Truera, a startup developing a machine learning model intelligence platform, today announced it has raised $12 million. The company says it will put the funds toward product R&D and expanding its geographic footprint.
According to a McKinsey survey, while 39% of executives say their companies recognize the risks associated with lack of “explainability,” or the ability to explain how AI models come to their decisions, only 21% say they’re taking steps to actively address the issue. But a separate report implies that “trustworthy AI” is fast becoming a business imperative. An IDC report found that business decision-makers believe fairness, explainability, robustness, data lineage, and transparency — including disclosures — are “critical requirements” in AI that need to be addressed now.
Truera’s technology builds on six years of AI explainability research undertaken at Carnegie Mellon University. With Truera, cofounders Shayak Sen and Anupam Datta, a professor at Carnegie Mellon University, sought to enable faster and more accurate AI-powered explanations and analytics.
Truera’s AI.Q platform, which can be deployed in either on-premise or private cloud software stacks, enables data scientists to explain model predictions and gain insights into behavior that might improve the development, governance, and operationalization of models. In addition to fairness, Truera performs sensitivity analyses to audit facets of model quality, including accuracy, stability, reliability, conceptual soundness, and analysis generalization, while supporting popular model development technologies like Facebook’s PyTorch and Google’s TensorFlow.
Truera is model-agnostic, meaning it works with all types of regression and classification models, including logistic regression, gradient-boosted and other tree ensemble models, and deep neural networks. And it can answer a broad set of questions about causal drivers of model predictions, as well as aggregate trends relevant to addressing unfair bias and instability. Moreover, Truera helps monitor models in production on a continuous basis, flagging any potential problems that could impact performance.
Truera competes to a degree with Fiddler, a Mountain View, California-based startup developing an “explainable” engine that is designed to analyze, validate, and manage AI solutions. There’s also Kyndi, which last June raised $20 million for explainable AI that analyzes documents. But Truera says its platform has already been deployed by a number of early Fortune 100 customers.
“Every company adopting machine learning struggles to build high-quality models that deliver on KPIs and that stakeholders can trust,” Truera CEO Will Uppington said. “Truera can help. Our model intelligence platform solves these issues during development, model review, testing, and machine learning monitoring. We’re thankful to Wing, Greylock, and Conversion Capital for supporting us on this journey.”
Conversion Capital and Greylock and new investors Data Community Fund, B Capital Group via the firm’s Ascent Fund, and Harpoon Ventures participated in Truera’s funding round. It brings the Redwood City, California-based startup’s total raised to $17.3 million.