A web app designed to empower your documents to speak volumes! With AskkkDoc, you can seamlessly process a variety of file formats, including PDFs, Word documents, and images, to extract insightful information and answer your queries.
Key Features:
- Frontend & Backend: Built with NextJS 14 App Router (React), TypeScript & and React Markdown.
- Styling: TailwindCSS and Shadcn UI.
- File Handling: Utilizes React Dropzone.
- Authentication & Database: Next Auth, Prisma ORM, and MySQL (TiDB).
- Storage & AI: AWS S3, OpenAI API, Pinecone, and Vercel's AI.
- File Processing: Langchain and Unstructured.
Core Functionality: Inspired by the functionalities of ChatPDF, AskkkDoc goes beyond just PDFs, allowing users to upload various document types – including images and Word documents. The AI will attempt to answer questions based on these documents, but please note that the answers are not guaranteed to be 100% correct, so it's advisable to verify them.
Limitations:
- Due to OpenAI API constraints, each user can upload up to 3 documents and ask questions. Unfortunately, due to some restrictions, I could not integrate Stripe for payments to let users buy tokens.
- After completing the whole website, I came to know that Unstructured is actually slow in processing file, which can have negatively affect the overall performance of a website in modern times, where most of the websites are blazingly fast. While the alternatives are cost real money, I chose to stay with this package since I did not want to make the most out of it.