SurfSense setup: a local NotebookLM alternative

In this Persian tutorial, the presenter introduces SurfSense as a NotebookLM alternative for document research and questions about uploaded sources. He describes desktop installers for Windows, macOS and Linux, plus open source code that developers can modify or deploy on a server. A suggested self-hosted use case is a company knowledge base where staff ask questions about PDF instructions.

The walkthrough covers the Windows installer, Google sign-in and model settings. The presenter explains how to add an Ollama model and discusses hosted model providers. Local LLM support is part of the tutorial, but the document chat example uses a hosted model. The example does not demonstrate a fully offline research pipeline.

Sources can come from web links, individual files or folders. He describes local document embedding and says a 200 to 300 page document can take several minutes, depending on storage, RAM and processor performance. He then shows a Persian book summary and explains how numbered citations open the source chunks behind an answer.

The overview also discusses folders, scheduled tasks, connectors and conversation memory. A network error interrupts one query, and the presenter attributes it to his VPN. The tutorial does not demonstrate a complete offline workflow.