Amazon Web Services launched a public preview that turns its managed virtual desktop service into agent infrastructure. Amazon WorkSpaces now lets AI agents authenticate via IAM, connect to desktop applications through the Model Context Protocol, and interact with legacy UIs using computer vision: clicking buttons, typing fields, scrolling pages.
The target problem is specific: according to a 2024 Gartner report, 75% of organizations run legacy applications that lack modern APIs, and 71% of Fortune 500 companies operate critical processes on mainframe systems without adequate programmatic access. WorkSpaces skips the modernization step entirely. The application doesn’t know an agent is driving it.
How It Works
Agents operate inside secure WorkSpaces environments rather than on local machines. The setup exposes three capabilities per stack:
- Computer input for clicking, typing, and scrolling
- Computer vision for capturing screenshots (how the agent “sees” the application)
- Screenshot storage for audit trails and debugging
Authentication runs through IAM. Full audit trails flow into CloudTrail and CloudWatch. Existing security controls and compliance policies carry over unchanged because agents run inside the same managed desktops employees already use.
WorkSpaces exposes a managed MCP endpoint per stack. Any agent framework that supports MCP (LangChain, CrewAI, Strands Agents, OpenClaw) can point to that endpoint and begin interacting with desktop applications installed on the fleet image.
First Customer Signal
Chris Noon, Director at Nuvens Consulting, told AWS: “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.”
Architecture Positioning
The launch puts AWS on a different side of the agent infrastructure debate than Nvidia and ServiceNow’s Project Arc (announced the same week). Arc runs agents locally on managed user desktops. WorkSpaces runs agents in cloud-managed virtual desktops. Both target legacy application access. The difference: Arc bets on local execution eliminating VPN and API intermediaries, while WorkSpaces bets on cloud isolation preserving existing compliance posture.
AWS demonstrated the system handling a prescription refill workflow inside a sample pharmacy application with no API: looking up a patient record, searching medication, placing the order, and confirming. The application was unmodified.
Availability
Public preview launched at no additional cost across 12 regions: US East (N. Virginia, Ohio), US West (Oregon), Canada (Central), Europe (Frankfurt, Ireland, Paris, London), and Asia (Tokyo, Mumbai, Sydney, Seoul, Singapore). AWS published a GitHub sample repository for getting started.