Infinidat has rolled out a new retrieval augmented generation (RAG) workflow architecture, offered as a consultancy service for its storage customers. This setup lets companies seamlessly integrate current, private data from various sources into their AI systems, drawing from any NFS storage within their organization.
This launch aligns with a growing trend where storage companies are tapping into AI workloads and addressing RAG challenges, especially in generative AI. Many businesses struggle with data that is incomplete, outdated, or lacking vital insights that only their private data can provide. When a company develops generative AI, it typically trains the AI using datasets that might be too general or outdated, which can hinder progress.
Eric Herzog, Infinidat’s chief marketing officer, pointed out that many organizations are running internal generative AI projects using private data. They want to protect their intellectual property while also ensuring accuracy and minimizing errors. For instance, a large enterprise that generates extensive data across sales, support, and operations would need to enhance its performance, heavily relying on efficient storage solutions.
Customers expect accurate data in almost real time, which allows AI to process intricate details about components—like specifications or suppliers—and update that information continually. To support this, Infinidat offers professional consulting services to help clients access data tailored for RAG usage, provided it’s in NFS file storage format. Herzog emphasized that Infinidat excels at quickly retrieving data and metadata for these purposes because of its focus on metadata and its unique “neural cache” within the InfuzeOS environment.
Infinidat’s storage solutions, whether all-flash or hybrid, cater mostly to high-end enterprises and service providers. Their hardware features triple-active controllers and a neural cache system that directs data to the fastest media, boasting a cache hit rate exceeding 90%.
With this emphasis on RAG capabilities, Infinidat joins other storage players who are keen to support AI initiatives. For instance, Pure Storage’s CEO, Charlie Giancarlo, showcased his company’s AI advancements at their Accelerate event in June, highlighting storage write speed and availability. Meanwhile, NetApp announced new data management features for AI, focusing on data classification through its OnTap operating system during its annual Insight event in September.