The Great Semantic Layer Debate: Five Issues for AI Innovators
Controversial Conversations are Coming at the AtScale Semantic Layer Summit 2025
In an era where AI agents promise to transform business operations, the upcoming Semantic Layer Summit 2025, on May 28th, is a pivotal forum for business and technical leaders. This virtual conference will explore how semantic layers serve as the critical bridge between raw data and business meaning for AI, and address important questions about the impact of semantic layers on AI.
But, first, let’s back up: What is a semantic layer, and why are they essential to an effective AI strategy?
Wait: What’s a Semantic Layer, Again?
Imagine you're in a foreign country where dozens of languages are spoken. You need to communicate with locals, access services, and understand important information, but you’re at a standstill. This is the challenge that modern AI agents face with enterprise data: databases, applications, tools, and custom code all "speak" different languages.
A semantic data layer acts as a universal translator in this multilingual data landscape. It sits between raw data, stored in various systems and formats, and AI agents — the strangers in a strange land. Just as a translator converts foreign languages into words you can understand, semantic layers transform technical data into AI-understandable concepts and terminology.
In business, without a semantic layer, AI agents can’t understand that a "customer" in one system is the same as a "client" in another, that "revenue" has a specific business definition, and that certain data should be treated as confidential.
Think of it as the difference between looking at sheet music versus an orchestra playing it. Raw data is the sheet music of a business—technically precise but requiring specialized knowledge to interpret. A semantic layer is like the orchestra that transforms code on paper into beautiful music that anyone can appreciate, regardless of their technical background or skill.
It’s not just an analogy: research shows that semantic layers make AI responses three times more accurate than direct database manipulation, and that’s a big deal.
Important Semantic Layer Issues for the AI Age
The opportunity isn’t lost on tech entrepreneurs. Since they’re viewed as the missing link between enterprise data and AI, Semantic Layers were the darling of Gartner’s Data Analytics Summit in Orlando last month, and there are as many ways to express data semantics as there are musical genres.
Here are five essential issues that will be tackled by leaders from Databricks, Snowflake, IBM, and, of course, the host of the Semantic Layer Summit, pioneer in semantic data, AtScale.
The Battle for Semantic Standardization
With numerous approaches to semantic understanding, can the industry converge on open standards that enable AI agents to access and interpret data across organizational boundaries seamlessly? Or, will proprietary semantic models lead to new forms of confusion and vendor lock-in? What is the role of Anthropic’s Model Context Protocol, or MCP?
Dael Williamson, EMEA Field CTO at Databricks, addresses these questions in his session, "Open Semantics: AI and Interoperability." For technical leaders, this debate has profound implications. AI agents require consistent, contextual understanding of data to operate effectively across systems. Without standardized semantic layers and programming interfaces, organizations risk creating isolated pools of intelligence—essentially replicating the data silos of the past in a new, more sophisticated form.
Natural Language Interfaces: Promise vs. Reality
"For over a decade, business users have been told this-or-that tool would take them to the promised land of Self Service Reporting," Joseph Hobbs from Databricks wonders: Will natural language finally deliver on the promise of democratized data access, and what role will human analysts play?
Natural language querying (NLQ) interprets human instructions and translates them into precise data operations. But skepticism remains: Will NLP truly empower business users or create new forms of technical debt that IT departments must manage?
The Automation Dilemma in Data Integration
"Driving Innovation with a GenAI-Powered Smart Data Fabric and Universal Semantic Layer" addresses a third controversial area: How to balance automation and human oversight.
Joe Lichtenberg of InterSystems will suggest that smart data fabrics provide the unified foundation that agents need to operate across organizational silos. However, the spread of AI agents raises new questions about governance and control. How much autonomy should agents have in interpreting and integrating data? What safeguards are necessary to prevent semantic drift and misinterpretation?
The Future of Human Analysts
Will GenAI ultimately replace traditional business intelligence (BI) analysts or merely augment their capabilities? The panel on "The Future of Semantics and BI with GenAI," featuring leaders from Snowflake, IBM, Databricks, and ThoughtSpot, explores this question.
It’s a tricky transformation to tackle. AI agents accelerate insight generation and decision-making, but also alter the nature of analytical work, risking the loss of creativity, instinct, and context that humans bring to critical business decisions.
The Real-World Battle Between Cost and Complexity
Multiple sessions address the practical challenges of modernizing legacy analytics systems. Robert Skerry and Pedro Alexandre SimĂ£o from Celfocus explain how Vodafone Portugal balances the costs of modernization with the weight of maintaining legacy systems. And Adam Walker from Telus explores how a semantic layer helped simplify its network analytics by abstracting away vendor-specific complexity in "Scaling Telecom Analytics with a Semantic Layer."
These real-world illustrate the implications of AI agent ecosystems on existing systems, and how semantic layers help fit old school thinking with new, innovative AI systems.
Join the Conversation! Navigating the Semantic Layer / AI Frontier
As AI agents become more central to business operations, the semantic layer that enables them to understand and interact with organizational data grows increasingly critical. The AtScale Semantic Layer Summit 2025 will bring these controversies into sharp focus, challenging business and technical leaders to rethink their data strategies for an agentic future.
For decision-makers attending the summit, the key will be balancing pragmatic near-term implementations with longer-term strategic considerations about interoperability, governance, and the evolving relationship between human and artificial intelligence. Those who successfully navigate these controversial waters will position their organizations to harness the full potential of AI agents and agentic applications in the years ahead.
To follow my takeaways from the Semantic Layer Conference, follow me on LinkedIn or subscribe to
!
It's all semantics!