Oracle adds a semantic layer to ground AI agents in the business context
InfoWorld ·
Oracle is expanding its Autonomous AI Lakehouse with a Business Semantic Layer designed to give AI agents a consistent understanding of enterprise data, a move aimed at improving the accuracy of AI-generated insights and helping organizations scale agentic applications across business functions. To ground AI agents in business context, the Business Semantic Layer uses catalog annotations, SQL annotations imported from Oracle AI Database, and AI-suggested analytic views to create reusable models of enterprise data that capture dimensions, measures, hierarchies, calculations, and key performance indicators (KPIs), Oracle said in a statement. These models can then be used across Oracle’s natural-language querying, data retrieval, and analytics capabilities, including Select AI, to help ensure that queries and insights are based on consistent business definitions. Oracle said that the new capabilities will be available within the next 12 months. Alongside the new layer, at the company’s annual AI World conference, Oracle also launched a Business Simulation Engine to help enterprises evaluate business scenarios and make more informed planning decisions by integrating Oracle Essbase into the Autonomous AI Lakehouse. Beyond forecasting and what-if analysis, the engine supports write-back, dynamic hierarchy rollups, advanced calculations, and allocations, the company said. Enterprise users can also ask questions in natural language, generate simulation reports, and perform Excel-based analysis on data stored in Oracle databases or in object storage via the new engine, it added. Additionally, Oracle is expanding Apache Iceberg support in the lakehouse to let customers create, populate, update, and delete Iceberg tables using SQL, while improving performance and functionality when accessing Iceberg data through external catalogs or its AI Data Catalog. Business context could improve AI reliability, support scale-up That ability to ground AI agents in business context will help data professionals and developers build more reliable AI-driven analytics and applications by using approved business metrics and definitions, reducing the time spent investigating and reconciling inconsistent results, said Amit Kumar Jena , AI development head at IT Consulting firm Kanerika. “We at Kanerika did something similar for a global pharma company on Fabric , a rival offering from Microsoft, modeling 40-plus entities and standardizing 85-plus KPIs before the agents went live. Investigation and reconciliation time fell by about 70%,” Jena said. These improvements in reliability will, in turn, help CIOs scale agentic applications across business functions and rely on AI-driven decision-making with greater confidence, said Manoj Chandra Jha , principal analyst at Nord-IQ Research. Similarly, the addition of the Business Simulation Engine, too, has the potential to help enterprises speed up decision-making by bringing analytics and planning capabilities together within the lakehouse. “By bringing analytics and planning closer together, the engine could help finance teams and business planners run what-if scenarios using enterprise data without repeatedly moving it between separate systems,” said Pareekh Jain , principal analyst at Pareekh Consulting. For CIOs, that could simplify data architecture and accelerate decisions, while developers could spend less time building integrations between analytics and planning tools and more time developing applications that turn business insights into actionable workflows, Jain added. The expanded Apache Iceberg support could further simplify data management for developers and data professionals by allowing them to create and modify Iceberg tables using familiar SQL, rather than relying on separate tools or workflows, Jha pointed out. For data engineers, specifically, this means being able to work with a single copy of the data using SQL while retaining the flexibility to use Oracle alongside platforms such as Apache Spark, Snowflake, or Databricks on the same tables, he added. That reduced data duplication, in turn, for CIOs, could reduce associated storage and processing costs, while giving greater flexibility to work across different data platforms, echoed Jain. New capabilities bring implementation and integration challenges However, each of the new capabilities added to the Autonomous AI Lakehouse comes with its own set of trade-offs. The Business Semantic Layer, to begin with, will require enterprises to define and maintain business metrics, relationships, and rules to ensure AI agents interpret data consistently, Jena said. For CIOs, this creates an additional governance responsibility because if the models are incomplete or outdated, AI-generated insights could still be inconsistent or misleading, while expanding the layer across business functions could require sustained work from data teams to model and validate new metrics and relationships, Jena added. For the Business Simulation Engine, meanwhile, the challenge will lie in how well it fits into existing enterprise workflows, supports established planning processes, and integrates with systems already in use, particularly as Oracle competes with dedicated planning platforms such as Anaplan, OneStream, and SAP Analytics Cloud, Jain pointed out. The expanded Iceberg support, too, comes with caveats around data portability and vendor lock-in. While open table formats can make data easier to access across platforms, enterprises may still depend on proprietary catalogs and semantic layers to manage and interpret it, Jha said. Oracle has not specified when the Business Simulation Engine or the expanded Apache Iceberg capabilities will become available to customers.