AI agents a new focus for backup and storage vendors
The next big generative AI gambit for enterprise IT vendors lies in AI agents, which storage and data protection vendors see as an opportunity to offer new services.
Storage and data protection vendors are seeing a new market opportunity in the dissemination of AI agents in enterprise organizations, and they are starting to respond.
Data storage company Ctera will release a new product so customers to create AI agents using data in its file and object gateway technology. Rubrik, a data protection vendor, has launched a service to protect AI agents, starting with Microsoft 365 Copilot.
AI agents are a premutation of the generative AI frenzy sweeping enterprise IT, according to analysts. Infrastructure platforms and service providers such as Oracle and Salesforce introduced AI agents this year. These agents, built off curated customer data collections, handle a variety of tasks from legal document collation to code troubleshooting.
Analysts expect generative AI agents will evolve these bots into agentic AI systems, where the AI bots that combine complex tasks autonomously to deliver an outcome usually moderated by a human. An example of an agentic AI system currently in production is self-driving cars.
Most companies likely cannot afford a multi-million-dollar investment in GPUs and other hardware for an in-house generative AI project, according to Mary Jander, an analyst at research firm Futuriom. They instead might turn to AI agents within vendor platforms to experiment with and adopt generative AI.
Vendors that handle data protection and storage needs are in a unique position in the agentic AI market, as these platforms already have a variety of places to access, analyze and prepare data for generative AI use, she said. Data backup vendors, already building connections with data management and security tools, could also find a niche in protecting these agents.
Jander said she expects more vendors to follow Ctera's lead to bring AI agents into their storage platforms to reach SMB and smaller enterprise customers.
"The [generative AI] technology is available only to the largest companies [currently]," Jander said. "We're starting to see the availability of an infrastructure for AI workloads, but we're still in a place of speculation and evaluation [for others]."
The different attributes between agentic AI versus generative AI.
Local talent
Ctera's upcoming Ctera Data Intelligence platform will incorporate an array of services so that customers can create AI agents.
AI agents, called AI Experts in Ctera's platform, can be created by administrators with specific characteristics and knowledge domains. Ctera's own retrieval-augmented generation engine can automatically update the agents as new data is added into platform.
These AI Experts can use cloud-based large language model (LLM) services from OpenAI and Microsoft or remain on premises with a customer's own language model. Companies can also incorporate identity policies so that AI Experts and users are limited to specific data sets for replies.
Generative AI projects live or die by the data available to them, said Aron Brand, CTO at Ctera. Storage gateways from companies such as Ctera already have visibility into enterprise data repositories and silos and should enable users to quickly draw out insights and value from those unstructured files.
"The thing that determines if an AI project is successful is the quality of the data," Brand said. "There's a lot of garbage in the file system of every organization, there needs to be a selection state where you curate the data for specific use cases."
Jander anticipates Ctera competitors, including Panzura and Nasuni, will similarly add AI agent capabilities to their platforms. Both are already adding features to enable data ingestion for LLMs.
"You're going to see a lot of companies involved in any way with storage really trying to become more involved in getting data ready and amenable for AI," Jander said.
Ctera Data Intelligence is available in early access to and will be generally available next year, according to Brand. It will be included with the Ctera file storage platform, priced by total customer storage used by the platform. Customers will need LLM specific tokens for transaction fees.
Self-policing
Rubrik sees a different opportunity in the rise of AI agents it believes the enterprise will need with last week's release of Rubrik Data Security Posture Management (DSPM) for Microsoft 365 Copilot.
This new Rubrik AI service, part of its Security Cloud platform, provides visibility of the Microsoft 365 data used by Copilot, alerting administrators to sensitive data distribution or improper permissions. As with other cloud services following the shared responsibility model, where customers are responsible for protecting cloud data, Copilot requires customers to monitor data confirmation and access to avoid misuses.
Enterprises with AI projects will likely need supplemental monitoring tools to track the deluge of unstructured data that could enter a generative AI project and contain sensitive or confidential information, according to Krista Case, an analyst at Futurum Group.
"It's using AI to protect AI," Case said. "That's become a general theme [in security]. AI is finding its way into our applications and devices, but we need AI to protect those applications."
Future adoption
AI agents in infrastructure platforms should be examined with a certain level of skepticism, according to Sid Nag, an analyst at Gartner.
Customers in the market for a storage or backup offering are buying such products for specific capabilities and should ask what generative AI would add or subtract from the product's purpose.
"The question to ask in all these situations is [what] value [is generative AI] trying to bring," he said.
Companies and analysts are also beginning to see the initial generative AI hype fizzle out. They are more closely examining what AI services will benefit the core business, Nag said.
Cohesity, another backup vendor, offers a generative AI service named Gaia that provides functionality similar to an agent AI for data stored within the Cohesity Data Cloud Platform. The vendor is looking to evolve Gaia into more of an assistant for backup and data management related tasks over knowledge vaults or agent services, according to Gregory Statton, vice president of AI solutions at Cohesity.
The question to ask in all these situations is [what] value [is generative AI] trying to bring?
Sid NagAnalyst, Gartner
The AI should enable backup administrators to perform their jobs better by breaking down specific alerts, logging and surfacing recent action histories as well as making remedy suggestions over a generalized data search, he said.
"Our [backup] administrators are already experts in the system," Statton said. "If anything, they'd want deeper intelligence that's more contextually aware of what they're working on."
Enterprises using platforms with generative AI or other AI capabilities should keep stock to avoid duplication and complexity, according to Mike Gualtieri, an analyst at Forrester Research.
"If every vendor is going to come up with some way of doing something, it becomes a mess at a larger enterprise," he said. "Any organization should step back and look at it strategically to see if there's one unifying [AI] system or a few."
Having AI agents closer to storage or infrastructure services isn't a bad idea, Gualtieri said, as performance or injecting new information into an agent's knowledge database can be more efficient.
Storage services will likely become a component of the stack managed through agentic AI in the future, he said, as automating the creation or deletion of storage resources and other tasks could free up administrators for more pressing tasks.
"[Agentic AI] is the last mile of automation [in] doing what a human could do," he said.
Tim McCarthy is a news writer for TechTarget Editorial covering cloud and data storage.
