We run stateful AI agents on a serverless infra (maritime.sh)

A new serverless infrastructure platform has been announced for running stateful AI agents, designed to provide each customer with their own dedicated micro-VM. This setup ensures complete isolation between customers, with no shared resources, thereby enhancing data security and privacy. The innovative architecture enables rapid provisioning, allowing developers to go from zero to a fully operational agent in about a minute with just one API call. Each micro-VM is tailored with customer-specific identifiers and lifecycle webhooks, ensuring a seamless integration for users. This advancement is significant for the AI/ML community as it addresses crucial scalability and privacy issues in deploying AI agents. The platform is built to scale effortlessly to thousands of agents, making it suitable for both small startups and larger enterprises. The flat pricing model simplifies budgeting for developers, while the system’s performance—offering 1 vCPU, 2 GB RAM, and 5 GB SSD for each agent—positions it as a versatile solution for various application needs. As developers increasingly seek robust yet flexible environments for AI deployment, this offering could reshape how AI services are delivered across industries.