SAS AI Navigator Lands on Microsoft Marketplace: Entra ID Requirement and 6-Month Free Offer

What launched, and when​

SAS announced on October 5, 2026 that AI Navigator is available to help AI, data, compliance and risk leaders "bring order to AI chaos." The timeline is slightly messier than a single press release suggests. SAS first unveiled it on April 28, 2026, at SAS Innovate in Dallas. A SAS community post says it was officially available on Microsoft Marketplace as of September 29, 2026. So the October announcement looks like the formal follow-through on a Q3 availability promise.

It is also not a Microsoft product. This is a third-party product sold through Microsoft's storefront, not a Microsoft service. Microsoft's role is as the sales channel, and Entra ID is a stated technical requirement.

What it does​

The pitch is centralised visibility. The Marketplace listing says it gives a unified view of AI across an organisation by centralising information about models, agents and use cases, whether built in house or sourced from third parties. SAS says organisations don't have to change how they build AI. The tool offers a view of the models and tools they already use, including LLMs, AI agents, and open-source or SAS models.

Governance is organised around use cases rather than individual models. SAS says this is where business impact occurs. Governance extends to the models and agents that power each use case, and to the policies applied to them. SAS's own example is a customer-facing chatbot. A company could govern the underlying agent or model, such as Claude or Microsoft Copilot, and apply policies to check that the use case meets regulatory expectations.

Other capabilities described by SAS:

• Lifecycle coverage: It supports the journey from experimentation through deployment to retirement, for assets built in-house or bought from third parties.
• Policy frameworks: The listing says you can work with any policy, standard or regulation, and names the EU AI Act, the NIST AI Risk Management Framework and internal policies such as a generative AI policy. You can create your own policies or buy them from a partner.
• Policy packs: SAS partners such as Protiviti and Asenion offer ready-made sets of compliance questions and risk classifications. These let you skip writing a policy from scratch. SAS has not published pack pricing or coverage details in the material I reviewed.
• Workflow tooling: A SAS quick-start post describes asset inventory, policy assessments, approval tracking, dashboards and alerts, and user roles and permissions. That is SAS enablement material, not independent testing.

A SAS roadmap post from before launch said the product stores metadata rather than copying models or datasets. It also said no on-premises option was planned at that point. Treat both as roadmap-era statements to confirm in current documentation. The same post warned that the roadmap can change and that buyers should decide based on existing functionality.

What admins need to know before buying​

Here is what the listing says about buying and deploying the product.

1. Entra ID is mandatory. The listing says organisations must use Microsoft Entra ID for identity management to implement SAS AI Navigator.
2. It's standalone. It operates independently and does not require any other SAS software.
3. Use a work account. A SAS community thread shows what happens otherwise. A reply there says purchases with a personal Microsoft account aren't supported, and the product must be bought with an organisational Entra ID (work) account. The same reply says the admin email entered during "Configure account" must use a domain belonging to that same organisation. In that thread, a buyer who used a personal account had a subscription stuck in a pending state with no invitation email. This is a single user's report with a SAS reply, not a documented failure rate. It is still a useful warning for anyone testing from a personal tenant.
4. Activation is by email. SAS sends an email shortly after purchase to activate the product.
5. The free period ends. The first six months are free, then pay-as-you-go pricing begins. The listing gives no rate in the material I reviewed. Ask for one before you begin an inventory you'll depend on.

The Marketplace page lists financial services, health care and life sciences, government and public sector, telecommunications, and manufacturing among supported industries.

The numbers behind the pitch​

SAS ties the launch to its 2026 Data and AI Impact Report: The New Economics of Trust, produced with research insights from IDC. Its headline claim is that organisations with trustworthy AI practices are 15 times more likely to report strong ROI. The details matter here:

• The figure is based on a global survey of 2,699 decision-makers across 28 countries, according to SAS's September report announcement.
• The 15x comes from a comparison of 62% versus 4% reporting strong or high ROI.
• "Leaders" are organisations that scored an average of 80 or higher out of 100 across five dimensions: data quality and governance, model governance and oversight, explainability and fairness, responsible AI policy, and audit and accountability.
• SAS also reports that 97.2% of users override AI recommendations at least sometimes. The top reason is that the AI couldn't explain its decision.
• SAS reports trust falling from 76% for generative AI to 66% for agentic AI.

These are vendor-published survey results. They show correlation among organisations with mature practices. They don't show that buying this product raises ROI or cuts override rates. A tool that inventories AI also doesn't make a model explainable. It records and organises, and the quality of the governance still depends on the people filling it in.

Early users and analyst view​

Erasmus University Medical Center in the Netherlands was an early tester. SAS says it runs the Responsible Ethical AI in Health Care Lab with SAS and Delft University of Technology. Dr. Michel van Genderen, an ICU and internal medicine physician and associate professor, called it a promising foundation for centralising AI governance. That is an early-tester endorsement, not proof of a finished deployment or measured outcomes.

IDC's Kathy Lange describes the problem as AI sprawl: models, agents and third-party tools proliferating across the enterprise. That is her assessment of the category, not a measured result for this product.

A practical evaluation checklist​

If you're an IT or compliance lead considering this, or any AI inventory tool, these questions matter more than the marketing:

• Scope: Which use cases, models and agents need tracking, including shadow AI and vendor-embedded Copilot features?
• Inventory method: How do assets get in? The roadmap mentioned future "connections" to reduce manual transfers, so ask what's available today.
• Policies: Which regulations apply to you, and do the Protiviti or Asenion packs cover them? Ask for samples.
• Identity: Is Entra ID acceptable for your tenant, and who will own the admin account?
• Cost: What does pay-as-you-go cost once the six free months are up?
• Assurance: What certifications exist? SAS's roadmap said compliance certifications were being pursued and that many wouldn't be in place at general availability.

Bottom line​

The news is real but modest. A third-party AI governance SaaS is now purchasable through Microsoft Marketplace, needs Entra ID, and is free for six months before metered billing. For Microsoft-centric shops juggling Copilot, Azure AI and other models, a central register of AI use is a reasonable idea. The compliance value depends on how well your organisation maintains it, and the six-month trial is the cheapest way to find out.