AWS is introducing a new feature in Amazon WorkSpaces that enables AI agents to run desktop applications independently. With this, the company is targeting organizations that are still heavily reliant on legacy software and applications without modern APIs.
With this expansion, AI agents can operate within the same virtual desktop environments that organizations already use for employees. According to AWS, this enables automating existing business processes without first modernizing applications or migrating them to new platforms.
Many large organizations still run business-critical processes on older software environments. According to figures AWS cites from Gartner research, a large portion of those applications lack programmable interfaces. This makes them difficult for modern AI systems and automation to access.
Instead of direct API integrations, AWS has agents operate through the application’s user interface. The AI agent gains access to a managed WorkSpaces desktop and can click, type, scroll, and analyze screen information via screenshots. In practice, the agent functions similarly to a human user at a desktop.
The solution leverages existing AWS services for authentication and logging. AI agents authenticate via AWS Identity and Access Management, while activities are logged via AWS CloudTrail and Amazon CloudWatch. Because agents operate within the existing WorkSpaces environment, security and compliance settings remain intact.
In addition, the service supports the Model Context Protocol, an open standard for AI agent integrations. This allows organizations to use a variety of agent frameworks, including LangChain, CrewAI, and Strands Agents.
AWS demonstrated the functionality with an example in which an AI agent independently processes a repeat prescription within a pharmacy application without API integrations. The agent retrieves patient data, selects medication, and completes the order entirely via the desktop interface. According to AWS, no modifications to the application were required.
Criticism of AI agents’ efficiency
However, the approach also has drawbacks, reports The Register. Research by AI company Reflex suggests that so-called computer-use agents can be significantly more expensive and less efficient than traditional API integrations. Because agents operate via screenshots and computer vision, many more AI operations are required to perform simple tasks. According to Reflex, a simple interaction with a user interface can already require hundreds of thousands of tokens. The company concludes that AI agents operating via desktop interfaces, therefore, remain structurally more expensive than software that communicates directly via APIs.
On the other hand, AWS specifically aims to help organizations that lack modern APIs. The WorkSpaces approach thus effectively serves as an alternative to costly legacy software modernization projects. Instead of modifying applications, AWS allows agents to operate through the same user interface that employees already use.
AWS also emphasizes the security benefits of isolated virtual desktop environments. AI agents run within separate WorkSpaces instances and therefore do not need direct access to local endpoints or internal corporate networks. Because WorkSpaces can be temporarily launched and shut down afterward, the infrastructure is also well-suited for short-term agent workloads.
AWS is not alone in this development. Microsoft is also working on support for AI agents within Windows 365 environments. This creates a new category of cloud desktop services in which AI systems not only call applications via APIs but actually operate software through the user interface.