SAS refreshes data management tools for AI governance

SAS has refreshed its Data Management portfolio, with updates focused on preparing and governing data for analytics, automation and AI.

The portfolio is cloud-native and built on its Viya data and AI platform. It adds and expands tools for data preparation, governance, AI-driven automation and analytics that run close to where data is stored.

The refresh comes as businesses try to move AI projects from trials into day-to-day use while managing fragmented data estates, manual engineering work and governance models that struggle to scale. SAS argued that weak data foundations are slowing deployment and increasing risk, even as AI software becomes more widely available.

Research cited by SAS pointed to the same problem. In a global study conducted with IDC, 49% of organisations said non-centralised or poorly optimised cloud data environments were the main barrier to AI progress, while 44% cited inadequate data governance processes.

Industry forecasts underline the issue. SAS cited Gartner research predicting that 60% of AI initiatives will fail because organisations lack AI-ready data.

Governance focus

At the centre of the refresh is an effort to make governance part of routine data work rather than a separate compliance layer. SAS said lineage, transparency and control are embedded directly into how data is accessed, prepared and used for analytics and AI, with the aim of giving organisations clearer audit trails and tighter oversight.

SAS is also trying to reduce the amount of data movement required for analytics projects. Many businesses still duplicate data across multiple systems to run models or analysis, adding delay, cost and governance concerns.

Its approach is to run analytics closer to the source data. One example is SAS SpeedyStore, a cloud-native analytical data platform integrated with Viya and intended to let analytics and AI operate alongside distributed data while preserving lineage and auditability.

That principle extends to third-party environments. SAS Data Accelerator is designed to let SAS analytics run inside cloud data warehouses and lakehouse systems already used by customers, allowing data to remain in place rather than being copied into another platform.

Viya also supports embedded analytics engines including DuckDB, allowing local analysis of open file formats such as Parquet, CSV and JSON within governed workflows.

Agents and copilots

Another part of the update focuses on data work that happens before AI applications are deployed. SAS said many assistants operate only after data has been prepared, which can leave gaps in lineage and oversight, so it is applying AI-driven assistance earlier in the data lifecycle.

The portfolio includes tools described as agents and copilots to help users understand, prepare and use data within governed workflows. The aim is to reduce manual effort while retaining transparency and human control.

Among them is SAS Viya Copilot for Data Discovery, which uses natural language prompts to help users explore governed data and analytics assets. It is intended to help users identify what data is available, how it can be used and whether it can be trusted.

SAS Viya Copilot for Code Assistance is aimed at developers working in SAS Studio. It helps users write, interpret and refine SAS and Python code through natural language prompts while staying inside the governed development environment.

The refresh also includes SAS Data Maker, which generates synthetic data intended to reflect the statistical, relational and temporal characteristics of real-world datasets. The tool is designed to let teams develop, test and collaborate without exposing sensitive information while maintaining privacy and auditability.

"A modern data platform is now a mission-critical requirement as organisations move toward agentic AI workflows with less human oversight," said Alyssa Farrell, Senior Director of Data and AI Strategy at SAS.

"SAS is redefining data management for the AI era by helping organisations optimise modern data estates, reduce complexity and unlock AI value, with governance and trust engineered directly into the foundation," Farrell said.