SAS supercharges Viya platform with AI agents, copilots, and synthetic data tools

At its Innovate conference, SAS on Wednesday announced a series of offerings across its portfolio that focus on AI in all its forms. They ranged from a series of enhancements to its Viya platform to a set of new custom AI models aimed at specific sectors, plus some governance resources to help organizations reduce risk from the technology.

SAS Viya and AI agents

The company announced new and enhanced components for the SAS Viya platform that are aimed at both developers and end users.

SAS Data Maker, a synthetic data generator which, SAS said, helps organizations tackle data privacy and scarcity challenges, has been enhanced with technology from the company’s recent acquisition of synthetic data pioneer Hazy. It looks at a source dataset that will be used as a basis for the synthetic data, automatically figures out the structure, and produces an entity relationship map that can then be tweaked and used to train the generative model to create high-quality synthetic data.

With the current release, it has been upgraded to produce multi-table data, and time series data. Data Maker has been in private preview, but is soon moving to public preview, and is expected to be generally available in the third quarter of this year.

SAS Viya Intelligent Decisioning, available now, helps users build and deploy intelligent AI agents via a low-code/no-code tool, with what the company describes as “The just right AI autonomy to human involvement ratio to strike the optimal oversight balance for task complexity, risk, and business goals.” For example, an agent vetting mortgage applications can flag specific denials for human review; the human can then query the agent to explore its thinking and make the final decision.

SAS Viya Copilot, now in private preview, is built on Microsoft Azure AI Services. It is an AI-driven conversational assistant embedded directly into the SAS Viya platform, to give developers, data scientists, and business users alike a personal assistant that accelerates analytical, business, and industry tasks. The initial Copilot offering in Model Studio includes AI-powered model development and code assistance for SAS users. It will be generally available in the third quarter.

Initially released in 2024, SAS Viya Workbench receives support for R language coding, SAS Enterprise Guide as an optional integrated development environment (IDE), and is now available on Microsoft Azure Marketplace as well as AWS Marketplace.

“While these updates may not be groundbreaking, they integrate features with built-in governance and ready-to-use models, crucial for enterprise use,” said Robert Kramer, VP & principal analyst, Enterprise Data, ERP & SCM, Moor Insights & Strategy. “Customers may be able to benefit from faster onboarding, easier collaboration, and more secure AI development, especially in regulated industries where auditability and model transparency matter.”

Prebuilt models

AI was also the basis for several other new SAS offerings.

In addition to Intelligent Decisioning, the company introduced six custom AI models to address specific processes in various industries.

“Our customer space is actually divided into two segments,” said VP of Applied AI and Modeling Udo Sglavo, during a media briefing. “One, they do have data science teams. The data science teams most of the time do an awesome job in their domain. The other camp, which is actually a little bit bigger, is they don’t have any data science resources, so they don’t know how to get started. We believe that with SAS models, we are addressing the needs of both segments of our market.”

He pointed out that, thanks to the models, companies with data science teams can shift their focus to strategic questions for the company, and leave the other questions to the model. For companies that don’t have a data science team, he said, “It’s a quick way to get started and see the impact of AI on your business models right away. So once again, value from the first day.”

Two of the models, AI-driven Entity Resolution and Document Analysis, are suitable for many industries. The Medication Adherence Risk model targets healthcare, Strategic Supply Chain Optimization serves manufacturing, and the final two, Payment Integrity for Food Assistance and Tax Compliance for Sales Tax, are for the public sector.

Later this year, four more models will join the group: Fraud Decisioning for Payments and Card Models for banking, Payment Integrity for HealthCare for healthcare, Worker Safety Monitoring for manufacturing, and Tax Compliance for Individual Income Tax for the public sector.

These models, said Sglavo, are lightweight and easy to deploy. “They basically live in a container, which you can plug into any ecosystem, just like SAS Viya. And you can get them up and running right away.” And, he added, “they are built around real-world industry cases, so we are turning practical use cases into software which you can productionize right away and get value right away.”

Kramer agreed. “The availability of pre-built AI models for applications such as fraud detection, supply chain planning, and health risk assessment should help organizations accelerate their AI adoption by providing ready-to-use solutions,” he said.

Governance

As AI becomes more pervasive in the enterprise, it also presents risk.

“It’s important that we’ve got a way to assess intended use and expected outcomes before deployment, and that we’ve got the ability to monitor for ongoing compliance,” noted Reggie Townsend, VP of the data ethics practice at SAS. “Now, this is a matter of oversight, for sure. This is a matter of operations, and this is a matter of organizational culture. And all of these things combined are what represent this new world of AI governance, where there’s a duality that exists between these conveniently accessible productivity boosters that the team has been talking about this morning, but we’re intersecting with inaccuracies and inconsistency and potential intellectual property leakage.”

To that end, SAS has introduced new governance resources to help organizations assess their current AI governance maturity in four essential areas: oversight, compliance, operations, and culture. Known as the AI Governance Map, the tool is the latest addition to the company’s suite of governance products. And more are on the way; SAS has announced an upcoming product, designed for executives, that it described as “a unified holistic AI governance solution” able to aggregate, orchestrate, and monitor AI systems, models, and agents.

Abhishek Punjani, research analyst – AI at Info-Tech Research Group, approved of the direction SAS has taken. “In the race to innovate with AI, many organizations made a fundamental misstep early on in their AI journey by putting innovation and speed above control, sometimes at the cost of long-term resilience,” he noted. “However, the tide is beginning to swing in a more responsible and balanced direction. With its latest and agentic AI innovations, SAS is at the forefront of the industry’s movement toward a more balanced and responsible path forward. Through a combination of ethical calibrations, SAS intends to create enterprise value through AI systems that are impactful and ethically sound, allowing for tailored levels of human oversight and intervention.”

Punjani also liked the approach taken with the rest of the SAS announcements.

“Building on this governance-first philosophy, SAS has also expanded its Viya platform with a suite of tools aimed at practical AI enablement,” he said. “SAS Data Maker addresses one of the most prominent issues in AI today, data scarcity and privacy, by generating secure synthetic data for safe model training. SAS Viya Intelligent Decisioning enables organizations to build AI agents with customized human involvement, allowing users to embed policy, logic, and rules into their AI agents for adaptive actions.”

“Together, these solutions mark a shift toward more grounded, enterprise-ready AI,” he said. “Rather than chasing scale alone, they reflect a growing focus on control and accountability, qualities that are becoming essential as AI becomes central to important business operations. As more organizations look for ways to move beyond experimentation, approaches like SAS’, which build governance and flexibility into the product itself, are reshaping what mainstream AI adoption looks like. It’s a reminder that the next phase of AI adoption won’t be driven by scale alone, but by how well these systems can integrate into business processes with clarity.”