SAS Viya Copilot for Code Assistance in SAS Data and AI Studio
Recent Library Articles
In this post, we’ll take a look at how SAS Viya Copilot for Code Assistance can help with some of those everyday tasks: understanding code, improving it, and making changes more confidently—all without leaving the SAS Data and AI Studio environment.
Clinicians and reviewers still hand-search large volumes of unstructured documents before a decision, which is slow and inconsistent. Deterministic business rules control the decision while AI extracts and summarises the supporting evidence, evaluates it against defined policies and presents an auditable result — aimed at trial recruitment, referral triage and surgical waitlist review.
... View more
Team Name
🧙🪄 FCL Supply Chain Wizards 🪄🧙
Track
Demand Planning (Retail and Consumer Goods)
Use Case
Modernizing promo forecasting model through modern SAS tools and exploring on-prem data feed to SAS cloud.
Technology
SAS Viya
Region
North America, Saskatchewan
Team lead
@MarkCantuba
Team members
@MarkCantuba
@JustRenn
@sophiafu
Social media handles
Is your team interested in participating in an interview?
N
Optional: Expand on your technology expertise
... View more
SAS Data Maker • Product Update
Conditional generation: more control over the synthetic data you create
When we published our last product roadmap update, conditional generation was one of the most anticipated capabilities on the SAS Data Maker roadmap. The goal was simple: give users the ability to generate synthetic data for specific scenarios, segments, and outcomes, rather than always generating data that mirrors the overall source population.
We're moving closer to that goal. Conditional generation gives organizations greater control over the data they create, while continuing to preserve the statistical characteristics and relationships learned during training.
Why conditional generation matters
Organizations often need data for specific business scenarios. Conditional generation helps teams create targeted synthetic datasets for purposes such as:
Testing an underwriting model using a particular risk segment
Studying rare patient outcomes in clinical research
Creating validation datasets for underrepresented populations
Simulating targeted business conditions before deployment
What's new
Conditional generation is now available from the Generation tab for generators trained using supported algorithms.
After training a generator, users can specify conditions when generating data, allowing them to create synthetic records that align with specific needs without retraining the generator.
Categorical conditions
Generate synthetic records where a categorical variable matches a specified value. For example:
TARGET = 1
SAS Data Maker generates synthetic observations that satisfy the condition while preserving the relationships found within the relevant subset of the source data.
Numerical, date, and time ranges
Generate synthetic data within a range specified for interval variables. For example:
Income between $75,000 and $100,000
January 1, 2025 through March 31, 2025
This enables users to focus on specific segments or business scenarios while maintaining realistic relationships across variables.
Multiple conditions
Users can also combine conditions using logical operators. For example:
TARGET = 1 AND Gender = 'M'
This allows for more precise scenario creation and supports increasingly specific analytical requirements.
Preserving relationships across data
Data rarely exists in a single table. That's why SAS Data Maker is designed to maintain consistency across related tables connected through primary key and foreign key relationships.
This means conditions applied to one table can propagate appropriately across related tables, helping maintain referential integrity throughout generated datasets.
Designed for reuse
One of the key design principles is the separation of training and generation. A generator is trained once.
After that, users can generate multiple synthetic datasets using different conditions without retraining. This reduces processing overhead and makes experimentation both easier and faster, allowing teams to explore multiple scenarios from the same trained generator.
Use cases
Rare events
Create synthetic data focused on low-frequency occurrences to support modeling, testing, and analysis.
Underrepresented segments
Generate data for specific segments that can support more balanced testing, development, and analysis.
Scenario creation and testing
Create datasets to target scenarios without extensive post-processing.
Looking ahead
Conditional generation represents a crucial step for making synthetic data more flexible, targeted, and useful within real-world analytical workflows.
The first implementation focuses on generation-time conditions, allowing users to apply filters after training and reuse generators across many business scenarios. As customer feedback continues to shape the roadmap, additional conditional generation capabilities may be introduced in future releases.
Interested in trying SAS Data Maker and conditional generation?
