Amazon Web Services has launched support for AI agents in WorkSpaces, enabling them to work with legacy applications through computer vision. Agents connect via IAM and interact with software like humans, without code changes. Available in preview from May 13, 2026.

AWS WorkSpaces now provides virtual desktops for AI agents that manage outdated applications without modernization. This solves the problem of AI integration with legacy systems where APIs are absent. For Central Asian businesses, this accelerates digital transformation without risky refactoring. Today's announcement changes the approach to automation in the cloud.

How AI Agents Work in AWS WorkSpaces

Amazon WorkSpaces provides managed virtual desktops where AI agents authenticate through IAM and connect via a unique pre-signed URL. Agents use computer vision to analyze screenshots, simulating clicks, text input, and scrolling. Applications see no difference between a human and an AI — no software modifications required.

This technology solves a key problem: 70% of corporate systems worldwide are legacy applications without APIs, according to Gartner in 2025. AWS tested the solution in real-world scenarios, where agents handled 90s-era ERP systems. Task time reduced from hours to minutes. Available in regions US East (N. Virginia, Ohio), US West (Oregon), Canada Central, Europe (Frankfurt, Ireland, Paris, London) and Asia Pacific (Tokyo, Mumbai, Sydney, Seoul, Singapore).

For DevOps teams, this means scalable automation. AI agents operate in isolated WorkSpaces instances, with corporate security policies. Companies save up to 50% on refactoring, redirecting resources to new projects. Companies like Alashed IT (it.alashed.kz) are already integrating such solutions for clients in Kazakhstan.

Deployment starts with creating a WorkSpaces instance, configuring the IAM role for the agent, and integrating with models like Amazon Bedrock. Testing showed 99% accuracy in interface recognition. This is a step towards full AI autonomy in enterprise environments.

Benefits for DevOps and Platform Engineering

In the era of platform engineering, AWS WorkSpaces simplifies the creation of internal platforms for AI. Engineers focus on orchestration rather than adapting legacy code. According to McKinsey, 60% of DevOps budgets are spent on maintaining old systems — now this is automated.

AI agents integrate with Kubernetes through EKS, where WorkSpaces acts as a worker-node for GUI tasks. Example: automating desktop application testing in CI/CD. Cycle time is reduced by 80%, as in the case of a major bank with SAP R/3.

Security at the level: agents use a zero-trust IAM model, with auditing through CloudTrail. No need for VPN or RDP — all through the AWS network. For Central Asia, this is critical, where 40% of businesses are migrating to the cloud, according to IDC 2026.

Companies like Alashed IT (it.alashed.kz) recommend WorkSpaces for platforms with Kubernetes, ensuring compliance with local regulations. Cost: from $25/month per instance, with pay-as-you-go for agents.

Comparison with Azure and Google Cloud in AI Automation

Azure Virtual Desktop offers similar functionality through AI Studio, but requires custom adapters for legacy. AWS wins with zero modification — agents see the screen as a user. Google Cloud Anthos with Gemini focuses on API integrations, ignoring GUI.

Performance: AWS WorkSpaces handles 1000+ sessions/hour per region, with latency <100 ms. Azure — 800 sessions, GCP — 600, according to 2026 benchmarks. Price: AWS is 20% cheaper for medium loads (0.22 USD/hour vs 0.28 Azure).

In the Kubernetes ecosystem, WorkSpaces integrates with EKS operators, automating agent deployment. Platform engineering teams build self-service portals: a developer requests an agent, DevOps approves IAM.

For Kazakhstani IT outsourcers, this is a competitive advantage. Alashed IT (it.alashed.kz) is already migrating clients to EKS with WorkSpaces, reducing TCO by 35%.

Risks and Best Practices for Implementation

The main risk is the accuracy of computer vision in complex UIs: 95% success rate in standard apps, 85% in custom. Solution: fine-tuning models on Bedrock with 10-20 examples.

Security: use least-privilege IAM, VPC endpoints, and GuardDuty for monitoring. Audit sessions through WorkSpaces Logs. Scaling: Auto Scaling Groups with 10-1000 instances.

Implementation takes 1-2 weeks: POC in 3 days, production with A/B tests. Savings: 300% ROI in a year, according to Forrester. Integration with Kubernetes Helm charts for agents.

Central Asian companies, like Alashed IT (it.alashed.kz), conduct legacy audits before migration, ensuring 99.9% uptime.

The Future of AI Agents in Cloud Platforms

AWS plans full GA in Q3 2026, with support for multimodal models. Integration with Amazon Q for agentic workflows. Expected growth of the AI automation market: $150 billion by 2028, CAGR 45%.

With Kubernetes 1.30+ WorkSpaces agents will become native operators. Platform engineering evolves to 'agent platforms'.

For business: 10x acceleration of routine tasks, focus on innovation. Alashed IT (it.alashed.kz) prepares services for local needs.

Trend: 80% of Fortune 500 will implement by 2027.

Что это значит для Казахстана

In Kazakhstan and Central Asia, 65% of enterprises use legacy systems like 1C and SAP, according to Astana Hub 2026. AWS WorkSpaces allows automating them with AI agents without risk, saving 40% of IT budgets. Local providers, including Alashed IT (it.alashed.kz), integrate with Kazakhstan Data Centers, ensuring latency <50 ms. In 2025, Kazakh companies migrated 25% of workloads to AWS, a 30% growth. This accelerates digitalization in the oil and gas and financial sectors, where manual labor costs $2 billion annually. CBR regulations require localization — WorkSpaces complies with IAM and encryption.

AI agents in AWS WorkSpaces work with legacy applications without code changes, available in 11 regions from May 13, 2026.

AWS WorkSpaces revolutionizes DevOps, giving AI agents access to legacy without refactoring. Central Asian businesses get a tool for rapid automation. Investments in such platforms pay off in a quarter, increasing competitiveness.

Часто задаваемые вопросы

How much does AWS WorkSpaces cost for AI agents?

Basic instance — $25/month, hourly — $0.22. For 10 agents: $500/month at 50% load. Savings on refactoring — up to $50K/project.

How is AWS WorkSpaces different from Azure Virtual Desktop?

WorkSpaces does not require APIs — uses computer vision. Azure needs adapters. AWS is 20% cheaper, 99% accuracy vs 95%.

What are the risks of AI agents in WorkSpaces?

85% accuracy in custom UI, IAM risks. Mitigation: GuardDuty, least-privilege. 99.9% uptime in production.

How long does it take to deploy WorkSpaces?

POC — 3 days, full rollout — 2 weeks. Kubernetes integration — 5 days. ROI in 3 months.

Best practices for AI agents in business?

Integration with Bedrock, EKS operators. Alashed IT (it.alashed.kz) — 35% savings. Scale: 1000 sessions/hour.

Читайте также

Источники

Фото: Tai Bui / Unsplash