Semantic intelligence is a vital ingredient of truly understanding what information means and the way it may be used.
Microsoft is now deeply integrating semantics and ontologies into its Material information platform with its new Material IQ expertise that it debuted on the Microsoft Ignite convention Tuesday.
Material IQ is a semantic intelligence layer designed to deal with a elementary drawback with enterprise AI brokers: Effectiveness relies upon not simply on dataset measurement however on how properly information displays precise enterprise operations. The brand new expertise creates a shared semantic construction that maps datasets to real-world entities, their relationships, hierarchies, and operational context. The semantic layer represents the most recent step in Microsoft's information platform technique, which not too long ago built-in LinkedIn's graph database expertise to supply context.
Microsoft can be increasing its information portfolio with a sequence of recent companies: Azure HorizonDB, a PostgreSQL-compatible service in early preview, in addition to SQL Server 2025 and Azure DocumentDB, which are actually usually accessible.
"When I think about what fabric does for customers, it gives customers a unified data platform so that they don't have to stitch together many, many, many different tools to get to business value," mentioned Arun Ulag, company vp of Azure Information at Microsoft.
Why semantic understanding issues for AI brokers
Conventional AI brokers battle with a elementary limitation: they will see patterns in information however don't perceive what that information represents in enterprise phrases. An agent would possibly analyze gross sales transactions with out understanding buyer hierarchies, seasonal patterns or product relationships. It could question stock ranges with out figuring out how manufacturing strains hook up with distribution networks or how provider relationships have an effect on availability.
This hole between uncooked information and enterprise which means is what causes unreliable predictions and poor automated selections. Ulag defined that Material IQ addresses this by offering a semantic layer that captures how organizations truly function.
This architectural strategy differs considerably from retrieval-augmented technology (RAG) and vector database methods that opponents have emphasised.
Whereas RAG pulls related paperwork to supply context, Material IQ creates a persistent semantic graph representing organizational construction, workflows and enterprise logic. Brokers don't simply retrieve data. They perceive relationships like which suppliers present which merchandise, how manufacturing strains hook up with stock methods or how buyer hierarchies map to gross sales territories.
From analytics semantic fashions to operational ontologies
Microsoft has invested in semantic fashions for over a decade by means of Energy BI. These fashions encapsulate enterprise logic and outline entities and relationships; they specify metrics and hierarchies; and so they hook up with various information sources throughout Azure, AWS, Google Cloud, on-premises methods, and SaaS platforms like Dynamics 365.
"We have 20 million semantic models that run in fabric today. Why? Because we built the semantic modeling layer into Power BI. So behind every Power BI report is a semantic model," Ulag mentioned. "These semantic models already encapsulate a lot of the business logic that mirrors what a customer cares about. What is the data that they care about? What are the metrics that they care about? How does the data relate to each other?"
The limitation of those semantic fashions has been their scope. They labored properly for enterprise intelligence, analytics, and visualization, however they solely operated inside particular person reviews or departmental boundaries. Material IQ removes these constraints.
"However, we've had a gap. These semantic models were only used for BI use cases," Ulag mentioned. "There's a much bigger opportunity out there, which is the opportunity to be able to take these semantic models and upgrade them into a full ontology."
Upgrading the semantic fashions to ontologies basically modifications what organizations can do with enterprise context and which means. "What does it do if you upgrade them into an ontology? What happens is that now you can connect data across your enterprise," Ulag mentioned.
He defined that the ontology additionally integrates with real-time information streams. Past connecting information, ontologies enable organizations to outline operational guidelines. This mix creates the inspiration for operational brokers that perceive enterprise context at a degree that conventional AI methods can’t obtain. Cross-enterprise information connections work along with real-time integration and rule definitions.
Operational brokers that perceive and act on enterprise operations
Material IQ allows a brand new class of brokers Microsoft calls "operational agents." These brokers can autonomously monitor information and take motion based mostly on the ontology's understanding of enterprise operations.
"We're also introducing something called operations agents in fabric that can watch your data for you, that can watch the rules that you're asking it to monitor. And it can autonomously take action under human supervision," Ulag mentioned.
Ulag offered a provide chain instance that illustrates the distinction from conventional approaches. A corporation can mannequin its provide chain and supply operations within the ontology. When real-time information exhibits congestion in a part of a metropolis, the operational agent can routinely reroute vans round the issue.
The ontologies created in Material IQ combine immediately with Microsoft's agent growth platforms. This offers enterprise context that makes brokers extra dependable and correct.
"It really takes the work that we've done in semantic models in fabric with unified data to a completely different level, allowing customers to be able to model their operations and take business actions," Ulag mentioned.
What this implies for enterprise AI methods
There appears to be a necessity for context engineering to higher allow agentic AI.
Semantics and their related ontologies just do that and extra. Context is about understanding why a request is being made, and semantics perceive the deeper which means. For enterprises combating AI agent reliability regardless of giant datasets, Material IQ represents a basically totally different strategy. It strikes past scaling compute or fine-tuning fashions. The vital query is whether or not enterprise context captured in ontologies would enhance agent effectiveness greater than conventional optimization paths.
The strategic guess Microsoft is making is evident: Semantic understanding of enterprise operations determines AI agent effectiveness. Entry to giant datasets alone will not be sufficient. Upgrading current semantic fashions into operational ontologies might present a sooner path to dependable brokers.
