New content-aware capabilities help IBM Storage Scale improve AI responses

Ask an AI chatbot who won last night’s football game, it will probably spit out the correct score. But ask an AI chatbot on a corporate website about one of the company’s recently announced products and you may get the reply, “Item not found.”

So much enterprise data is unstructured — locked inside emails, PDFs, presentations, audio and video files, and other opaque formats — is hard to interpret. The large language models (LLMs) underlying AI chatbots weren’t trained on this data, potentially limiting their value.

With today’s announcement of content-aware IBM Storage Scale, we believe we’ve found a way to address this problem. By applying innovative natural language processing techniques, we’ve developed ways to much more efficiently extract the semantic meaning from all kinds of content, making it easier to update AI tools to improve the quality of their answers.

Content-aware storage (CAS) is engineered to enhance the value of customers’ AI applications for faster time to insights, reduced costs, improved performance, enhanced security, and streamlined operations.

These new capabilities are based on work done at IBM Research and NVIDIA, leveraging new approaches to how words and their meanings can be represented as series of numbers called vectors. This is the magic that allows generative AI applications to retrieve information that doesn’t just contain the right keywords but relevant meaning. It builds on retrieval augmented generation (RAG) to help improve the accuracy of AI tools and allow AI models to stay updated with new information without retraining.

The new content-aware storage capabilities in Storage Scale make the RAG process far more powerful. We built upon the latest NVIDIA Blueprint, AI-Q, which uses NeMo Retriever designed to extract information from text, charts, graphs and even images. We automated and accelerated RAG processing to position organizations to derive greater business value from their existing data stores.

For instance, content-aware Storage Scale can watch folders on other storage systems or in the cloud to identify changes as they occur, automatically run pre-built pipelines, and update just the changes to the vector database, ensuring that data is current for AI applications efficiently using NVIDIA accelerated computing. Storage Scale also includes innovative global data abstraction services engineered to provide connectivity from multiple data sources and multiple locations to bring data into your AI factory from IBM and third-party storage environments.

For organizations aiming to optimize the business value of their data stores, content-aware storage presents many use cases:

• Improving AI assistant accuracy: CAS helps ensure that AI assistants have the latest, contextual information, which can help improve accuracy and address hallucinations.
• Up-to-date data insights: CAS watches folders for changes and updates the RAG database as data changes, enabling AI tools to extract insights as events unfold.
• Streamlined AI data pipelines – By embedding compute, data pipelines, and vector database capabilities within the storage system, CAS reduces data movement and latency to increase efficiency.
• Enhanced search: By using natural language processing, CAS extracts semantic meaning from unstructured data, facilitating the search for relevant information beyond keyword-matching.

Storage Scale also supports the new NVIDIA AI Data Platform, which brings enterprise storage into the era of agentic AI and helps IT leaders build distributed systems that unlock the value of business data to fuel data-driven actions. The platform is a customizable reference design for integrating NVIDIA accelerated computing, networking and AI software with enterprise storage, transforming data into actionable intelligence.

With the new content-aware capabilities in Storage Scale, generally available 27 March 2025, data engineers can now easily integrate compute, data pipelines and a vector database inside the storage system for efficiency gains and faster access to data and time-to-value.

To learn more, attend the webinar on “AI First Storage – Enhancing AI Results with Content-Aware Storage,” on 10 April 2025 at 9:00am ET.

Register here for the April 10 webinar