Mednova Assistant — RAG chatbot on n8n + Pinecone
A Retrieval-Augmented Generation chatbot that answers from company knowledge — n8n workflows, a Pinecone vector store, OpenAI embeddings, and context-aware responses in real time.
Answers from your own knowledge, not the model's
Clients needed a chatbot that answers from their own knowledge base with accurate, current context — not generic LLM output.
We designed and implemented a Retrieval-Augmented Generation chatbot that answers user questions from company knowledge data. The automation workflow runs on n8n, with Pinecone as the vector database storing and retrieving embeddings.
The system processes documents, generates embeddings with OpenAI, and retrieves relevant context for each answer. Automated ingestion pipelines (for example a Google Drive file trigger), query processing, and API endpoints let the chatbot respond with context in real time.
It shipped as the Mednova Assistant — an AI chatbot on the Mednova healthcare site answering questions about services, location and bookings such as telemedicine consultations, with optional logging to Google Sheets for leads and interests.
The assistant in place
Answering service, location and booking questions on the live healthcare site.
What we built
Grounded answers with retrieved context
Questions are answered from retrieved company knowledge via OpenAI, covering healthcare cases like services, bookings and location.
- —Retrieval-grounded answers
- —Healthcare use cases
- —Mednova Assistant
Automation workflows and a vector store
n8n orchestrates the pipeline, Pinecone stores embeddings, OpenAI handles embedding and generation, with optional Google Sheets logging.
- —n8n workflows
- —Pinecone embeddings
- —Google Sheets logging
Documents in, answers out
Automated ingestion from Google Drive, document loaders and text splitters, plus API endpoints for real-time responses.
- —Google Drive trigger
- —Loaders & text splitters
- —API endpoints
More screens
More cases
Have a similar project in mind?
The first call is a free 30-minute technical scoping session.
Book a call