AI-Powered Document Chat with Nextcloud Files using LangChain & OpenAI

This n8n workflow powers an intelligent chatbot for querying documents stored in Nextcloud folders. A public chat trigger captures user messages, feeding them into a LangChain AI agent powered by OpenAI. The agent maintains short-term memory for contextual conversations and dynamically calls a 'Nextcloud Tool' sub-workflow when file access is needed—simply by referencing the folder path in queries.

The 'Nextcloud Tool' sub-workflow is invoked with a folder path parameter. It lists all files via Nextcloud API, filters for supported formats (PDF, Markdown, DOCX), downloads them, and extracts plain text: PDFs use built-in extraction, Markdown is read raw, and DOCX leverages the community 'word2text' node. All texts are aggregated into a single output field, providing rich context back to the AI agent for precise, informed responses.

Benefits include seamless document Q&A without manual uploads, reduced research time, and scalable knowledge access for teams. Use cases: internal knowledge bases, legal/contract reviews, research assistance, or customer support on shared docs. High accuracy from RAG-like retrieval enhances AI reliability over generic models.