paperless-ai is an AI-powered assistant for Paperless-ngx that automates document classification, tagging, and natural language search. It connects directly to a Paperless-ngx instance, monitors for new or updated documents, and uses configurable AI models to assign titles, tags, types, and correspondents automatically. The tool also provides a real-time chat interface that lets users ask questions about any document and receive context-aware answers.
Beyond automated classification, paperless-ai offers several distinguishing capabilities. Every AI request, raw response, and applied metadata change is recorded in a local database, creating a full audit trail for review and compliance. Documents can be processed manually on demand through a web interface, and conversational document retrieval sends both the query and full document text to the language model in a single request, eliminating the need for a separate retrieval store. The system uses polling-based monitoring and exposes a REST API for integration, with template-based prompts that enforce consistent output formatting.
The assistant supports natural language document search that understands context beyond simple keywords, and allows users to select and customize different AI models and processing parameters to suit their specific needs.