SAS: AI Agents for Industry-Specific Challenges

The Silicon Valley sprint to build ever-larger models is exposing a widening gap in sector-specific expertise, with many organisations struggling to turn general-purpose AI into measurable results at scale.

SAS is expanding a suite of purpose-built, domain-tuned solutions aimed at the toughest operational challenges, forming part of a US$1bn commitment to industry solutions throughout 2026.

Manisha Khanna, Global Market Strategy Lead, Applied AI at SAS, says: “When organisations are left stitching together ad-hoc AI frameworks and experiments, they often fail to achieve the competitive edge they are looking for when they invest in AI.”

Balancing demand and operations

A clear target is sales and operations planning (S&OP), a linchpin process for retailers and manufacturers that has long relied on complex spreadsheets and multi-day meetings. Because it is so resource intensive, many companies only run S&OP monthly.

SAS’s Supply Chain Agent compresses that cadence by continuously balancing demand and operations in near-real-time. It allows teams to optimise inventory, production and logistics decisions as conditions change.

Business users can interact with the agent through a chat interface to explore scenarios, such as the impact of a 15% drop in demand. The agent returns recommended actions and explanations of how it reached those conclusions to build trust.

Kathy Lange, Research Director at IDC’s AI, Data and Automation Software practice, says: “Current pre-packaged agents tend to tackle basic processes; with Supply Chain Agent, SAS is compressing a very complex process, which could deliver significant value. This offering positions SAS to bring its longstanding supply chain knowledge to a new generation of agentic AI solutions.”

Digital twins and worker safety

SAS is using digital twins and synthetic data to improve efficiency and safety in industrial environments. Using Unreal Engine, the company creates virtual replicas of factory floors to simulate scenarios without risking physical assets.

Debuted at SAS Innovate 2025, these digital twins provide a proving ground to explore “what if” questions. Teams can test process changes, equipment layouts or staffing models before deploying them in the real world.

One major medical device sterilisation provider is using a SAS digital twin to identify bottlenecks that slow down life-saving services. The result is a data-driven way to prioritise fixes and remove constraints.

SAS Worker Safety also uses synthetic data and digital twins to train computer vision models on rare but plausible accidents. These include forklift collisions or machinery failures that are difficult to capture in real life.

Official UK figures for 2024 to 2025 show that 124 workers were fatally injured, with falls and machinery accidents making up a significant share. This underscores the need for better prevention tools on the shop floor.

The new offering creates realistic footage for safety model training without involving real employees. Using synthetic data also ensures that no personally identifiable information is exposed.

Financial fraud decisioning

To tackle rising financial fraud, SAS is launching Fraud Decisioning for Payments for real-time detection across a range of transactions. It is designed to protect customers while reducing false positives that frustrate users.

According to a study by SAS and the Association of Certified Fraud Professionals, 75% of anti-fraud professionals report a surge in scams targeting consumers. Additionally, 55% anticipate a significant rise in deepfake social engineering and Gen AI document fraud over the next two years.

Global banks, insurers and financial services organisations trust SAS for its fraud detection models, which protect consumer assets and identities. The ready-to-run models can be integrated without wholesale changes to existing systems.

Fraud Decisioning for Payments is trained on patterns from a consortium dataset contributed by major financial institutions. It spans credit card, debit card and ATM fraud, as well as digital wallet risk.

It also addresses emerging vectors such as money mule detection. By deploying these models on the SAS platform, institutions do not need to start from zero.

Across supply chains, industrial operations and financial services, a consistent theme emerges: impact depends less on raw model power and more on domain fluency, data pipelines and workflow integration.

By providing accelerators designed for specific functions that integrate with existing processes, SAS aims to bridge the gap between experimentation and practical deployment. 

The focus is on targeted, transparent tools that compress complex processes into timely, trusted decisions.