SAS adds Copilot & agentic AI tools to Viya platform

SAS has expanded its Viya platform with new AI assistants and agentic AI tools, adding a Copilot offering, a Model Context Protocol Server and an Agentic AI Accelerator.

The additions target business and analytics teams looking to move AI projects beyond isolated trials and into governed use within existing workflows.

SAS Viya Copilot is a family of AI assistants embedded across the analytics life cycle. The tools are designed to help users analyse data, build models and support decisions from within the platform rather than through a separate chat interface.

According to SAS, the Copilot tools support a range of existing tasks in Viya applications. These include question-and-answer support for data discovery, model pipeline development, model management, decision intelligence and environment management, along with AI-generated SAS and Python code, documentation, model guidance, conversational dashboard creation and AI-assisted visual investigation.

SAS is also introducing a Model Context Protocol Server based on the open MCP standard. It is intended to let organisations expose SAS analytics, models and decisioning tools to external AI agents without rebuilding the underlying logic in another system.

Alongside that, SAS has launched the SAS Agentic AI Accelerator, a framework for building, governing and deploying AI agents within Viya. The package includes code, components, interfaces and working practices for teams using no-code, low-code and developer-led approaches.

Current Viya users can access these agent-focused tools through GitHub. SAS is positioning the move as a way to connect AI-driven insights, decisions and actions across business systems and teams.

Embedded assistants

The Copilot element appears central to the expansion. SAS said it integrates Microsoft Foundry, allowing users to work with AI support inside analytics workflows through natural language prompts in coding environments and visual interfaces.

Rather than limiting the offering to general-purpose assistance, SAS is also rolling out industry-specific versions. The first include SAS Asset and Liability Management Copilot for financial risk work and SAS Health Clinical Data Discovery Copilot for clinical data discovery and analysis.

The financial services version is designed to guide analysts through scenario configuration, execution and interpretation for risk workflows. In healthcare, the clinical data tool is intended to support data exploration, cohort creation and quality investigation through natural language queries.

SAS plans to extend industry Copilot functions to financial crime prevention in banking and to planning and supply chain optimisation in manufacturing.

Governance focus

Governance is central to the launch, particularly as businesses weigh how to use generative and agentic AI in production settings. SAS argued that combining assistants, agents and internal controls can help organisations maintain human oversight while automating more work.

Jared Peterson, Senior Vice President, Global Engineering at SAS, outlined that position in comments accompanying the launch.

"The role of human expertise in operationalising agentic AI is not diminished by automation; it's elevated," said Peterson. "With SAS Viya, organisations can pair copilots and agents with human judgment, trusted data and enterprise governance, so AI doesn't just generate outputs but drives responsible, real-world decisions."

Another part of that governance strategy is SAS Retrieval Agent Manager, known as RAM. The no-code product is built on retrieval-augmented generation and is designed to turn unstructured data into responses that reflect enterprise context.

RAM is currently sold as a standalone product, but SAS intends to bring those functions into Viya. That would give agents and assistants access to more context from internal data sources when generating responses and recommendations.

Broader platform push

The launch reflects a wider shift among software suppliers to package AI assistants and autonomous agents as built-in features of analytics and business software. Vendors are increasingly trying to distinguish between broad consumer-style chatbots and tools tied to specific data, rules and workflows inside companies.

For SAS, the message is that domain-specific analytics and governance should sit at the centre of these systems. It said Viya Copilot is designed to execute analytics tasks grounded in industry data models, terminology and working practices, rather than relying mainly on the general knowledge of a large language model.

The changes also show how open standards are starting to shape the market. By using MCP, SAS is aligning its platform with an emerging way of connecting AI models and agents to external software tools, which could make it easier for customers to link Viya functions with third-party interfaces.

SAS said the latest additions are meant to help organisations design AI systems that operate across teams and processes while remaining subject to internal controls.