Backup vendors build generative AI inroads with partnerships
Data backup vendors see generative AI as a complement to existing standards, but experts say the potential is still a way off and likely to be democratized across sellers.
The partnerships between data protection vendors and generative AI platforms are starting to blossom, but industry analysts say enterprise needs and demands are still nascent.
Early partnerships are leaning on the popular OpenAI generative AI data sets as well as machine learning models under development by cloud hyperscalers such as Vertex AI by Google Cloud.
Unlike the pattern recognition technology of machine learning, generative AI offers the ability to analyze data at scale while creating new, unique results such as reports, images and more.
Enterprise data backup customers interested in hearing the AI pitch from their technology partners and vendors should carefully consider what data is being used to power the services, said Krista Macomber, an analyst at Futurum Group.
Generative AI is only as good as the data it's built on.
Krista MacomberAnalyst, Futurum Group
"If we're going to start using generative AI in IT processes, we need to be able to do so in a way that is secure and draws on quality AI," Macomber said. "Generative AI is only as good as the data it's built on."
AI incident response
Rubrik, with its roots in data backup and disaster recovery, rolled out a new integration last month for the Rubrik Security Cloud platform for Microsoft Sentinel and Azure OpenAI Service.
The integration enables Microsoft Sentinel, a cloud security information and event management web console, to collect information and trends from a customer's data on premises and in Microsoft Azure. This information is then fed to the Azure OpenAI Service to generate suggested tasks and incident response flow charts for recovery and forensic investigations following a ransomware attack or data breach. Other capabilities include code generation to investigate the incident and generate reports within Sentinel.
The capabilities won't replace the specific skills of security or backup employees anytime soon, said Anneka Gupta, chief product officer at Rubrik, but will complement their investigations by automating administrative drudgery.
"We don't think generative AI is going to replace humans, but augment them," she said. "We think this is true across IT ops and security ops."
Rubrik and Microsoft have a longstanding relationship, with Microsoft making a significant equity investment in Rubrik several years prior, but Gupta said Rubrik is interested in and willing to work with other generative AI partners in the future.
"We're open to partnering with everyone," Gupta said. "We do not have an exclusive relationship with Microsoft."
The Microsoft Sentinel dashboard now integrates with alerts and generative AI capabilities from Rubrik Security Cloud.
Another data protection vendor willing to play the AI suitor field is Cohesity, which in June unveiled a new partnership with Google's Vertex AI, a machine learning service, just a few months after announcing a partnership with Microsoft Azure OpenAI.
Around the same time, Cohesity rebranded its AI products as Turing. The new product line aims to provide a variety of AI models to help detect anomalous activity and ongoing ransomware attacks.
AI parlor tricks
Capabilities such as task automation and remediation suggestions are useful AI tools, but they are also just refinements of existing capabilities within data protection platforms, said Johnny Yu, research manager at IDC.
Most of these generative AI capabilities will not cause customers to jump ship for another backup vendor, he said. Instead, they might cause a customer to ask when and how such capabilities will be integrated into their existing data protection platform.
"Pushing [AI] forward is great, but how do you market that?" Yu said. "It's going to be hard to make a customer switch to another data protection vendor."
As the limits of generative AI's capabilities become clearer, so too should enterprise customer skepticism when hearing a data protection vendor's AI pitch, he added.
Most purported AI capabilities are still reactive in nature, similar to machine learning of years past, Yu said. Such capabilities are useful for pattern recognition and scanning massive data sets to find issues such as ransomware payloads, but lack the capability to preemptively stop a payload detonation or halt a hacking breach in real time.
As those capabilities become more democratized among vendors, Yu said, data protection customers should expect their platform to adopt the capabilities or be left behind.
"Pretty much all the other vendors will follow suit," he said. "It's the natural flow of things."
Tim McCarthy is a journalist from the Merrimack Valley of Massachusetts. He covers cloud and data storage news.
