SAS Viya Copilot Explained: Building Machine Learning Pipelines in Minutes, Not Hours
SAS Viya Copilot is a set of software as a service (SaaS) features and capabilities that use large language models (LLMs) to provide users with a more intuitive and accessible way to work with SAS Viya offerings. SAS Viya Copilot is for developers, data scientists, citizen data scientists, and business analysts who are writing code, analyzing data, building machine learning model pipelines, and doing more across the data and AI life cycle.
SAS Viya Copilot is powered by a combination of proprietary SAS technology and Microsoft Azure technology to help you securely access, use, and manage your data. The Microsoft Azure technology used for SAS Viya Copilot centers on cloud-based infrastructure for building and deploying scalable applications. For more information, visit the SAS Viya Copilot page on the SAS website.
Your Guide to SAS Viya Copilot for Model Pipeline Development
Here our focus is on SAS Viya Copilot for Model Pipeline Development which is available with SAS Model Studio. SAS Viya Copilot in Model Studio acts as an intelligent assistant to streamline and enhance the process of building machine learning models.
To help you get started, here are some FAQs about SAS Viya Copilot for model pipeline development.
What is SAS Viya Copilot?
SAS Viya Copilot for model pipeline development is a Generative AI-powered assistant designed to accelerate the model building process. Users have everything they need to build powerful machine learning models with ease. Copilot’s generative AI can analyze model performance, provide model documentation, and explain how models work and how they use underlying data. It also recommends next best steps in the modeling process and can automatically add nodes to the model pipeline, helping users move from experimentation to production faster and with greater confidence.
It is designed to boost productivity, reduce manual effort and shorten time-to-value while supporting human-in-the-loop oversight for responsible AI use.
What are the requirements for using SAS Viya Copilot?
To use SAS Viya Copilot, you must be running SAS Viya 4 version 2025.12 or later, and your SAS Viya environment must be connected to the internet.
SAS Viya Copilot is not enabled by default. An administrator must explicitly turn it on through SAS Environment Manager by enabling the SAS Viya Copilot feature. For detailed instructions, refer to the SAS Viya Copilot page in SAS Environment Manager: User’s Guide.
To open the SAS Viya Copilot side pane in SAS Viya Machine Learning, click the Copilot icon in the upper-right corner of the window. The SAS Viya Copilot side pane appears.
The Copilot icon appears only when SAS Viya Copilot is enabled. If the icon does not appear, contact your SAS Viya Administrator.
Currently, SAS Viya Copilot for Visual Machine Learning supports natural-language English only. When you open the SAS Viya Copilot side pane, it appears in English regardless of your browser's locale setting.
How SAS Viya Copilot works?
Select any image to see a larger version.
Mobile users: To view the images, select the "Full" version at the bottom of the page.
- SAS Viya Copilot receives a prompt from you. The more concise the prompt is, the more likely that SAS Viya Copilot can provide you with useful information. Your prompt should include text or other content that is discoverable on the SAS Viya 4 platform.
- SAS Viya Copilot processes your prompt and passes an augmented prompt to the LLM in Microsoft Azure.
- SAS Viya Copilot receives the response from the LLM and performs post-processing activities.
- SAS Viya Copilot returns the response to you for your review and assessment.
How SAS Viya Copilot for model pipeline development help users?
There are many ways in which SAS Viya Copilot can assist you as you develop your pipelines in SAS Viya Machine Learning. Here are some of the common tasks with which it can help:
- getting specific information about SAS Visual Machine Learning
- adding a specific node to your pipeline
- determining the next node or nodes to add to your pipeline from the Data node or a data preprocessing node
- running a node
- getting information about a node, its parameters, or the algorithms that is uses
- improving nodes in your pipeline
- getting an analysis or summary of node results
Thus, SAS Viya Copilot helps with building machine learning models and gaining insights through a natural language interface. You can simply ask SAS Viya Copilot to guide you through workflows. It leverages GenAI and recommends the best next steps, reduces the quantity of steps for tasks that are manual or time-consuming, and completes the tasks to increase your decision making and productivity.
You can watch an overview Demonstration of SAS Viya Copilot in Model Studio on the SAS Support Communities.
Stay Tuned!
For a deeper dive into the key capabilities of SAS Viya Copilot for Model Pipeline Development, watch for the upcoming 3-part series:
- SAS Viya Copilot in Action (Part 1): Intelligent Assistance for Building Model Studio Pipelines
- SAS Viya Copilot in Action (Part 2): Smart Recommendations for the Next Nodes in Your Model Studio Pipeline
- SAS Viya Copilot in Action (Part 3): Improving and Optimizing Nodes in Your Model Studio Pipeline
How can I get the best suggestions from SAS Viya Copilot?
