Business intelligence is evolving fast. In 2025, AI, machine learning, and real-time analytics will fundamentally reshape the way enterprises ask questions, deliver answers, and act on data. Dashboards are no longer just reporting tools; they’re interactive, intelligent systems. And business users aren’t waiting for insights; they’re generating them on demand.
But there’s one truth I’ve seen consistently across some of the world’s most data-forward organizations: no matter how powerful the AI or how sleek the interface, it all falls apart without a solid data foundation. That foundation is the semantic layer.
We saw this vision come to life at the recent 2025 Semantic Layer Summit, where data and analytics professionals joined us to explore the future of enterprise intelligence. The summit spotlighted a powerful shift: as Generative AI continues to revolutionize analytics, semantic layers are emerging as the critical layer that makes it all work. Sessions from leaders at IBM, Databricks, Snowflake, and ThoughtSpot reinforced the same point: semantic layers are no longer a “nice-to-have.” They’re essential infrastructure.
Real-world case studies from companies like Vodafone and TELUS showed the tangible impact, faster time to insight, stronger data governance, and a more empowered workforce. As the summit made clear, the organizations investing in semantic layers today are laying the groundwork for a more agile, trustworthy, and AI-ready BI tomorrow.
The summit, and the conversations it sparked about what’s next, got me thinking about the broader future of BI and where the rest of this year is headed. In the following sections, I’ll unpack the key trends shaping this evolution and explore why semantic layers are becoming the essential engine behind modern, AI-powered business intelligence.
Generative AI, Augmented Analytics & the Rise of MCP
Generative AI has gone from research project to real-world productivity driver. From Snowflake Cortex Analyst and Databricks Genie to the emergence of Model Context Protocols (MCPs), we’re seeing tools that allow business users to “talk to their data” and receive charts, forecasts, or narratives on the fly.
But there’s a problem: most GenAI tools lack context. They can guess at a question’s meaning, but without a defined, governed model of your business, they can’t guarantee accurate answers. As Dael Williamson, EMEA Field CTO at Databricks, put it:
When your CEO wants to ask a natural language question, the process to get an answer can be incredibly slow and fragmented. The real challenge is bridging the gap between technical teams who organize the data and business leaders who just need answers.
That’s where semantic layers shine, and now, with the announcement of AtScale’s MCP server, that power is reaching a new level.
MCP is a new standard protocol for connecting LLMs and autonomous agents directly to governed, semantic-rich metadata. AtScale’s implementation of the MCP server enables enterprises to securely expose their semantic models to any AI agent that supports MCP—including Claude, ChatGPT Enterprise, and custom-built copilots, while inheriting the same security, governance, and consistency already in place for BI tools.
Instead of hardcoding rules into every interface or worrying about metric drift across tools, organizations can now define rules once and use them anywhere across BI dashboards, LLMs, embedded apps, and autonomous agents.
This opens up use cases far beyond chatbots:
- Inventory alert agents
- Forecast optimization copilots
- Internal LLMs trained on secure, governed metadata
- Autonomous reporting assistants for operations, sales, and finance
AtScale customers are already exploring these paths, not by sending sensitive data to a third-party LLM, but by keeping their data and logic in-house, while empowering agents through AtScale’s MCP server.
Natural Language Query (NLQ) in BI
In a world where users expect to “Google” their enterprise data, Natural Language Query (NLQ) is redefining how business teams interact with analytics. But querying with plain language isn’t simple, especially when different teams define the same metric in different ways. One group says “customer churn,” another says “attrition rate.”
Without a consistent, governed understanding of business terms, NLQ quickly breaks down.
That’s where the semantic layer comes in. By creating a centralized, reusable model of your business logic, the semantic layer gives NLQ systems the context they need to return accurate, trusted answers—every time.
And with the rise of AI agents powered by MCP (Metric Context Protocol), semantic models aren’t just powering dashboards; they’re enabling machines to reason, act, and speak the language of your business.
As Adam Walker from TELUS explained:
“We needed to standardize KPIs across four different hardware vendors and multiple generations of network technology. Our customers expect the same TELUS experience everywhere. The semantic layer lets us hide complexity and ensure that no matter where someone is analyzing the data, they’re getting the right definitions.”
