Sensible Instruct

The Multimodal Engine uses LLMs to extract data from non-text and partial text images embedded in documents, such as pictures, charts, graphs, and handwriting. It also improves data extraction accuracy from documents with challenging layouts, like overlapping lines, non-standard checkboxes, and signatures.

The Multimodal Engine parameter sends an image of the document region containing the target data to a multimodal LLM. You can ask questions about non-text and partial text images, and Sensible automatically selects a relevant excerpt and surrounding context to send as an image to the multimodal LLM based on your natural language queries.

Examples include extracting data from images embedded in a document, such as facts about a building's characteristics or community amenities, and extracting data from documents with complex and imprecise layouts, such as handwritten forms with overlapping lines and checkboxes.

To enable the Multimodal Engine parameter, you need to configure it within the LLM Query Group method. You can set the region to 'automatic' or define a custom extraction region to accurately extract the data.

Sensible supports various file types, including PDF, TIFF, JPEG, PNG, Word documents, and Excel files.

Sensible sends an image of the relevant region to the LLM, which uses context to holistically process and extract data similarly to the way a human would, improving extraction accuracy for challenging or complex layouts.

Yes, the Multimodal Engine can be used to directly extract data from visual charts in a financial report, making it a powerful tool for automating data extraction from a wider range of documents.