Relativity6 launched Marvin, a state-of-the-art LLM-powered Industry Classification product. This new, self-learning, self-training product achieves up to 95% industry classification accuracy across small business classes with only the business name and address as inputs.
“We’ve reached a level of accuracy and precision around industry classification that was never thought possible, even just 6 months ago. Marvin represents a real-world, high value application of purpose-built AI. We’ve harnessed the incredible capabilities of LLMs to unlock one of the most sought after but elusive opportunities in the data space, industry classification,” said Alan Ringvald, CEO of Relativity6.
The 33.2M small businesses in the US are the great unknown in the data world. 99% of those small businesses have less than 20 employees. Those small businesses are in a constant state of change, making them incredibly challenging to classify. Misclassification accounts for tens of billions of dollars of underwriting and compliance losses each year.
“Before Marvin, the best models out there hovered around 60% accuracy, but we’re achieving significantly higher recall and precision across small businesses classes. Our system is self-learning and automatically retrieves its own updated training data, so we’re confident near 100% accuracy is technically achievable,” said Abraham Rodriguez, CTO of Relativity6.
SOURCE: PRNewswire