NetApp IT is unlocking the power of AI analytics

At NetApp, we’re not just talking about AI—we’re actively exploring, testing, and implementing it to solve real business problems. As the hype around generative AI continues to swell, we’re taking a more focused approach, identifying high-impact use cases and driving meaningful results across our data ecosystem. Here’s how NetApp IT approaches AI analytics and what we’ve learned.

One of our first challenges was enabling chat-based interactions with documents in NetAIChat, our internal generative AI tool. Users could upload files—Word, PowerPoint, Excel—and ask questions about them. This worked well for most formats, but Excel posed a unique problem: generative AI models struggle with numbers. 

Instead of giving up, we pivoted. We replaced natural language interpretation with a new capability—Text-to-SQL. When a user uploads structured data, the system translates their question into SQL queries. The result? Accurate, actionable analytics from spreadsheets and structured files previously misunderstood by large language models (LLMs). 

Our next step was unlocking insight from our enterprise data warehouse, hosted on Snowflake. Rather than relying on human analysts to generate reports, we’re building a system where AI determines the data needed to answer a question and generates responses on the fly. For example, a request like “Show me bookings in Germany for Product X over the past quarter” would no longer require a ticket and turnaround time—it could be handled in seconds. 

  The goal isn’t just speed; it’s scale. We want to free up our analysts to focus on strategic work, not manual report creation.

While historical reporting is valuable, there’s a growing demand for real-time insights from transactional systems like sales and supply chain. We’re currently exploring how to integrate AI with those systems without compromising security or access controls. 

  Security is a critical concern. We’re not just looking at what data can be accessed; we’re ensuring only the right people can access it. Generative AI must honor data permissions just like any other enterprise system. That means no AI-driven shortcuts around sensitive information. 

NetApp also invests in traditional machine learning for use cases like capacity planning and ransomware detection. Here, historical data is used to train models to forecast future behavior, such as identifying early warning signs of ransomware based on anomalous system activity. 

  These models require specialized expertise and careful oversight, but they complement our generative AI efforts. Together, they help us answer “what happened?” and “what’s likely to happen next?”

We’ve also established an AI Governance Committee to ensure we’re using these technologies responsibly. This committee grew organically and now works with executive leadership to prioritize use cases that offer tangible business value, not just science experiments. 

  One of the biggest challenges we’ve encountered is managing expectations. Executives often want AI-driven answers with the same level of precision as human-generated reports. But generative AI isn’t built for 100% accuracy. It’s best suited for ad hoc insights, brainstorming, and decision support—not financial reporting. 

As we continue to develop AI tools, we’re refining ways to improve confidence in AI-generated answers, while communicating their limitations.

I believe AI will become as ubiquitous as the internet, but it will take time. Currently, NetApp is ahead of the curve, but the field is evolving rapidly. In FY26, our priority is to double down on proven use cases, operationalize them at scale, and ensure we invest in what delivers the most value.  
  
Our CEO recently emphasized in a companywide communication that “AI isn’t just a trend. It’s a strategic lever for driving efficiency.” At NetApp IT, we’re turning that vision into reality. 

For more information, please visit NetApp AI data management for AI applications.