Hercules AI Launches ‘Assembly Line’ Approach to GenAI

Combined with its RosettaStoneLLM, the approach speeds GenAI development in regulated industries.

(Image source: Hercules AI website.)

Hercules AI (Campbell, Calif.), initially known as Zero Systems, has introduced an assembly line approach to building generative AI applications, paired with the large language model “RosettaStoneLLM” to target highly regulated sectors such as finance, insurance, and legal services. This approach seeks to transform AI application development, focusing on automating complex workflows that require human-like cognitive decision-making.

The new offering addresses a gap between the excitement about Generative AI and a rate iof adoption in enterprises has been slower than anticipated. “Many AI initiatives stall at the prototype phase,” a Hercules statement says. “Enterprises face a tough choice: invest heavily in building their AI systems or risk exposing sensitive data using external models. This issue is especially acute in highly regulated sectors where data security and compliance are critical.”

Alex Babin, CEO, Hercules.

Many enterprises face the challenge of transforming large volumes of structured data, such as database exports and spreadsheets, to align with their internal workflows, the vendor says. Converting millions of CSV files and Excel spreadsheets daily, this traditionally manual and costly process results in significant yearly expenses for large organizations, particularly in sectors such as insurance, healthcare, and finance.

As part of its “assembly line” engine for AI Solutions, Hercules says it is is releasing the first enterprise-grade LLM of 7 billion parameters that “speaks” with spreadsheets, called “RosettaStoneLLM.” It has been trained on a vast amount of data to perform tasks like mapping, transforming, and verifying structured data.

“Over the past eight months, in collaboration with our clients and partners, large consulting and insurance companies, our team has been focusing on merging top-tier structured data transformation expertise with the capabilities of our Hercules engine,” comments Alex Babin, CEO, Hercules.

Outperforming GPT-4

Hercules asserts that results on enterprise datasets demonstrate that “RosettaStoneLLM” outperforms general models like GPT-4 by 10 to 30 percent in tasks like entity mapping and code generation for data transformation, both in processing speed and output quality.

Hercules’ assembly line method, combined with a set of vertically oriented models, addresses the challenge of enterprise-grade, scalable, and reliable AI solutions within the enterprises’ existing security perimeters. The company reports that this approach has led to a five-fold annual growth and partnerships with numerous Fortune 500, insurance, and financial services companies and some of the top law firms in the U.S.

2024 AI Trends Reshaping Insurance Operations

Anthony R. O’Donnell // Anthony O'Donnell is Executive Editor of Insurance Innovation Reporter. For nearly two decades, he has been an observer and commentator on the use of information technology in the insurance industry, following industry trends and writing about the use of IT across all sectors of the insurance industry. He can be reached at AnthODonnell@IIReporter.com or (503) 936-2803.

Leave a Comment

(required)