Why Structured Data May Be AI’s Next Enterprise Frontier
Enterprise AI (artificial intelligence) is expanding in a new direction — away from just language and unstructured content, and toward the structured data that actually runs businesses. In a recent article for Forbes, AI expert Ron Schmelzer explains that while large language models have dominated the AI conversation, the next frontier may be much more grounded: relational, structured enterprise data. The article highlights a wave of new enterprise-focused AI initiatives from vendors like Snowflake, Oracle, SAP, and Kumo, all aimed at bringing AI closer to the databases, transaction systems, and data warehouses where core business operations live. The idea is simple but important — most companies don’t run on text; they run on structured records like orders, payments, shipments, and customer histories. Traditional machine learning has long worked in this space, but it’s been slow and resource-heavy. Teams typically need to extract data from multiple systems, clean it, engineer features, and build custom models for each use case. The result is powerful but hard to scale. New “structured AI” approaches aim to change that by making models that understand relational data natively — working across tables, keys, and linked entities without requiring heavy transformation into text or manual feature engineering. Vendors argue this could speed up deployment and make predictive analytics far more accessible. The key distinction is capability. While large language models are strong at language tasks like summarization and coding, they are less precise when forced to interpret structured business systems. Structured models, on the other hand, are designed for outcomes like fraud detection, churn prediction, and supply chain optimization — where relationships between data points matter more than individual records. Schmelzer’s key takeaway is that enterprise AI is starting to split into layers. Language models will handle interaction and reasoning, while structured models will focus on prediction inside the systems where business value is actually created.
