SAS has introduced the SAS Viya MCP Server, a new layer designed to bridge large language models and enterprise analytics.

This article explains the problem it solves in modern LLM applications and the value it brings to governed, reliable AI-driven analytics workflows.

What is it and why does it matter? 

The business problem 

Remember the first GPT models that were released but could not solve simple mathematical questions. These models were text models; they could predict what word should come next based on the words it already generated. It was not trained on mathematicallogic but on word logic. It could understand that after ‘18 x 18 =’ there should be a number, but to choose the number it would just recall the structure of earlier texts and could come up with an arbitrary number. It did not ‘understand’ math. The same holds for writing programming languages to connect to databases, APIs, or other internal data sources; GPT models do not understand what is in your database inherently, which can result in erroneous SQL queries or JSON messages that are not compliant with the expected outcome in a workflow. 

Why does this matter? 

Often textual data is not the only source of data an enterprise owns. We, people, love to structure data and have been doing so over the last decades in order to generate insights ourselves. As a result, multiple databases, API’s, interfaces, and data storages exist in most companies, each with its own datasets, often consisting of structured data like tables, JSON messages, etc.  

At the same time, following best practices, AI agents are being set up in more deterministic frameworks these days. Upon giving a task to an AI agent, it will be forced to write an approach with steps first and then execute these steps. However, if a user asks for a revenue forecast for 2026, and the AI agent determines to predict this by making a forecast based on revenues in 2024 and 2025, one of the steps will be to extract the historical revenue information from a database. It should write the correct code to extract this information, not making mistakes like we saw in the early days of GPT models and their math capabilities. 

What the MCP solves 

The math problem was solved by OpenAI in an elegant way: instead of teaching a new GPT model how to do math, it just taught a new model when to call WolframAlpha, an online calculator, give it the input of the math problem, and return the output to the user. 

In order for an AI Agent to talk with SAS Viya, it should be familiar with the SAS REST APIs to retrieve these accurate results. The SAS Viya MCP Server is a dictionary that translates human language (‘run a linear regression’) into a SAS execution in a standardized way, using SAS to do the statistical work at the core, but initiated by an assignment written in human language and returning results that are accompanied by  interpretation and context, delivering a far more flexible ‘interface’ than a dashboard.  

Why use this MCP? 

In order to make an AI agent ‘talk with SAS’, this MCP server gives some big advantages: 

• No custom integration necessary, plug and play on every agent with any LLM 

• Uses the SAS Viya Authentication; every user can still access only their own data 

• Trustworthy and error-free output, as the validated SAS procedures are used at the core to calculate and visualize results 

• Results, tables, graphs: all can be sent back to the end user in the interface of their preference 

• Full accountability trail with comprehensive logging and auditability for every agent interaction 

Do you have questions or would you like to plan a free demo? Reach out directly to paul@notilyze.com  

About Notilyze

We are Notilyze, a SAS® Partner with only one goal: to help organizations with data analytics, and become more data-driven and successful. Our aim is to bring data analytics within reach of every department and every company. We do this by offering SAS® as a Service. Notilyze Cloud is the quickest solution to start with AI on SAS® Viya®. In addition, we offer IT solutions to properly integrate analytics into existing processes, and the experience of our SAS® specialists.