LLM-based methods

Extract free text from unstructured documents using large language model (LLM)-based SenseML methods. For example, extract information from legal paragraphs in contracts and leases, or results from research papers.

These methods are low-code alternatives to layout-based methods for structured documents, for example, tax documents or insurance forms.

The following topics describe how to author LLM-based methods using the SenseML editor. For information about authoring LLM-based methods using a visual tool instead of JSON, see Prompt tips.

MethodExample use caseNotes
List method"For each vehicle in an auto insurance declaration, extract the VIN, model, and year."Extracts a list of data out of a document, where you don't know how the data are represented.
NLP Table method"For each transaction in a bank statement table, extract the date and amount."Extracts a list of data out of a document, where you know they're in a table.
Query Group method"When does the policy period end?"
"What are the last 4 numbers of the account?"
Extracts a single fact or data point.
Summarizer computed field method + Topic method"list the rents, how often the rent must be paid, and when the rent is due"More configurable alternative to the List method.

Notes

For layout-based extraction methods, see methods.