Docugami, a Business Document Foundation Model for Generative AI 

Docugami is a proprietary Business Document Foundation Model, a Large Language Model (LLM) for Generative AI applied to your own business documents, where the human-in-the-loop is the Business User. Docugami is cross-segment and in the market today in a variety of vertical industry segments, including Commercial Insurance, Commercial Real Estate, Technology, a wide range of Professional Services sectors, and more. 

Docugami is more specifically a family of LLMs trained on millions of Business Documents, ranging from 2.7B parameter to 20B parameter models, with multi-modal inputs in the vision and text domains. Because Docugami is designed and trained exclusively for business scenarios, our LLMs are a better choice for business documents from a cost and performance perspective than 100 billion+ parameters models, while achieving parity in accuracy. 

Docugami's founders recognized several years ago that the time was right to create an AI transformation to change the world of business documents. 

Since its very inception, Docugami’s science and innovation have been standing on the shoulders of giants, including the seminal Attention Is All You Need (Vaswani, et al) paper, first among further research leading to the advent of Generative Pretrained Transformers (GPT), and Declarative Markup, that enabled the semantic representation of documents as data at scale. 

Unlike the present publicized uses of Generative AI, such as writing a customer service response letter in the style of William Faulkner, drawing pictures of a cat in the style of Salvador Dali, or passing an Advanced Placement biology exam, Docugami takes a very different direction and focuses on generating Business Documents and Data. 

Starting with your documents in any file format, our Foundation Model generates a Document XML Knowledge Graph Representation of your Business Documents in their entirety, which Docugami uses to generate Spreadsheets, Business Reports, Abstracts, new Business Documents through interactive Assisted Authoring, or Data ready for use by Enterprise Systems.  

The different and patented direction on Generative AI taken by Docugami means that the content generated is derived exclusively from your own business documents – not some random content on the Internet, minimizing any potential for erroneous or inappropriate content generation, including the type of AI hallucinations that have captured headlines. 

Docugami is founded on the principles of understanding the unique “Small Data” of individual companies and providing immediate value to frontline business users without costly investments in IT projects or training. The business user, as the human-in-the-loop, points to her business documents and provides very minimal feedback (“few shot”) to the Docugami Foundation Model that, in turn, creates a new fine-tuned Foundation Model with the user’s data (the “Small Data”). Each customer’s data and feedback are completely separate and secure. 

Docugami is both a Foundation Model and an Application for Business Users in the market today, and our API is currently available to select customers on the Docugami Playground.