RLAMA

Point RLAMA at a folder of PDFs and notes, then ask questions in plain language and get answers drawn from those files.

RLAMA is an open-source tool for building document question-answering systems (RAG) with local AI models, run from the command line.

Processing stays 100% local, so no data goes to external servers.

It works with Ollama models and can also connect to OpenAI if you prefer a hosted model.

The CLI is free and handles several document formats.

It also has web crawling and directory watching, so a knowledge base can pick up new material from a website or a folder you point it at.

RLAMA also supports AI agents and crews for more complex tasks.

The catch is the command line: you need to be comfortable in a terminal.

A visual builder called RLAMA Unlimited offers no-code RAG creation, but it requires a subscription.

Expect some rough edges too, such as Ollama not being reachable, text extraction problems, and answers that miss relevant passages.