Databricks today announced that it has acquired Okera, an AI-focused data management platform. The two companies did not disclose the purchase price. According to Crunchbase, Okera previously raised just under $30 million. Investors include Felicis, Bessemer Venture Partners, Cyber Mentor Fund, ClearSky and Emergent Ventures.
Data governance was already a hot topic, but the recent focus on AI has highlighted some of the shortcomings of the previous approach to it, Databricks points out in today’s announcement. “Historically, data management technologies, regardless of sophistication, relied on enforcing control at a narrow waist layer and required workloads to fit into the ‘walled garden’ at this layer,” the company explains in a blog post. This approach doesn’t work anymore in the age of large language models (LLMs) because the number of assets is growing so fast (in part because so many of them are generated automatically) and because the AI landscape is changing so quickly, standard access controls can’t quickly pick up on those changes. enough.
Okera then uses an AI-powered system that can automatically detect and categorize personally identifiable information, tag it and apply rules to it (with an emphasis on metadata), using a no-code interface.
As the Databricks team emphasized, this is one reason the company was interested in acquiring Okera, but another reason is its service isolation technology, which can impose governance oversight on arbitrary workloads without any significant additional costs. This technology is still in private preview but is likely one of the main reasons why Databricks acquired the company.
Databricks, which launched its own LLM a few weeks ago, plans to integrate Okera’s technology into its module catalog, an existing governance solution for data and AI assets. The company also noted that the acquisition will enable Databricks to unveil additional APIs that its data management partners can use to provide solutions to their customers.
With this acquisition, Databricks also brings Okera co-founder and CEO Nong Li to the board. Lee created the Apache Parquet data storage format and was actually a brief engineer at Databricks between working at Cloudera and before starting at Okera, where he was the founding CTO and became CEO in February 2022.
“As data continues to grow in volume, speed, and variety across different applications, CIOs, CMOs, and CEOs across the board must balance these two often conflicting initiatives — not to mention that managing access policies across multiple clouds has historically been painful and time-consuming. Too long,” Lee wrote in today’s announcement. “Many organizations do not have the technical talent to manage access policies at scale, especially with the proliferation of LLMs. What they need is a modern AI-focused governance solution. We couldn’t be more excited to join the Databricks team and bring our expertise in building secure governance solutions and scalable and simple for some of the most forward-thinking organizations in the world.”
If you know more about this acquisition, you can contact Frederic at Signal at (860) 208-3416 or via email (frederic@techcrunch.com). You can also reach us via SecureDrop.