Composable BI, Data Governance, and the Agentic Future
With the rise of composable architectures and cloud-native services, BI stacks are becoming more modular, but also more fragmented. The risk? Rebuilding logic in every new tool, dashboard, or interface.
MCP eliminates that redundancy. By providing a single entry point to governed logic, AtScale’s MCP server acts as a semantic API layer, letting multiple agents, apps, and services all consume the same definitions, hierarchies, and business rules.
As our CTO, Dave Mariani, put it:
“The real differentiator isn’t just having an MCP server—it’s what you can do with it. AtScale can enrich LLMs with deep metadata, query history, and semantic context that spans every BI tool and user across the business. That’s how you build AI agents that act with intelligence and trust.”
Whether you’re trying to protect PII from leaking into AI prompts, build an internal co-pilot that understands your metrics, or empower non-technical users with safe NLQ, AtScale with MCP provides the foundation.
Collaborative BI and Decision Intelligence
Today’s business questions don’t respect departmental boundaries. Sales, marketing, and operations ask questions about the same customer journey. That means BI tools must support real-time collaboration, and the semantic layer must serve as the single source of truth that ties it all together.
At TELUS, the semantic layer became the foundation for this type of teamwork:
“Our semantic layer lets us scale up analytics with each new use case we add. That makes our teams more effective and creates a better experience for our customers.”
–Adam Walker, Senior Design Specialist, TELUS
Real-Time and Embedded Analytics
When data is embedded into operational systems, such as CRMs or customer service platforms, it must be both fast and trustworthy. There’s no time to check five dashboards or validate numbers when a customer is on the line.
Semantic layers enable live querying across cloud warehouses without requiring data movement or the creation of copies. That means faster insights, tighter loops, and confident decisions made at the edge.
Self-Service BI and Data Democratization
For years, “self-service” has been the holy grail of BI, but it’s been out of reach. That’s changing.
Rick Ramaker from The Home Depot shared how the AtScale semantic layer closed this gap for them:
“We wanted to enable thousands of users to explore data independently, but without risking inconsistent results. The semantic layer made self-service safe. It gave business users autonomy while protecting the integrity of our data.”
Modern BI demands: the freedom to explore, coupled with the guardrails to protect trust.
Explainable AI and Transparency in BI
In the age of AI, explanations matter. If a model recommends a course of action, leaders need to understand how it arrived at that conclusion, especially in regulated industries.
Semantic layers store every insight’s business logic, data lineage, and definitions. This transparency fosters trust in automated recommendations, making it easier to audit, improve, and align AI systems with business goals.
Data Observability and Data Quality
You can’t scale intelligence on top of chaos. Poor data quality leads to missed opportunities and bad decisions.
As Dael Williamson noted:
“We discovered that semantic data dramatically improves model accuracy. Good governance and structure are key to scaling AI”.
Semantic layers act as preventative data care, enforcing standards, surfacing issues, and maintaining integrity across massive pipelines.
Cloud-Native and API-Driven BI
Enterprises are increasingly assembling modular, cloud-native analytics stacks. The challenge? Making sure the pieces work together.
Semantic layers enable composability, acting as a universal interface between data sources, BI tools, and APIs. With a semantic layer in place, organizations can evolve their tech stack without having to rebuild their logic from scratch.
Data Governance and Compliance
Whether GDPR, HIPAA, or internal policy, governance is at the top of data leaders’ minds. The semantic layer allows you to enforce access control, row-level security, and auditability at the logic layer, streamlining compliance across the entire analytics ecosystem.
The Expanding Role of Semantic Layers
If there’s one trend that unites them all, it’s this: semantic layers are no longer optional. They:
- Power GenAI and NLP with context
- Enable governed self-service and collaboration
- Drive real-time, multi-cloud analytics
- Provide transparency and trust for AI and BI
- Enable secure, flexible AI agent access via MCP
At AtScale, we’re seeing semantic layers shift from center-of-excellence projects to enterprise-wide platforms. One of our customers even described it as the “operating system for data access.”
And it’s only the beginning. Now, with the MCP server added to the mix, the semantic layer serves as the interface for both humans and machines.
SHARE
Case Study: Vodafone Portugal Modernizes Data Analytics