Docvia - AI Chat bot integration (SaaS)
About Project
This project is an AI-driven knowledge management platform that allows users to transform static Google Docs into an interactive chatbot. The core functionality centers on a Retrieval-Augmented Generation (RAG) system which acts as a dynamic brain for the AI. Users can sync their documentation in a read-only format, and the system processes this information to provide precise, context-aware answers to user queries. To make this technology accessible to other developers, I architected and published a production-ready React npm package. This allows for seamless integration into external Next.js and React applications with a single import statement. To support the business logic of the platform, I developed a sophisticated SaaS billing engine using Stripe that automates the entire lifecycle of a subscription, including upgrades, downgrades, and failed payment handling, while utilizing feature-flagging to enforce plan limits such as restricting free users to a single active application. The most significant technical challenge I encountered during development was establishing a secure and user-friendly connection between the platform and a user's Google Docs library. My initial architectural approach required users to navigate the Google Cloud Console to generate and upload a service account credential JSON file. However, I quickly realized that this was a major friction point because the process is highly technical and intimidating for non-developers, which would have severely limited the platform's adoption. To solve this, I pivoted to implementing a Google OAuth2 integration. This allowed me to replace a complex manual setup with a familiar and secure "Sign in with Google" flow. By requesting specific scopes, the platform can now gain authorized read-only access to a user’s documents with a single click, providing a much smoother onboarding experience while maintaining high security standards.