Backup vendors embrace GenAI, but features remain immature
Data backup and disaster recovery vendors are keeping up with the GenAI hype by quickly releasing new features -- but the use cases are limited, and their value remains unclear.
Vendors in the storage and backup infrastructure market are rolling out chatbot capabilities to establish a foothold, however small, in the generative AI hype for enterprise IT.
Three vendors with storage and backup infrastructure offerings have released copilot services or released guidance on how to connect a customer's IT into GenAI tools within the past month.
GenAI assistants or copilots gather and corroborate information through large language models (LLMs). GenAI creation platforms offered by companies such as Microsoft and AWS incorporate these LLMs, adding supporting cloud infrastructure or software, to abstract and expedite the creating copilot services for other customers. While similar in function, copilots are different from Microsoft Copilot, the gargantuan IT vendor's own GenAI assistant built on OpenAI LLMs -- and perhaps the best-known version of the trend. Microsoft Copilot also includes a variety of other AI-related services beyond the chatbot.
The uptick in copilot services continues the ongoing rush of GenAI products from 2023, but they have yet to prove their value for enterprise IT infrastructure, according to Jerome Wendt, founder and president at Data Center intelligence Group.
You have to quantify what you want AI to do and what problem you're trying to solve.
Jerome WendtFounder and president, Data Center Intelligence Group
Storage or backup buyers can use vendor copilots to automate certain tasks, such as identifying performance bottlenecks or creating incident summaries, but the true value and capabilities of these services remains to be seen, Wendt and other industry analysts said.
"You have to quantify what you want AI to do and what problem you're trying to solve," Wendt said. "[With AI,] the vendors will be more than happy to define it for you."
AI for sale
Nasuni, a hybrid cloud file namespace platform vendor, released guidance in April on how to connect unstructured data it manages with Microsoft Copilot Studio. Copilot Studio is a low-code SaaS tool to build and customize GenAI chatbots using OpenAI LLMs.
The guidance can help users build copilots from their own data using Nasuni IQ, a suite of metadata and data management capabilities added to the platform earlier this year, according to Nasuni spokespeople. Nasuni not only developed the guidance, but used it to build its own internal AI assistant, Ask Nasuni, which is integrated into Microsoft Teams.
Veritas, a data backup and recovery vendor, released the Veritas Alta Copilot to users of Veritas Alta View last week. The Alta Copilot, built using the Azure OpenAI Service, provides capabilities such as report generation and identification of potential issues within a customer's backup configurations.
Finally, Dell Technologies this week released Apex Backup Services AI for its cloud backup and disaster recovery offering. This assistant is a white-label version of Druva's Dru AI service, built on the Amazon Bedrock GenAI platform.
These implementations, and similar copilots released in the past year, have yet to fully capitalize on their promise of automation, said Mike Matchett, founder and analyst at Small World Big Data.
Most business-applicable implementations of GenAI within IT infrastructure are still playing catch-up to analyst timelines of capabilities and availability, he said.
"I was predicting we'd be here six months ago, but [vendors are] just rolling them out now," Matchett said. "I'm not sure [sellers] have proven their value and that this isn't just AI washing."
Generative implementations
Report generation or a natural language interface to communicate with operations manuals is still a benefit to most infrastructure IT teams, said Christophe Bertrand, an analyst at TechTarget's Enterprise Strategy Group.
GenAI assistants could help generalist IT employees manage backup software and conduct data backups using tools such as the Veritas Alta Copilot. These assistants help to simplify interactions and could eliminate a support backlog by empowering less specialized IT staff to handle an issue or, as the technology evolves, operate the entire backup environment.
"There are no more Ph.D.'s in backup," Bertrand said. "A lot of folks running the environment are IT generalists."
Infrastructure teams will likely find specialized GenAI tools such as HYCU Inc.'s API connection creation feature more useful than generic services, Matchett said.
HYCU uses Anthropic's Claude LLM to create code and connect SaaS application data via an API into its R-Cloud backup platform. HYCU's capability does not make changes to a customer's data stores or configurations without direct action by a user.
"It's a very constrained, very focused application," Matchett said. "It's hard to imagine it's going to get into too much trouble."
Code or configuration copilots might be a better use of GenAI, but come with their own legal and technical risks when used, he added.
Buyer beware
Customers investing in GenAI should understand clearly what party is responsible for issues that arise due to inaccurate or nonsensical outputs or the use of illicit data, said Steve McDowell, founder and analyst at NAND Research.
Companies such as Microsoft, Adobe and IBM provide indemnity clauses for their GenAI services, where the vendors will bear some responsibility for legal risks that arise from using their tools. Indemnity protections can still be narrow, however, as vendors might have specific legal liability thresholds.
"If you don't have the indemnification [contracts], it's a risk to use these tools," McDowell said. "There's still a lot of hesitation, and rightfully so."
Ultimately, using copilots and following their outputs will come down to the organization and individual employees, Wendt said. Customers skittish about GenAI platforms existing alongside confidential or proprietary data need to consider if the benefits outweigh the risks and their trust in the vendor.
"There is so much more we don't understand about AI than we do," Wendt said. "These companies are going to be gathering information about your environment in order to answer questions about it. The question is, do they care and is that something you want them to do?"
Tim McCarthy is a news writer for TechTarget Editorial covering cloud and data storage.
