WrenAI local setup and PostgreSQL dashboard tutorial

This Italian-language tutorial introduces WrenAI as an open source tool for generative business intelligence. The speaker positions it alongside Power BI and Tableau, then demonstrates how natural-language questions produce SQL queries, tables and charts that users can pin to a dashboard.

Setup starts with Docker Desktop and the WrenAI launcher. The walkthrough describes launcher downloads for macOS, Linux and Windows, and uses Windows for the demonstration. After model configuration, the launcher downloads container images and starts the interface at localhost:3000. The speaker selects an OpenAI API key for this session and explains that an Ollama connection instead needs a host and port. The demonstrated setup therefore does not establish fully offline operation or the speaker's claim of complete privacy.

The first example uses a sample human-resources database. The tutorial covers semantic relationships, editable data models and a knowledge section that stores question-and-SQL pairs. It then connects PostgreSQL data representing a simulated shop and asks for monthly sales, best-selling products and revenue by geography and product category.

For users planning to run models locally, the speaker warns that results depend on model capability, including function calling. He recommends large models and cites GPT OSS with 120 billion parameters, while acknowledging the resources needed to run it locally.