I spent a meaningful chapter of my life at Microsoft, working closely with brilliant minds and incredible technology. Tools like SQL Server Analysis Services (SSAS) and SQL Server Machine Learning Services (MLS) were part of my everyday toolkit—and I’ll be the first to say they’re excellent at what they’re built for: structured analytics, business intelligence, and model execution within the SQL Server ecosystem.

But here’s the thing. The way people interact with data has evolved. Business users don’t want to write Data Analysis Expressions (DAX) or Multidimensional Expressions (MDX). They don’t want to call a data team just to ask, “How is the customer retention metric between last year and this year?” They want answers—accurately, instantly, intuitively, and in their own words.

Microsoft SQL Server Analysis Services

Microsoft created SQL Server Analysis Services (SSAS) to provide a powerful online analytical processing (OLAP) and data mining tool that enables organizations to analyze and make sense of complex information spread across multiple databases or disparate data sources. SSAS was designed to complement the SQL Server relational database engine, which excels at transactional processing but is not optimized for complex analytical queries and aggregations on large volumes of data.

The origins of SSAS trace back to Microsoft's 1996 acquisition of OLAP technology from Panorama Software, aiming to enter the OLAP market and enhance its business intelligence offerings. The first version, OLAP Services, shipped with SQL Server 7.0 in 1998, providing multidimensional analysis capabilities. Microsoft renamed it Analysis Services in SQL Server 2000, adding data mining features to complete its analytical capabilities and support richer BI solutions.

The core motivation was to enable enterprises to build enterprise-grade semantic data models on top of data warehouses, facilitating scalable, fast, and flexible data analysis and reporting. SSAS supports both multidimensional and tabular models, allowing organizations to choose the best approach based on their data volume and analytical needs. This semantic layer enables client applications like Power BI and Excel to deliver deeper insights with better performance than querying raw data sources directly.

Microsoft SQL Server Machine Learning Services

SQL Server Machine Learning Services is a feature within SQL Server that enables organizations to run Python and R scripts directly inside the database, leveraging open-source machine learning frameworks alongside relational data. This in-database execution removes the need to move large datasets across systems, streamlining advanced analytics and machine learning workflows.

Microsoft created SQL Server Machine Learning Services (originally introduced as R Services in SQL Server 2016) to bring advanced analytics and machine learning capabilities directly into the SQL Server database engine. The main reasons were to enable in-database processing of R and later Python scripts, eliminating the need to move large volumes of data out of the database for analysis, which reduces latency, enhances security, and simplifies compliance.

By integrating machine learning into the database, Microsoft aimed to allow data scientists and developers to prepare data, train, evaluate, and deploy machine learning models within the same environment where the data resides. This integration supports full dataset processing at scale, real-time scoring, and operationalization of models through stored procedures, making predictive analytics more efficient and accessible to enterprises.

The Natural Language Shift: A New Standard for Data Access

Today, data needs to be conversational. Whether it’s a frontline manager or a marketing analyst, the expectation is this: “I should be able to ask a question in plain English—and get a meaningful answer from my data.”

Natural Language Search (NLS) has moved from being a “nice-to-have” to a critical capability. But the legacy infrastructure many companies still rely on—SSAS, MLS, and similar technologies—just weren’t designed for this.

Why SSAS and MLS Struggle with Natural Language Search?

