, while gaining a greater understanding of the material .
UX & UI
After discussing my initial design exploration with our developer, I came to the realization that Chrome extensions are unable to adapt to the UI within the container, making the spacing and sizing of the elements appear misaligned.
As a result, I had to design the extension with a fixed container size, ensuring consistent screen dimensions across different visual elements within the extension container.
Implementation
Building LightScan posed several technical challenges. One of the biggest hurdles was working with vector databases, which required a deep understanding of database design and optimization.
To overcome this challenge, I used a serverless architecture for the APIs and hosted the extension on the Chrome Store. I also utilized Pinecone DB for vector storage and Langchain to sanitize & answer user questions and give the AI access to all the Pinecone vectors.
Although building LightScan was challenging, I successfully developed the app without encountering any scaling issues during the testing phase. However, finding a way to monetize the app is still an ongoing challenge.
I learned the importance of working with vector databases and the challenges that come with it. Going forward, I will continue to improve my skills in this area to ensure the success of future projects.
Feedback
After analyzing the survey synthesis and time constraints, we discovered that the majority of students prefer using online digital articles as opposed to research material in PDF form. We decided to temporarily shelve the upload process for future iterations.
We interviewed two potential users, masters students, who both have to read extensive digital source material for research. Both described analyzing source material as “tedious”. We implemented a test that measured:
- How long it took for them to read and summarize an article with and without the extension
- How long it took identify key concepts with and without the extension
- Did the extension help the student understand deeper concepts within the material
- Is the user happy with the information provided by the extension
It took the students about 3-5 minutes to read the provided article and an additional 1-2 minutes to summarize the content of the article without the extension. It took them 27 seconds - 1 minute for the same process with LightScan . Both users stated that the research process was easier when utilizing the extension.
Future Steps
LightScan is currently available for free in the Chrome Web Store. Our next step is to monetize the app using a freemium model. This means providing a basic version of the extension for free and charging for premium features.The decision on which premium features to offer will be based on user feedback and reviews from the app store. Tentative premium features include a web application for uploading PDF documents for users to siphon, and generating study materials based on content analysis and user input.
The extension currently uses an open AI key to function. After significant usage, the user is charged for this function. This will also play a role in our monetization.
Learnings
Product Manager Learnings:
Curtis Barber
- Effectively managing a cross-functional team across different time zones while adhering to agile practices (EST & IST) .
- Identifying and respecting the different work styles of my team members to establish trust.
- Building a functional product roadmap that garnered a functional MVP.
- Organizing the team’s tasks in a prioritized backlog with user stories using #Asana and the ICE feature prioritization method.
- Knowing that clearly articulating myself to my team is imperative for successful sprints and product launches.
Designer Learnings:
Kevin Wang
- It is essential to familiarize oneself with the platform for which I am designing before diving into the design process.
- In the face of uncertainties, it is advisable to begin the design process by identifying and leveraging the known constraints as a solid starting point.
Developer Learnings:
Pratik Ravariya
- Prioritizing user feedback and incorporating it into the development process.
- Effective time management and goal-setting as a solo developer.
- Continuous learning and keeping up-to-date with the latest technologies and industry trends.
Developers Learnings:
&
Full Team Learning
Our goal was to create a working MVP, which fostered open communication and a willingness to compromise to improve the product for the user. We faced challenges such as time zone differences, technology setbacks, product changes, and team transitions. In the midst of these challenges, we learned to pivot, push through, and apply our skills to successfully ship our product.