The company claims that early results show RosettaStoneLLM can perform tasks like entity mapping and code generation up to 30% better than general models of GPT-4. The model is designed to convert large volumes of structured data into a format that aligns with a company's internal workflows. Hercules AI, based in Campbell, California, has raised $12.1 million in venture funding and is used by Fortune 1000 companies and 30% of the top law firms in the U.S.
Key takeaways:
- Hercules AI, previously known as Zero Systems, has developed a new methodology for quickly deploying virtual AI workers in the enterprise, using an 'assembly line' approach.
- The company has introduced RosettaStoneLLM, a model designed to help companies automate complex workflows that require cognitive decision-making.
- The RosettaStoneLLM is built off Mistral-7B and WizardCoder-13B and can convert structured data from within spreadsheets so it can be mapped and transformed by the AI.
- Hercules AI has raised $12.1 million in venture funding and is used by Fortune 1000 companies and 30 percent of the top law firms in the U.S.