Enterprise AI is entering a more difficult phase. The question is no longer simply whether companies can build generative AI applications, but whether they can trust those systems enough to deploy them broadly. SAS and Amazon Web Services are putting that challenge at the center of an upcoming AI Enterprise Conference panel, as enterprises move from AI experimentation toward production deployments and increasingly autonomous agentic systems.
The enterprise AI market is moving into a new phase where technical capability is no longer the only obstacle to adoption.
Companies have access to increasingly capable large language models, cloud AI infrastructure and development platforms. What remains harder is establishing enough confidence in those systems to let them influence operational decisions, interact with business data and eventually act with greater autonomy.
That is the issue SAS and Amazon Web Services plan to address at the AI Enterprise Conference in New York on September 1, where representatives from the two companies will participate in a panel focused on implementing trusted AI systems at scale.
The discussion coincides with the second annual SAS Data and AI Impact Report, supported by IDC, which SAS says will examine the relationship between trust and returns on AI investment, enterprise AI adoption and the emerging use of agentic AI.
The timing is significant. Organizations are increasingly moving beyond proof-of-concept projects, but scaling AI introduces problems that are difficult to solve through model performance alone: data quality, governance, security, explainability, accountability and operational controls.
AI governance is becoming an operating problem
For several years, enterprise AI governance was often discussed as a policy exercise. Organizations created responsible-AI principles, drafted usage guidelines and established review committees.
Production AI changes the equation.
A model that generates internal marketing copy presents a different risk profile from an AI system that evaluates loan applications, assists clinicians or makes recommendations inside a financial-services workflow. An autonomous agent that can call APIs or modify records introduces another layer of risk because the system can potentially take actions rather than simply generate information.
SAS is framing this transition as a move from AI experimentation toward production deployment.
“What we’re seeing now is the struggle to trust AI enough to meaningfully scale with it,” Diana Rothfuss, global product marketing director for financial services at SAS, said in the announcement.
That trust cannot be created by a single security feature. It depends on the infrastructure surrounding the model: reliable data, access controls, monitoring, governance processes and clear accountability when an AI system produces an incorrect or harmful outcome.
Agentic AI raises the stakes
The focus on agentic AI is particularly important.
Generative AI applications typically respond to prompts. AI agents can go further by planning tasks, calling external tools, retrieving information and taking actions based on objectives.
That makes agentic systems potentially more useful—and potentially more difficult to govern.
Enterprise teams deploying agents need to know what information an agent can access, what actions it is authorized to perform, which tools it can invoke and how its decisions can be audited. Organizations also need mechanisms to stop or constrain an agent when its behavior falls outside predefined boundaries.
This creates a convergence between AI development and conventional enterprise security architecture.
Microsoft, Google, Amazon and NVIDIA are all building increasingly sophisticated AI infrastructure, while enterprise software companies such as Salesforce and Adobe are embedding AI agents and copilots directly into business applications. As AI becomes a layer inside existing software rather than a standalone experiment, governance increasingly needs to follow the technology into those workflows.
SAS and AWS are building an ecosystem around the problem
The SAS-AWS relationship gives the announcement a broader infrastructure dimension.
SAS is an official member of the AWS Partner Network and entered a strategic collaboration agreement with AWS in 2023. The companies have been integrating SAS data and AI software with AWS technologies and making several SAS offerings available through AWS Marketplace.
Those products include SAS Viya, the company’s cloud-native data and AI platform; SAS Viya Workbench, an on-demand development environment; SAS Customer Intelligence 360; and SAS Hosted Managed Services.
The significance is less about individual product availability than the direction of enterprise AI architecture. Companies increasingly want AI development, data management, analytics and deployment infrastructure to operate within the same cloud environment as their existing workloads.
For enterprise technology teams, that can simplify deployment and data access. It can also make governance more centralized if security and policy controls are integrated into the underlying platform.
But cloud integration does not automatically create trustworthy AI.
Enterprises still need to determine whether models are appropriate for specific use cases, whether training and operational data are reliable, whether outputs can be audited and how humans remain accountable for AI-assisted decisions.
Trust could become an AI purchasing criterion
The next stage of enterprise AI competition may therefore involve less emphasis on raw model capability and more on operational trust.
IDC’s involvement in the SAS research is notable because enterprises are increasingly looking for measurable evidence that AI investments produce business value. Gartner has similarly identified AI governance and risk management as increasingly important as organizations move AI into business-critical workflows.
The market is consequently developing around several layers: model infrastructure from hyperscalers and AI companies, data and analytics platforms from vendors such as SAS, application-level AI from enterprise software providers, and security and governance technologies that attempt to connect them.
The winning architecture may not be the one with the most sophisticated model. It may be the one that gives enterprises sufficient visibility and control to use advanced models safely at scale.
That becomes even more important as agentic AI moves into areas such as customer service, financial analysis, supply-chain operations and software development.
For CIOs, chief data officers and AI leaders, the practical challenge is turning “trustworthy AI” from a corporate principle into an engineering discipline. That means defining who owns an AI system, what data it can use, what actions it can take, how its performance is monitored and what happens when it fails.
SAS and AWS are positioning their partnership around that transition. The larger market question is whether enterprise AI can move from impressive demonstrations to dependable infrastructure.
The answer will likely depend as much on governance and operational design as on the next generation of models.
Market Landscape
Enterprise AI platforms are converging around several technology layers:
- AI development platforms: SAS Viya, Microsoft Azure AI, Google Cloud Vertex AI and AWS AI services provide environments for developing and deploying enterprise models.
- Cloud AI infrastructure: AWS, Microsoft Azure and Google Cloud provide the compute, storage and managed AI services required to operate large-scale workloads.
- Generative and agentic AI: LLMs are increasingly being embedded into enterprise applications, while agentic systems introduce tool use and autonomous workflows.
- AI governance: Enterprises are developing controls around data lineage, model monitoring, security, explainability, compliance and human oversight.
- AI-powered applications: Salesforce, Adobe and other enterprise software vendors are incorporating AI directly into CRM, marketing, content and workflow products.
The competitive differentiator is shifting toward integration. Enterprises increasingly need AI systems that connect cleanly to existing data estates, cloud infrastructure, identity systems and business applications while maintaining auditable controls.
Top Insights
- SAS and AWS will focus on trusted AI at the AI Enterprise Conference as organizations shift from experimentation toward production-scale generative and agentic AI deployments.
- SAS’s latest IDC-supported research will examine AI investment returns, adoption trends and agentic AI, highlighting trust as an enterprise scaling challenge.
- SAS Viya and related offerings on AWS provide a cloud-based foundation for enterprise data science, AI development, customer intelligence and managed services.
- Agentic AI increases governance requirements because autonomous systems can access data, invoke tools and take actions beyond simply generating model responses.
- Enterprise AI buyers are increasingly evaluating governance, monitoring, security and accountability alongside model performance when selecting production AI infrastructure.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
