March 2026

In the last month, Sensible added support for extracting data from multi-document email attachments ("portfolios") and for testing email extractions in a development environment. We also updated Anthropic LLM model versions and made minor UX improvements.

Improvement: Portfolio attachments for email extraction

Sensible now supports data extraction for portfolio email attachments, or attachments that contain multiple documents combined in a single file. When you specify portfolio mode for your email processor, Sensible automatically segments each document in a multi-document file attachment by page range, and classifies each one against the document types you configure for the processor. For more information, see Getting started with email extraction.

Improvement: Environment targeting for email extraction

You can now direct email extractions to a specific environment by prepending the environment name to your email processor address. For example, to use your development environment's document type configs and webhook, forward emails to [email protected]. If you omit the environment prefix, Sensible defaults to the production environment. For more information about configuring your email processor for environments, contact Sensible.

Improvement: LLM model version updates

Where applicable, Sensible updated the Query Group and NLP Table methods to Claude Haiku 4.5 from Claude 3 Haiku. For more information, see LLM models.

Improvement: Filter portfolio extractions by configuration

In the Extraction history tab, you can now filter portfolio extractions by configuration names:

Previously, this filter returned results only for single-document extractions. You can use the same filtering option in the List extractions endpoint by passing a comma-delimited list of configuration UUIDs to the configuration_id parameter.