1. No Native NLP Capabilities SSAS is phenomenal with structured data models. But ask it to interpret a sentence like “Which regions had the highest YoY revenue growth last quarter?”—and it won’t know where to begin. MLS allows Python or R scripting, but you’re left building natural language logic from scratch. No semantic parsing, no pre-built NLP features.
2. Rigid and Technical Interfaces MDX and DAX aren’t exactly user-friendly, especially for non-technical users. MLS scripts, while powerful, require explicit task programming. There’s no fluidity—no way to dynamically interpret the variety and ambiguity of human language.
3. No Understanding of Intent or Context Humans are nuanced. We say things like “top-performing products” or “last month’s churn.” Systems like SSAS and MLS have no intent detection or semantic recognition. They can’t map synonyms or understand conversational context.
4. Cumbersome Handling of Unstructured Data Most organizations now rely on both structured and unstructured data—emails, CRM notes, chat logs. SSAS is optimized for cubes and tables. MLS on the other side can technically process text, but only through painstaking integration with external NLP libraries.
5. Performance and Scalability Issues Modern NLS systems need speed. Parsing, entity recognition, query generation—these must happen in near-real-time. SSAS and MLS weren’t built for that. Heavy NLP workloads within MLS, in particular, can strain database resources.
6. Disconnected User Experience Today’s data users expect to ask questions through an intuitive natural language search interfaces similar to OpenAI and Perplexity or through integration into productivity applications like Slack, Teams, or even voice assistants. SSAS and MLS are tied to BI dashboards and lack native APIs for these modern channels.

Summary: Where Legacy Tools Miss the Mark

Capability	SSAS/MLS	Natural Language Search Needs	Key Gap
NLP Understanding	❌ Not available	✅ Native NLP support	Cannot interpret human language
Intent Recognition	❌ No semantic engine	✅ Contextual and adaptive	No intent or contextual parsing
Unstructured Data Handling	⚠️ Manual & inefficient	✅ Native support for mixed data	Poor text data support
Real-Time Performance	❌ Not optimized	✅ Millisecond-level responsiveness	High latency, not scalable for NLS
Query Flexibility	❌ Requires MDX/DAX/scripts	✅ Open, intuitive input	No dynamic translation of questions to queries
Integration with Chat/Voice Systems	❌ Not available	✅ Seamless conversational UX	No native chatbot or assistant integrations

At Tursio, we have set out with a bold mission: To build the AI for future 100x workers.

Here’s how we’re doing it:

Accuracy You Can Trust

Tursio’s proprietary NLP engine is trained on business language across domains—sales, finance, marketing, operations. We use semantic models that don’t just understand keywords, but the meaning behind your question. This ensures your query always maps to the right dataset, metrics, and timeframes.

Speed That Matches Your Workflow

Our platform delivers real-time responses, optimized to run at scale. Whether your data lives in SQL Server, Snowflake, or BigQuery, Tursio interprets and executes your queries interactively. Ask a question. Get accurate answer. Keep moving.

Enterprise-Grade Security

As a former Microsoft engineer, I know how critical trust and data privacy are. That’s why we’ve built Tursio with enterprise-grade encryption, fine-grained access control, and full audit trails. Your data stays protected, and your governance rules stay intact.

Structured Data, Unleashed

Tursio is purpose-built to work with structured datasets—across cloud data warehouses like Snowflake and BigQuery, as well as on-premises systems like SQL Server. We eliminate the friction between data and the decision-makers by turning rigid queries into human conversations.

Empowering Data Access Across the Organization

Tursio isn’t just a NLS tool—it’s a transformation layer. Imagine:

• A sales rep asking, “What were my top accounts last quarter by revenue?”—and getting an instant answer.
• A finance analyst typing, “Compare spend vs. budget for Q1 across departments,”—no SQL analyst needed.
• A customer success manager querying, “How many clients churned after their renewal date?”—without opening a dashboard.

This is what data accessibility should look like. No code. No delays. No bottlenecks.

Conclusion: The Future of Querying Is NLS

For folks who are still navigating SSAS cubes and ML scripts—I see you. I’ve been there, and I understand the immense value those tools offer. But times are changing.

Natural Language Search is not a future feature. It’s a present-day necessity.

And legacy systems weren’t built for this new world.

At Tursio, we’re not here to replace everything you’ve built—we’re here to make it smarter, faster, and radically more accessible. We want to unlock the full potential of your structured data and enable every employee—whether they code or not.

So, here’s the real question:

Can your current system answer your next business question the moment you think of it?

If not, it’s time to see what Tursio can do.