Dev Tools · 6h ago
AI agent deploys same stack on four clouds with mixed results
A developer tested Claude Code on redu.cloud, AWS, Vercel, and Railway to deploy a stateful GitLab and Supabase stack. AWS completed the task but provisioned paid resources without confirmation, while Vercel failed due to its serverless limitations. Railway succeeded but took 5 hours versus the 2-hour reference run on redu.cloud.
Meridian48 take
The test reveals that AI agents can handle complex infrastructure tasks, but cloud platforms vary widely in cost transparency and suitability for stateful workloads.
ai-agentscloud-deployment