Explore SAS Data Maker
... View more
SAS Innovate 2026 · Product Roadmap Recap
Data Maker: What's New and What's Next
Sundaresh Sankaran | Edie Moyers
Data is at the center of every AI initiative, and yet for most organizations, getting to the right data is anything but straightforward. Controlled and restricted access to data, regulatory constraints, imbalanced datasets, the cost of collecting rare events, the risks of sharing sensitive information across teams or consortium partners and these aren't edge cases. They're daily realities that slow down analytical work before it even begins.
SAS Data Maker was built to work around these constraints through high-quality synthetic data generation: data that retains the statistical characteristics of your original data without carrying its privacy risks or access complications. This post covers what it does today, what's coming next and where the product is heading as part of the broader SAS analytics ecosystem.
The Vision
Governed, Accessible Synthetic Data Generation
The vision for SAS Data Maker was to create a standalone offering built on the principle that synthetic data generation should be governed, accessible, and deeply integrated into how analytical teams actually work, not treated as a one-off workaround.
Data Maker follows a split-plane architecture: the application runs in your own Azure tenant, keeping your data secure, while SAS handles updates and administration. There is no SAS license required to get started, and you pay only when you generate output that you find useful. The intent is to lower the barrier to entry as much as possible and let the value speak for itself.
The broader direction is toward making synthetic data a natural, governed part of the analytics lifecycle available at the right moment in the right form, with clear separation of duties, and without requiring data to move unnecessarily across environments.
The Now
Data Maker Capabilities Available Today
Project-Based Synthetic Data Generation
Data Maker is organized around projects, each scoped to a specific business domain. You bring in your data across a range of supported formats: Parquet, SAS datasets, CSVs, zipped CSVs, and the platform automatically analyzes it, identifies relationships between entities, and draws an entity relationship diagram on the fly. You remain in control of key data configuration decisions; Data Maker surfaces the analysis but doesn't make consequential choices for you.
Training a Data Generator
Once your data is configured, you train a generator: a compiled, executable program that can be called repeatedly on demand to produce synthetic data. SAS Data Maker offers a range of generation algorithms with configurable settings, including differential privacy, which introduces controlled randomness into the training process to further reduce disclosure risk. The result is a reusable asset, not a one-time output.
Evaluation Metrics
After training, you receive a set of metrics that assess two dimensions simultaneously: how similar the synthetic data is to the original, and how much information about the original data it leaks. Finding the right balance between those two dimensions depends on your intended use. Volume testing tolerates lower similarity; model training demands higher fidelity. Data Maker gives you the visibility to make that call.
One-Click Data Generation
Once a generator is trained, consumers of the data can generate what they need through a simple interface: specify the number of observations, click generate, and the output lands in a location ready for downstream use for modeling, validation, testing, or sharing across teams.
The Next
What's on the Roadmap
All of the following directions have emerged from customer feedback gathered through private previews and one on one conversations. These are themes, not final feature specifications, and they reflect where the product is actively heading.
Deeper Integration with the Analytics Lifecycle
Right now, Data Maker operates somewhat independently from the rest of the analytical workflow. The near-term direction is to change that by making synthetic data generation available as a natural step at each phase of the analytics lifecycle, rather than a separate activity you must retrace your steps to access.
Concretely, this means connecting Data Maker to upstream data profiling and information governance tools so that projects can arrive pre-populated with data profiles and quality assessments, rather than requiring users to rediscover that information inside Data Maker. It also means making synthetic data easily accessible during the modeling phase, both for augmenting training data when volumes are thin and more importantly for generating holdout and validation samples without compromising the original training data. Many organizations find themselves in a bind where all available data goes to training because more training data means better models, leaving nothing clean for validation. Synthetic holdout data addresses this directly. The same logic extends to testing decision flows, business rules, and complex segmented scenarios downstream.
Conditional Generation
Conditional generation gives you control over the shape of the data you produce, rather than always generating data that mirrors the distribution of the original. This can mean boosting underrepresented classes, generating data for specific scenarios or segments, or creating simulation data for hypothesis testing. Conditional generation is expected to be available within the next few months, packaged as optional controls at both the training and generation stages.
Separation of Duties and a Generator Marketplace
This is one of the more significant architectural directions for enterprise customers. Today, the same person or team that has access to original data also controls generation and distribution of synthetic data: a scope of access that creates both compliance concerns and data leakage risk.
