May 2023

In the last month, we released a new API for classifying a document by its user-defined type, improvements to the Zapier-Sensible integration, to the quick extraction tab in the Sensible app, and to natural-languages methods.

New feature: Classify API

You can now score a document's similarity to the document types and reference documents you defined in your Sensible account.

Use this API:

  • In an extraction workflow. For example, determine which documents to extract prior to calling a Sensible extraction endpoint.
  • Outside an extraction workflow. For example, to determine where to route each document or to label each document in a system of record

For more information, see Classify a document by its document type.

Improvement: Zapier integration

Previously, you could only specify single-value fields as spreadsheet output in Zapier. Now, you can configure the Sensible Trigger to output tables, sections, and other complex output as spreadsheet output using Sensible's JSON-to-spreadsheet conversion rules.

To use this option, specify the Create Excel output in the Sensible trigger. See slide 3 in the following slides.

For a full tutorial, see Advanced Zapier tutorial.

Improvement: Removed limits on file size for quick extraction

You can now upload large documents to the quick extraction tab in the Sensible app The following limits are now removed:

  • there's no limit for a document's file size
  • the 30-second time-out limit is extended to 15 minutes in the background. If the quick extraction tab reaches its timeout of 5 minutes, check the recent extractions tab after 15 minutes to see if the extraction completed in the background.

Improvement: Include page location in natural-language prompts

You can now add page location information to disambiguate the data you want to extract in the Query and List methods. Use location information if there are multiple data candidates for extraction, or if you need to otherwise narrow down your search. For example, for the Query method, you can ask questions such as:

  • "address in the top left of the first page of the document"

  • "What is the medical paid value on the last claim of the second page?"

  • "consumer electronics device with highest sales mentioned near end of document"

Bug fix: Removed column recognition from multi-line cell recognition in Text Table

In December 2022 Sensible introduced an improvement to take a line's membership in a column into account in order to more accurately recognize multi-line cells in Text Tables. This month, Sensible removed this improvement because it introduced bugs.