datasette-extract 0.3a0

2026-04-01 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, quick

Summary

Datasette-extract 0.3a0, released on April 1st, 2026, introduces new capabilities for importing unstructured data, including text and images, into structured tables. This update integrates with datasette-llm, allowing users to manage and configure the large language models (LLMs) available for extraction tasks. Specifically, model availability for extraction can now be controlled via the `extract` purpose within datasette-llm's model configuration. This enhancement streamlines the process of defining which LLMs are utilized for converting diverse unstructured inputs into a structured database format, improving flexibility and control for data professionals.

Key takeaway

For data engineers and analysts working with Datasette, this update means you can now precisely control the LLMs used for extracting structured data from unstructured sources. Configure your preferred models through datasette-llm's `extract` purpose to optimize performance and cost for your specific text and image processing needs. This integration simplifies managing your data pipeline's AI components.

Key insights

Datasette-extract 0.3a0 integrates with datasette-llm for configurable unstructured data extraction.

Principles

• Unstructured data can be structured programmatically.
• LLM configurations should be manageable.

Method

Datasette-extract uses datasette-llm to define and manage LLM configurations for data extraction, specifying models via the `extract` purpose in LLM model settings.

In practice

• Configure LLMs for extraction using datasette-llm.
• Import text and images into structured tables.

Topics

• datasette-extract
• Unstructured Data Extraction
• Structured Tables
• datasette-llm
• Model Configuration

Code references

• datasette/datasette-extract
• datasette/datasette-llm

Best for: AI Engineer, NLP Engineer, Computer Vision Engineer, Data Scientist, Data Engineer, Machine Learning Engineer

Related on AIssential

• datasette-enrichments-llm 0.2a0 — datasette-enrichments-llm 0.2a0 integrates datasette-llm for enhanced LLM configuration and management.
• Automate PDF Data Extraction with n8n EASILY! (Open source) — Combine Unstruct and n8n to automate local, free, and open-source structured data extraction from documents.
• datasette 1.0a31 — Datasette 1.0a31 expands data interaction by enabling SQL write queries and persistent query storage.
• Ai_extract in Databricks — Databricks' `AI_extract` function, combined with regex and SQL, efficiently extracts structured data from PDF-derived HTML tables.
• GliNER2: Extracting Structured Information from Text — GliNER2 unifies multiple NLP extraction tasks into a single, CPU-efficient, schema-driven framework.
• LlamaSheets | AI Parsing and Extraction for Spreadsheets — Llama Sheets converts unstructured spreadsheet data into structured Parquet files for AI agent processing.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.