GLYNT’s goal is consistent and sensible field names that make “Document data is as easy as any other data source”
GLYNT Has a Single API But Serves Many Verticals
GLYNT is a machine learning system that can be applied to capture document data on any loosely structured document. The GLYNT user wants key-value pairs from the document data, eg Field Name, Data Value.
Based on a unique approach to machine learning known as “few shot” learning, GLYNT needs very few documents for training. When presented with an entirely new form or invoice, GLYNT needs just a handful of marked examples. First results are in minutes, and average 98% accuracy. When the documents are near replicas of each other, such as W-2 forms, GLYNT needs 3 – 4 documents for training. When the documents are varied, even within a type, GLYNT needs 10 – 12 documents for training.
Because of its extreme flexibility, GLYNT is easily trained for loosely structured documents across many verticals, including forms, invoices, bank statements and lab results. All are served by the same API. Please see the API documentation here.
Within the single API framework, the user is able to name the fields. So a single API integration can provide data for many different document types. This schema separates the API itself from the field naming conventions.
Some users will want to name fields in a certain way, but many others prefer to use GLYNT’s prebuilt Field Name Lists (aka Field List). With a known Field List in hand, the user can quickly integrate the GLYNT API and map the output to current databases.
Thus, GLYNT has a single API, but numerous prebuilt Field Lists and serves many verticals.