Paper AI Tools Competition: Drive Data Innovation
Tool title
Paper AI
What is your tool?
Just a one-liner. You'll give a more detailed description below.
The ultimate AI-Powered & Open-Source research companion that streamlines literature reviews, enhances real-time collaboration, and offers a smart voice assistant to revolutionize your research experience.
AI-Powered & Open Source Collaborative Research Platform
Proposal abstract:
Describe your proposed tool. We want to understand how it works, how it reaches the learner populations and learning outcomes relevant to your track, and how the technology functions.
Paper AI is an innovative web application that transforms academic research by integrating artificial intelligence, collaborative tools, and an AI voice assistant similar to Apple's Siri. Users can upload research papers and interact with them by annotating, highlighting, and adding personal notes. When text is highlighted, the integrated AI—powered by OpenAI's latest o1 language models and the AI/ML API—instantly provides explanations and summaries, enhancing comprehension of complex concepts.
A standout feature is the AI Voice Assistant, powered by ElevenLabs, which allows users to interact with the platform using voice commands. For example, a user can click the assistant button and say, "Hey, any updates from my co-researcher Wei Tang?" This feature streamlines communication and keeps researchers informed in real-time, making the research process more efficient and engaging.
The tool reaches learners—including researchers, students, and academics—by being globally accessible online with multilingual support. It addresses challenges like information overload and time-consuming literature reviews, improving learning outcomes such as increased comprehension and reduced research time. By fostering a shared learning ecosystem, it encourages collaboration across disciplines and geographic boundaries.
Technologically, Paper AI is built with TypeScript and Tailwind CSS, and deployed on Vercel for optimal performance. It utilizes the open-source Adobe PDF Embed API as the core of the platform, allowing seamless interaction with PDF documents. The AI functionalities are powered by OpenAI's latest o1 language models and the AI/ML API, providing advanced natural language processing capabilities.
The AI Voice Assistant leverages ElevenLabs' technology for realistic and contextually aware speech synthesis across multiple languages. For data storage and synchronization, the platform uses Google's Cloud Services—including Firebase Auth, Firebase Realtime Database, and Storage—ensuring scalability and reliability. User authentication and management are handled by Clerk, offering comprehensive embeddable UIs and APIs for a seamless user experience.
To enhance performance on the client side, Paper AI employs IndexedDB for storing audio results from the AI Voice Assistant, ensuring that user data is not collected and maintaining data security. This allows for high-performance searches and efficient data retrieval without compromising user privacy.
What is your approach to Learning Engineering?
Our approach to Learning Engineering leverages anonymized user interaction data to enhance learning outcomes and drive continuous improvement. We collect data on highlights, annotations, requests for AI-generated explanations or summaries, voice queries to the AI Voice Assistant, and collaborative activities like sharing notes. This provides valuable insights into areas where users need more understanding or face challenges. By analyzing these patterns with machine learning algorithms, we improve the accuracy of AI responses and personalize learning experiences tailored to individual behaviors. Researchers can utilize aggregated data to identify common learning challenges, analyze the effectiveness of different interaction methods, and measure the impact of collaboration on learning outcomes. Data privacy is ensured through anonymization, user-controlled data sharing preferences, and compliance with international regulations.
How can your tool scale?
Please summarize how your tool can grow rapidly in new contexts, markets, and/or populations. If there are any significant cost considerations required to scale your tool (e.g., hardware) please note them in this section.
Built on scalable cloud infrastructure like Google's Cloud Services and Vercel, it efficiently handles increasing user loads. The platform's multilingual support and AI Voice Assistant—powered by ElevenLabs—enable expansion into non-English-speaking markets, reaching diverse populations. Utilizing scalable AI services like OpenAI's latest language models and the AI/ML API ensures our AI capabilities adjust to demand, facilitating efficient scaling. Integration with Clerk for user authentication allows seamless onboarding of large user bases. Cost considerations are minimized through pay-as-you-go pricing models offered by our technology providers, and efficient resource management—such as using IndexedDB for client-side storage—reduces server load and data transfer costs. Collaborative features promote organic growth as users share annotations and attract others, while adoption by educational institutions and research organizations further accelerates expansion.
Why is this the right team?
Please highlight content or technical expertise related to your team.
Our team is uniquely qualified to execute this project. Co-founder, CEO, and CTO, Ibrohim Abdivokhidov is a senior Computer Science student with extensive experience in AI, Neuroscience, and Quantum Computing. He has authored over five research papers, participated in 60+ hackathons, and collaborated with 100+ industry leaders. Co-founder, CLO, and COO, Wei Bing Tan holds an LLB (Hons) from the University of London with top accolades in Criminal and Intellectual Property Law. She has strong tech skills from coding bootcamps and AI hackathons. Together, we blend technical, legal, and operational expertise, making us the right team for Paper AI.
Extra prize (Dataset):
Our dataset focuses on medical imaging diagnostics in the field of neurology, specifically Brain MRI scans. It is designed to enhance AI models for improving diagnostic accuracy and accessibility in underserved rural areas.
(a) Topic Area: The dataset pertains to medical education and healthcare diagnostics, with an emphasis on brain imaging and neurological conditions.
(b) Relevant Research Questions: The data helps address key research questions such as:
- How can AI improve the accuracy of Brain MRI diagnostics from 84% to 95%?
- What patterns in MRI scans are indicative of specific neurological disorders?
- How can AI reduce diagnostic waiting times from days to minutes?
- How can AI assist in making advanced diagnostics accessible in regions lacking medical specialists?
(c) Target Populations and Representation: The target population includes patients undergoing Brain MRI scans, particularly from underserved and rural areas where access to advanced diagnostics and radiologists is limited. The data represents a diverse patient demographic, capturing a wide range of neurological conditions and anomalies, thus providing a comprehensive foundation for training AI models to generalize across populations.
(d) Potential Data Size: The dataset comprises over 3 million medical records, including detailed MRI observations and conclusions. This extensive dataset allows for robust training of AI models, enhancing their accuracy and reliability in real-world clinical settings.
In summary, this dataset is instrumental in developing AI solutions that improve diagnostic accuracy, reduce waiting times, and make advanced Brain MRI diagnostics more accessible, thereby addressing critical gaps in global healthcare.