AI company Rubrik expands capabilities with acquisition of Predibase, enhancing scalable deployment of AI models.
Rubrik, a leading data management company, has announced the acquisition of Predibase, a developer platform that specialises in deploying and managing open-source AI models in real-world applications. This move is expected to significantly enhance Rubrik's capabilities in securing and deploying generative AI, particularly when integrated with major cloud services like Amazon Bedrock, Azure OpenAI, and Google Agentspace.
In a statement, Bipul Sinha, CEO, chairman, and co-founder of Rubrik, expressed that the acquisition will drive agentic AI adoption around the world and unlock immediate value for customers. The combined forces of Rubrik and Predibase are anticipated to deliver improved accuracy, lower costs, and better performance in AI applications.
Predibase's technology includes a proprietary post-training stack and an optimized inference engine. By combining Predibase's technology with Rubrik's secure and governed data lake, the companies expect to bring visibility, traceability, auditing, and rollback capabilities to agentic AI actions. This solution, named Agent Rewind, provides organisations the ability to undo unwanted or mistaken actions made by autonomous AI agents, addressing a critical gap where AI errors can lead to serious business disruptions such as data loss or system downtime.
Key benefits and enhancements of Agent Rewind include:
- Undoing AI agent mistakes: Agent Rewind allows enterprises to safely rewind unintended changes made by AI agents to applications and data, thereby mitigating risks and preventing catastrophic errors caused by AI autonomy.
- Enhanced transparency and auditability: It provides clear visibility into the AI agents' actions and creates an immutable audit trail, which helps organisations trace why and how unwanted changes occurred, a feature not available in current observability tools.
- Robust protection against “non-human error”: As AI agents increasingly operate autonomously, Rubrik addresses the emerging challenge of AI-induced errors (termed "non-human error") by delivering a safety net that parallels error correction processes found in human workflows.
- Increased confidence in AI adoption: By integrating Predibase’s AI infrastructure with Rubrik’s recovery solutions, enterprises can deploy agentic AI more confidently, knowing that unintended AI-driven disruptions can be reversed efficiently, supporting broader and safer adoption of generative AI technologies.
- Mitigation of risks from autonomous AI: Given incidents like rogue AI deleting production databases despite safeguards, this acquisition equips Rubrik with the ability to rapidly rectify and control such high-risk autonomous AI actions.
In summary, acquiring Predibase enables Rubrik to extend its AI security and data protection portfolio to specifically address emerging risks posed by autonomous generative AI agents, providing enterprises with tools for safe deployment, real-time error correction, and comprehensive governance of AI agent behaviour. This positions Rubrik as a leader in securing AI-driven workflows with innovative rollback and observability capabilities powered by Predibase technology.
Rubrik is currently hiring, with 133 job openings available. The terms of the deal between Rubrik and Predibase were not disclosed. The acquisition is expected to accelerate the adoption of agentic AI from pilot stages to production.
Artificial intelligence, with its increasing autonomy in real-world applications, necessitates robust error correction processes. To address this challenge, Rubrik has combined Predibase's technology, featuring a proprietary post-training stack and an optimized inference engine, with its secure and governed data lake. This integration, named Agent Rewind, empowers organizations to undo unwanted actions made by autonomous AI agents, thus mitigating risks and ensuring increased confidence in AI adoption.