Hercules AI Launches Innovative Assembly Line Method for Developing Enterprise-Grade Generative AI Applications

Hercules AI, a generative AI company focused on shaping the future of work, has developed an efficient methodology for deploying virtual AI workers within enterprises. Previously known as Zero Systems, the company has also launched RosettaStoneLLM, a model designed to help businesses in regulated industries automate complex workflows that involve cognitive decision-making.

This innovative “assembly line” approach allows organizations to select prefabricated components to create and deploy virtual AI workers efficiently. All components are pre-made, tested, and configured, eliminating the need for custom orders. Hercules AI asserts that this method results in high-quality, cost-effective, and easily scalable AI agents. Organizations only need to fine-tune the model to ensure it adheres to their specific workflows.

While many companies may prefer customized bots tailored to their specific needs, achieving this in regulated sectors can demand significant time and financial resources, potentially inconveniencing customers. However, industries like finance, insurance, and legal services may be more receptive, as all components are vetted by regulating bodies to ensure compliance and data security. If this model works for others in regulated environments, why not adopt it? The sole customization allowed would be with the large language model (LLM).

Additionally, Hercules AI offers RosettaStoneLLM, built on Mistral-7B and WizardCoder-13B, featuring 7 billion parameters. This LLM enables businesses to convert structured data from various spreadsheets into formats that the AI can utilize. For regulated firms with extensive databases and spreadsheets, converting these files into actionable system-wide data can be costly and labor-intensive. RosettaStoneLLM is specifically designed to transform large volumes of structured data to align with internal workflows seamlessly.

Early results suggest that RosettaStoneLLM can outperform general GPT-4 models in tasks like entity mapping and code generation by up to 30%.

As a Hercules AI spokesperson explained: “In sectors like insurance and finance, structured data is vital for operations. For instance, when an insurance firm prepares a healthcare insurance quote, the client submits HR database information that often varies in format and structure, along with unique naming conventions. Traditionally, converting this data for internal analysis is time-consuming. RosettaStone can accomplish this transformation in seconds, a task that would typically take hours for humans to complete.”

Based in Campbell, California, Hercules AI has raised $12.1 million in venture funding. While specific customer numbers remain undisclosed, the company has highlighted its use among Fortune 1000 companies and 30% of the top law firms in the U.S.