PandasAI is a Python library that makes data analysis conversational: connect a database, data lake or simple CSV and parquet files, then ask questions in natural language and get answers, tables and charts back. Instead of writing pandas or SQL by hand, the library uses a large language model with retrieval techniques to generate and execute the code for you. It has earned roughly 24,000 stars on GitHub and is one of the best-known natural-language interfaces to data.
PandasAI wraps your data sources in semantic dataframes that an LLM can query. Ask a question such as the average revenue by region, and it produces the answer; ask for a plot, and it returns a chart. Multiple dataframes can be queried together, so questions that span employees and salaries, or orders and customers, work naturally. It connects to SQL databases, CSV and parquet files and larger data lakes, works with many LLM providers through LiteLLM, and can run generated code inside an optional Docker sandbox for safer execution. The core library is open source, while the company behind it, Sinaptik AI, also offers a managed PandasAI Cloud and self-hosted enterprise editions for organizations that need governance and support.
The PandasAI library is free and open source under the MIT license (an enterprise directory carries its own license). Managed PandasAI Cloud and self-hosted enterprise offerings are paid, making the overall product freemium; you also supply your own LLM API key when self-hosting the library.
PandasAI is for analysts, data scientists and business users who want answers from data without writing SQL or pandas code for every question, and for developers embedding conversational analytics into their own products.