Sensible Instruct

What is Sensible Instruct?

Sensible Instruct is a document extraction tool built with an API-first approach, designed to be integrated into other systems. It pulls structured data from documents such as resumes, contracts, and invoices using GPT-4’s natural language processing, and it supports SenseML, a configuration language for customizing how extraction works.

Key capabilities include:

• Real-time processing of incoming documents
• Targeted extraction by defining specific regions within documents
• Data redaction and transformation to mask or modify information
• Versioning and environment testing for managing and validating setups
• Custom data regions and granular access control
• Confidence signals and validation checks to gauge extraction reliability
• Support for emails, spreadsheets, and images
• Zapier integration, a dedicated dashboard, and high-volume extraction capacity

The platform provides enterprise-grade security and regulatory compliance. Its interface is accessible to non-developers, and it delivers results without requiring you to train models yourself.

Who created Sensible Instruct?

Sensible Instruct was founded by Josh Lewis, who also serves as CEO. He started Sensible after experiencing firsthand how challenging document parsing can be for developers. The company’s goal is to simplify document processing by combining advanced parsing techniques with comprehensive APIs that fit into existing tech stacks.

What is Sensible Instruct used for?

Typical use cases span several industries:

• Building permit processing — extracting key details from permit applications
• Medical research — gathering and organizing data from research documents
• Insurance claims — extracting information from claim forms
• Financial services — simplifying extraction from financial documents and reports

Getting started

Development happens through the SensibleSDK: import it into your code, create an instance with your API key, and call the extract() function with a document URL and document type (such as a resume or invoice). An environment parameter distinguishes testing from production, and the wait_for() function returns results once processing completes.