In brief: Datasette Agent combines the LLM framework with the Datasette database platform and enables natural language questions against SQLite databases.
The first version of Datasette Agent is now available – an AI assistant that provides a conversational interface for SQL-based database queries and can be extended through a plug-in system.
Datasette Agent provides a conversational interface that allows users to ask questions about data stored in Datasette. The system automatically generates SQL queries and executes them against SQLite databases. In the live demo at agent.datasette.io, the system runs on Gemini 3.1 Flash-Lite and uses sample databases such as the World Resources Institute’s Global Power Plants Dataset and a backup of a personal blog.
The system is designed as a plug-in architecture from the outset. To date, three plug-ins have been released: datasette-agent-charts extends the assistant with diagram generation via Observable Plot, datasette-agent-openai-imagegen adds a tool for image generation via ChatGPT Images 2.0, and datasette-agent-sprites enables code execution in a persistent sandbox. The system can also be run locally with open-weight models such as Gemma 4 26B.
For practitioners, this represents a new way to make data holdings accessible through natural language queries without having to write SQL manually. The system relies on reliable tool calls and correct SQLite generation – requirements that open-weight models from the past six months are increasingly meeting. Implementation for Datasette Cloud is planned.
Source: ainews-dev.lumi-systems.io · Published 21 May 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.5.2.