Rubrik Acquires Predibase to Fuel Generative AI Expansion

Rubrik is set to acquire Predibase, a startup founded by a former Google Cloud AI product manager, marking a significant step to mitigate challenges in adopting agentic AI within enterprises. The Silicon Valley-based data protection company aims to facilitate the transition from initial AI prototypes to fully scalable applications that prioritize trust and security, as stated by Chief Business Officer Mike Tornincasa. This acquisition is poised to further bolster Rubrik’s cyber resilience platform by integrating advanced AI capabilities.

“We perceive generative agentic AI as essential for driving productivity,” Tornincasa emphasized. “Two primary factors are constraining enterprise adoption: governed and trusted data, which we provide, and improving accuracy alongside cost and deployment optimization. A third barrier is the talent gap; with Predibase, we aim to address the second and third challenges.”

Founded in 2021, Predibase currently employs 25 individuals and has secured a total of $28.5 million in funding, including a recent $12.2 million Series A round led by Felicis. The company has been under the leadership of Devvret Rishi, who was previously the chief product officer before stepping into the role of CEO in October 2023. Rishi articulated the company’s mission as democratizing access to AI technologies and enabling organizations to transition from proofs-of-concept to operational generative AI applications.

While enterprises often embark on generative AI initiatives targeting internal applications, moving towards production-ready solutions for external use cases remains challenging. Key issues include securely accessing proprietary datasets, maintaining governance compliance, and ensuring the accuracy of generative models. Rishi noted that identifying unlocked potential use cases extends beyond internal applications to external ones, emphasizing a need for better integration solutions to bridge the gap.

At the heart of this acquisition is Rubrik’s data lake, which intelligently consolidates enterprise data and metadata while adhering strictly to compliance and governance standards, thereby ensuring that AI models access high-fidelity data. Tornincasa suggests that Predibase’s model-hosting layer will empower customers to customize and deploy AI models tailored to their proprietary data securely.

Despite advancements, organizations frequently grapple with scalability issues in AI deployments, often achieving only 70% to 80% accuracy, which Rishi identifies as inadequate for many production scenarios. Predibase aims to enhance model performance through a post-training stack that allows enterprises to refine models to exceed the 90% accuracy benchmark crucial for large-scale application reliability.

Enterprises are particularly cautious about widespread AI adoption, requiring assurances that the data utilized is accurate and well-governed. Rubrik’s robust identity-based access management system plays a pivotal role in safeguarding data, ensuring that access is strictly regulated. Tornincasa posits that this foundational architecture will be vital as enterprises seek to implement secure AI applications.

While advanced technology firms possess the in-house resources to deploy sophisticated AI models, many enterprises do not. This acquisition allows Rubrik to leverage Predibase’s platform to extend AI capabilities to less technically advanced organizations, reducing the operational burden and simplifying the integration process. “Predibase has succeeded primarily with innovative companies that are quick to adopt AI,” Tornincasa noted. “Yet the broader market lacks the talent and infrastructure necessary to fully utilize these new capabilities.”