2

Video Document Chatbot - MyRandy

Full-stack Web Developer - Kytzo. Developed a full-stack Next.js application with TypeScript for a video document chatbot. Utilized LangChain to handle large quantities of texts (txt, pdf, word) and generate LLM models. Employed ChromaDB as a database to store embedded documents and MongoDB to store user data. Integrated GPT-4 to use the generated models and provide context to users concerning one or multiple text files. Implemented Stripe as the payment gateway. Created custom chatbots for clients, generating a new app automatically based on user specifications. Clients can access their personal chatbot, which accepts videos as input. Videos can be uploaded manually or via multiple platform integrations (Vimeo, YouTube, Wistia). A Python script retrieves the transcripts, stores them in ChromaDB, and uses LangChain and GPT-4 to create the customized chatbot with context from the user's videos. Deployed the apps on Vercel, handled production errors, and worked on maintenance.

Project Details

Detailed information about the project.

Project Goals
To create a customizable video document chatbot that leverages LangChain and GPT-4 to provide context from various text formats and video transcripts, ensuring seamless integration and user-friendly access.
Challenges Faced
Faced challenges in handling large quantities of diverse text formats, integrating multiple video platforms, and ensuring secure and efficient storage and retrieval of embedded documents and user data.
Solutions Implemented
Successfully implemented LangChain for text handling, integrated multiple video platforms for seamless video uploads, utilized ChromaDB and MongoDB for efficient data storage, and used GPT-4 for generating contextual chatbots. Overcame integration and data management challenges through detailed planning, thorough testing, and optimization of the application. Ensured secure payment processing with Stripe and deployed the application on Vercel for reliable performance and maintenance.
Technologies Used
Next.js, TypeScript, LangChain, ChromaDB, MongoDB, GPT-4, Python, Stripe, Vercel
Start Date
6/1/2023
End Date
10/31/2023

Project Screenshots

A visual overview of the project.