MLJAR Studio: Local AI Data Analyst with Reproducible Notebooks

MLJAR Studio is a new desktop application that combines natural language interaction with local Python execution to produce reproducible Jupyter notebooks. Built by the creator of mljar-supervised, an open-source AutoML library for tabular data, the app allows users to converse with their data in plain English. The AI generates Python code, runs it locally, and saves the entire conversation as an *.ipynb file, enabling inspection, modification, and re-execution. The app automatically sets up a local Python environment on Mac, Windows, and Linux, installs missing packages on the fly, and includes built-in AutoML for classification, regression, and multiclass tasks. It works with standard Python libraries (pandas, matplotlib) and supports various data file formats: CSV, Excel, Stata, Parquet. Database connectivity includes PostgreSQL, MySQL, SQL Server, Snowflake, Databricks, and Supabase. For AI models, users can run Ollama locally (zero data egress), bring their own OpenAI key, or use the MLJAR AI add-on. The tool addresses the common problem of AI-generated analysis being a black box by ensuring every output is a reproducible notebook.