#108
C+
Rank #226
DocsGPT
An open-source RAG chatbot that lets users query their own documentation or file uploads through an LLM-backed Q&A interface. Standout trait is the batteries-included ingestion pipeline for turning docs into a searchable vector index.
55.8
/ 100
⚙ Full Stack
Flask/FastAPI backend pairing a vector store (FAISS or Qdrant) with MongoDB for chat/session metadata, and a React frontend. LLM calls default to a hosted API but can be pointed at self-hosted or local models.
📈 Scaling Analysis
Reasonable for small-to-mid document sets and query volume since retrieval and generation are decoupled, but the default FAISS/local vector store doesn't horizontally scale — swapping to a clustered vector DB is required before internet-scale traffic. Session state in Mongo is a sound, if unremarkable, scaling foundation.
🚀 Running on Nexlayer
Needs at minimum an app pod, a Mongo pod (`mongo.pod:27017`), and a vector-store pod; all should use mirror.gcr.io base images. Set the LLM endpoint env var to a Nexlayer-hosted or local model to avoid forcing paid API costs on every deploy.