Sprout.ai has launched a new machine learning-based claims automation product that can “review and assess claims without ever having seen a claim of that type before”.
The new ‘zero-shot model’ solution is currently in use by existing Sprout.ai customers, and the company reports that it offers “a 50% increase in claim handler efficiency, with a 90% reduction in training data required”.
Zero-shot learning is a method of machine learning in which an artificial intelligence (AI) model can classify a previously unencountered object or entity by connecting it to ones it has encountered previously. For example, if an AI has never seen a zebra before, but understands the concepts of ‘horse’ and ‘stripes’, it can identify one on the basis that ‘zebras look like horses with stripes’.
Likewise, in the context of processing insurance claims, Sprout.ai’s solution is able to classify unusual claims by noting their similarity to others that it has encountered – minimising manual input and increasing efficiency.
The solution includes:
- A built-in quality-checking function to help customers validate and improve how quickly their claim can be processed, prior to submission
- An extraction engine that uses computer vision and natural language processing (NLP) to extract and classify data from both structured and unstructured documents
- A validation engine that also uses NLP, and works with the extraction engine to automate checking on all extracted data, ensuring accuracy.
The launch of the new product follows the success of a recent £5.4 million funding round led by Amadeus Capital Partners and Praetura Ventures.