probe.io

THE PROBLEM:
Businesses are sitting on a goldmine of data in the form of unstructured documents.  Getting value from these unstructured documents requires hiring talent in machine learning or artificial intelligence to design methods of extracting the required information.  Most companies don't have the ability to hire and retain such talent, and those that do are faced with the problem of creating an all purpose solution (one that is applicable for a variety of documents).

IDEA:
GPT-3 offers a generalized solution to extracting value from unstructured data with a surprisingly deep domain-specific understanding.  Our startup proposes to utilize GPT-3 and vector embeddings to allow businesses to focus on the questions they need answered from their documents, not the underlying technical requirements to achieve that goal. 

BUSINESS MODEL:
API, either subscription or pay as you go.

POTENTIAL CUSTOMERS:
The use case for our startup spans many industries, but insurance is one that we've decided to focus on initially due to the urgent need.  Insurance companies are currently outsourcing workers that can manually extract information from documents such as loss adjuster reports.  They are spending this money on outsourcing to gain key insights into past claims which then informs their risk pricing models when quoting/underwriting new clients or when assessing YoY increases to their existing clients premiums.  This one use case demonstrates that the questions they are answering with this outsourcing produce more value than the cost of the labor involved in getting those answers.