Dataiku launches LLM Guard Services to enhance generative AI deployment

Artificial intelligence data analytics startup Dataiku Inc. today announced the launch of a new suite designed to advance enterprise generative AI deployments at scale from proof-of-concept to full production.

Called LLM Guard Services, the new suite offers generative AI deployment that does not compromise on cost, quality or safety. The suite includes three solutions — Cost Guard, Safe Guard and Quality Guard — that integrate with the Dataiku LLM Mesh for building and managing enterprise-grade generative AI applications that will remain effective and relevant over time.

The suite aims to tackle the issue wherein enterprise leaders want to use fewer tools to reduce the burden of scaling projects with siloed systems, but many do not have specific applications or processes for managing large language models. Available as a fully integrated suite within the Dataiku Universal AI Platform, LLM Guard Services addresses the challenge and mitigates common risks when building, deploying and managing generative AI in the enterprise.

“As the AI hype cycle follows its course, the excitement of two years ago has given way to frustration bordering on disillusionment today,” said Chief Executive Florian Douetteau. “However, the issue is not the abilities of generative AI, but its reliability. Ensuring that generative AI applications deliver consistent performance in terms of cost, quality and safety is essential for the technology to deliver its full potential in the enterprise.”

Dataiku’s LLM Guard Services offer oversight for enterprises using LLMs, focusing on three key areas. Cost Guard monitors and traces LLM usage to ensure budget management for generative AI, while Safe Guard protects against data misuse by evaluating requests for sensitive information and preventing leakage through customizable tools.

The suite also includes Quality Guard, which provides quality assurance by automatically evaluating LLMs without coding and is tailored to each use case. In doing so, it ensures objective and scalable assessments, helping maximize response quality in enterprise AI applications.

Through the use of Quality Guard, Dataiku customers can automatically compute standard LLM evaluation metrics, including LLM-as-a-judge techniques like answer relevancy, answer correctness and context precision, along with support for statistical techniques such as BERT, Rouge and Bleu. Quality Guard also democratizes generative AI applications, allowing any stakeholder to understand the move from proof-of-concept experiments to enterprise-grade applications with a consistent methodology for evaluating quality.

Dataiku is a venture capital-backed startup that has raised approximately $1.04 billion over nine rounds, including a round of $200 million in December 2022. Investors in the company include Tiger Global Management, Iconiq Capital, CapitalG LP, Battery Ventures LP, Dawn Capital LLP, Insight Partners, Eurazeo SE, Lightrock LLP and FirstMark.

Jed Dougherty, vice president of platform strategy at Dataiku, spoke with theCUBE, SiliconANGLE Media’s livestreaming studio, in January 2023 to discuss how Dataiku aims to teach everyone how to analyze with its machine learning platform:

Image: Dataiku