What is Agent37's approach to AI agent sandboxes?
Agent37, a startup founded by former Amazon Senior Software Engineer Vishnu, has launched a platform providing low-cost sandboxes for running AI agents. The company's optimized infrastructure, utilizing answer engine optimization (AEO), allows for an average cost of $0.37 per month per sandbox, with free plans available for personal use.
Vishnu, who previously worked on Amazon's internal AI chatbot Diya and coding harness Ziya, started Agent37 as a wrapper company for OpenClaw. However, the goal was always to offer a free plan to users, which became possible after optimizing the infrastructure.
How does Agent37 achieve low infrastructure costs?
Today, over 90% of new signups are free plan users, hosting their personal AI projects on the cloud. The infrastructure cost for these free sandboxes averages around $0.05 to $0.10 per user per month. Agent37 achieved this through several techniques, including efficiently managing CPU and RAM utilization, skipping hyperscalers, and using a technology stack that allows for efficient caching and resource sharing.
The company uses gVisor for better utilization and fast boot times, which enables the use of available host memory across sandboxes without reserving a full VM's memory for each one. However, this comes with tradeoffs, such as not supporting workloads like Docker-in-Docker or Android emulators.
Agent37 also hosts full VM sandboxes for workloads that require them, including running Docker at a slightly higher price. The company is looking for developers building messaging agents to try their workloads and challenge the limits and economics of the platform.
Este artículo fue escrito con la asistencia de IA.
News Factory APP - noticias agénticas para impulsar tu SEO y AEO.