Getting the best suggestions from SAS Viya Copilot can be an iterative process. Your first prompt, or question, might not get the best results. If the reply is not what you are expecting, you can try rewriting your question or asking follow-up questions.
You can improve the quality of the AI-generated responses from SAS Viya Copilot by following a few best practices in mind when you are writing prompts:
- Write Prompts as Questions – Write the prompt the way you would ask a question. To ensure that the replies from SAS Viya Copilot are as relevant as possible, try to use terms that are consistent with those in your SAS Viya application. If a word might apply to more than one area of the application, be as specific as you can be about what you are referring to.
- Reduce Ambiguity in Your Prompts – To increase the likelihood of useful results, eliminate ambiguity by being specific and including context that helps define what you are asking about.
Does SAS Viya Copilot support conversational context and short-term memory?
If your first prompt does not return the information that you need, you can follow up with additional context.
SAS Viya Copilot retains the most recent questions that you asked in the current session and might use them to add additional context when it processes your prompt. You can clear the current conversation and start a new one with the copilot if you do not want the previous prompts and answers to be considered.
Prompts are retained only for the current session and are not stored after a session ends.
What guardrails does SAS Viya Copilot use to promote safe, ethical, and trustworthy AI?
SAS Viya Copilot uses the same guardrails as Microsoft Azure OpenAI, which include a content filtering system running both prompts and completions through classification models to prevent harmful content. Azure OpenAI also monitors for behaviors or content that may violate product terms. More details can be found in the Transparency Note for OpenAI Service.
SAS Viya Copilot rejects prompts that violate the following guardrails:
- Hate: Content attacking or discriminating against individuals or groups based on race, ethnicity, gender identity, sexual orientation, religion, appearance, disability, or harassment.
- Violence: Language related to harming people or property, weapons, bullying, terrorism, or stalking.
- Sexual content: Includes vulgarity, nudity, pornography, abuse, child exploitation, or sexual assault.
- Self-harm: Content promoting injury or suicide, including eating disorders and bullying.
- Protected material: Text or code from copyrighted sources without proper attribution.
- Groundedness: Ensures responses are factual and based on user-provided sources.
SAS Viya Copilot uses Azure AI Content Safety’s Prompt Shields, which block unsafe or adversarial inputs before content generation. These safety measures restrict LLM behavior, though adversarial inputs may still bypass them, making specialized filters essential.
Should I review every GenAI response from SAS Viya Copilot?
Human oversight is critical in AI-assisted processes. SAS products let users review, modify, or discard AI-generated content, ensuring human judgment remains central and AI acts only as an assistant. Clear visibility of AI contributions keeps users as the ultimate decision makers, reducing overreliance and promoting alignment with user intent.
You are responsible for all LLM-generated responses shared with end users or customers. Review them for accuracy, helpfulness, and alignment with your company’s values. This represents a classic human-in-the-loop scenario, with continuous human oversight and the ability for humans to intervene or override decisions at any point to address accuracy and bias.
Accuracy: LLMs can hallucinate or provide outdated information. Verify key details before acting on or publishing responses, such as ensuring technical support suggestions match current SAS Support knowledge.
Bias: AI may reflect human biases or generate harmful content. Check responses to ensure they are appropriate and safe for your audience.
For more on SAS’s commitment to security, privacy, compliance, and quality, see the SAS Trust Center.
Is user data protected when using SAS Viya Copilot?
Yes, your data is secured.
- No customer data is used to train any LLM.
- The SAS Viya platform securely transmits a request to the SAS Viya Copilot service in the Microsoft Azure environment specific to SAS, and the request results in one or more prompts to an LLM hosted in that Azure environment.
- SAS and Microsoft do not retain your data and content; it is stored only transiently for processing the request.
How does SAS Viya Copilot support Agentic AI across the Data and AI Lifecycle?
With the SAS Viya 2025.12 release, SAS Viya Copilot for Code Assistance and Model Pipeline Development is now available in SAS Viya 4. Additional copilots will be introduced over time, expanding agentic AI capabilities across more stages of the data and AI lifecycle. Together, these copilots will evolve into agents, forming a unified agentic AI system that spans the entire analytics lifecycle.
SAS Viya Copilot uses generative AI to turn natural language into analytics, insights and recommendations. Built on transparent, secure, human-controlled architecture, it lays the foundation for agentic AI, where trusted agents will automate tasks and orchestrate decisions while keeping humans in charge.
For a deeper perspective on how agentic AI will shape this evolution, see the post Agentic AI: Powering the Next Evolution in Data and AI Lifecycle.
Find more articles from SAS Global Enablement and Learning here.