The roadmap addresses this by separating the generator persona (those with access to original data who train and publish generators) from the consumer persona (those who need synthetic data for analytical work but should never touch original data). The mechanism is an internal generator marketplace: after training a generator, the data owner publishes it to a shared catalog where consumers can browse available generators, understand what data they were trained on, understand what use cases they address, and generate data from them without ever being exposed to the underlying original data.
This also enables generators to be moved to lower environments (dev, test, staging) rather than moving synthetic data itself, which reduces data movement costs and closes a security loophole that exists when data generated in a production environment is transferred to a development environment.
API Access and Programmatic Integration
As Data Maker moves toward deeper integration with the analytics lifecycle, a clean, standardized API becomes essential, one that other analytical applications can consume without needing bespoke integrations. The direction is toward a consumable API paired with Python notebooks and other accessible interfaces, maintaining the no-code experience for those who prefer it while opening up programmatic access for teams that work that way. This also enables practical capabilities like scheduling generation jobs.
Additional Data Quality Controls
Several smaller but meaningful improvements are on the roadmap, including a provenance variable that tags each row of synthetic data as synthetic vs. original, making it easy to segment results when combining datasets later. Also in scope are user-defined rules to eliminate nonsensical synthetic records. For example, filtering out male patients with a pregnancy indicator, or loan applications with negative amounts which arise naturally from the probabilistic nature of synthetic generation.
Early Data Understanding with DuckDB-Inspired Approaches
Generating high-quality synthetic data requires a thorough understanding of the original data, and that analysis currently requires working with the full dataset. The roadmap includes an approach inspired by DuckDB's philosophy of generating metadata and statistical understanding early and efficiently, so that Data Maker can reach a solid understanding of data structure and characteristics with less raw data processing upfront.
The Future
Longer-Term Direction
Two themes surfaced in customer conversations that are under active consideration for the longer term.
On-Premise Deployment
On-premise deployment is the most frequently requested item from customers with strict data residency requirements. The current split-plane architecture is tied to cloud environments, and extending it to on-premise or OpenShift-style clustered environments is on the radar, though not yet on a committed timeline. If this is a requirement for your organization, the team wants to hear from you.
Deeper Viya Integration
Deeper Viya integration is directionally clear, but the path is being considered carefully: rather than building a tight integration only with Viya, the preference is to first establish an open, standardized API that any analytical application can consume, then layer in tighter Viya integrations where they add the most value. This keeps Data Maker useful as standalone and as an integrated component.
Get Started
Where to Go From Here
Try it today. SAS Data Maker is available on the Azure Marketplace right now and you pay only when you generate output you find useful. You can also scan the QR code to learn more about Data Maker.
Try before you buy through Managed Trials. SAS offers a Managed Trials program where you can bring your own data or use industry-aligned sample data in a Data Maker environment hosted by SAS, explore the product for a limited period at no cost, and then decide whether to install it yourself. This is a low-commitment way to evaluate the product against your actual use cases.
We'll continue posting updates here in SAS Support Communities as new capabilities make their way into the product. Customer feedback is actively shaping what gets built, so if you have thoughts, bring them to the comments below.
Watch the Full Roadmap Session
Catch the complete SAS Innovate 2026 Data Maker product roadmap presentation on SAS Support Communities.
Watch on Communities →
#SASDataMaker #SyntheticData #DataPrivacy #DataGovernance #SASViya #AI #MachineLearning #SASInnovate
... View more
Team Name InTheBud Track Health care & Life Sciences Use Case A clinical intelligence tool that transcribes ambient EMS radio reports into structured patient snapshots and a risk-ranked queue before arrival at the emergency department. Technology SAS Viya, SAS Visual Analytics, Node.js Region NA Team lead Ishaan Patel Team members Vrishank Ghosh, Dhananjay Raghu, Mayank Vachanni, Anand Chunduri (updating @'s once available) Social media handles [email protected], [email protected], [email protected], [email protected], [email protected] Is your team interested in participating in an interview? N Optional: Expand on your technology expertise AI/ML integration workflows, backend development, data analytics, quantitative risk
... View more