Bridging the Legacy Gap: Amazon WorkSpaces Now Enables AI Agents to Operate Desktop Apps

For years, enterprises have faced a frustrating paradox: they possess powerful AI agents capable of complex reasoning, but those agents are effectively locked out of the incredibly software that runs the business. Most critical business workflows still rely on legacy desktop applications, mainframes, and proprietary tools that lack modern APIs. This “last-mile challenge” has forced companies to choose between stalling their AI adoption or investing in risky, expensive modernization projects.

AWS is solving this problem by turning the virtual desktop into an AI interface. Amazon WorkSpaces now allows AI agents to securely access and operate desktop applications exactly as a human would—by pointing, clicking, and navigating the user interface—without requiring a single line of code to be changed in the underlying legacy software.

The Legacy Application Bottleneck

The scale of the legacy software problem is immense. According to a 2024 Gartner report, 75% of organizations operate legacy applications that lack modern APIs. Even more striking, 71% of Fortune 500 companies rely on mainframe systems for critical processes that offer no adequate programmatic access.

When an AI agent cannot “talk” to a system via an API, it cannot automate the workflow. Until now, the only solution was to build a custom API or migrate the entire system to a modern cloud architecture. By allowing agents to operate within a managed WorkSpaces environment, Amazon is removing the need for application migrations or new infrastructure management.

How AI Agents Interact with the Desktop

Rather than attempting to integrate with the backend code of an application, AI agents in WorkSpaces use a combination of computer vision and input simulation to interact with the frontend.

Core Capabilities

• Computer Input: Agents can click, type, and scroll within the desktop environment to navigate menus and enter data.
• Computer Vision: Agents capture screenshots of the desktop, allowing them to “see” the application’s state and UI elements.
• Model Context Protocol (MCP): WorkSpaces supports the industry-standard MCP, ensuring compatibility with various agent frameworks such as LangChain, CrewAI, and Strands Agents.

Enterprise-Grade Governance

Security is often the primary concern when granting AI agents access to sensitive business systems. Amazon has integrated these agents into the existing AWS security ecosystem:

• Authentication: Agents authenticate through AWS Identity and Access Management (IAM).
• Observability: Full audit trails are available via AWS CloudTrail and Amazon CloudWatch, providing visibility into exactly what the agent did.
• Isolation: Because agents operate within secure, managed WorkSpaces environments rather than on local machines, existing compliance policies remain intact.

“WorkSpaces lets our clients give AI agents the same secure, governed desktop environment their employees already use — no custom API integrations, full audit trails, and enterprise-grade isolation out of the box. For regulated industries, that’s not a nice-to-have — it’s the baseline.”
— Chris Noon, Director, Nuvens Consulting

Real-World Application: From Pharmacy to Finance

The ability to automate “API-less” software opens the door for high-impact automation in regulated industries. For example, an agent built with the Strands Agent SDK and Amazon Bedrock can handle a prescription refill by looking up patient records, searching for medication, and placing orders within a sample pharmacy system—all without the software knowing an agent is driving the interface.

Similar workflows can be applied to:

• Financial Services: Trade settlement and back-office operations.
• Healthcare: Claims processing and patient data management.
• HR: Candidate screening and onboarding via proprietary legacy tools.

Availability and Deployment

This feature is currently available in public preview at no additional cost. It is deployed across several global regions, including:

• North America: US East (N. Virginia, Ohio), US West (Oregon), and Canada (Central).
• Europe: Frankfurt, Ireland, Paris, and London.
• Asia: Tokyo, Mumbai, Sydney, Seoul, and Singapore.

The Path Forward

The introduction of AI agent capabilities into Amazon WorkSpaces signals a shift in how enterprises approach digital transformation. Instead of the “rip and replace” strategy that often leads to costly failures, organizations can now layer intelligence on top of their existing investments. As AI agents move from simple chat interfaces to active operators of business software, the boundary between human-led and AI-augmented workflows will continue to blur, driving a new era of enterprise productivity.