Novita Sandbox delivers system-level isolation with sub-200ms startup, enabling safe and scalable deployment of autonomous agent systems.

SAN FRANCISCO, April 28, 2026 /PRNewswire/ — As AI rapidly shifts from copilots to autonomous systems, infrastructure requirements are entering a new phase. Today’s agents don’t only assist. They execute code, call APIs, browse the web, and interact with live environments. That evolution turns security from a secondary concern into a foundational requirement.

Frameworks such as OpenClaw and Hermes Agent are accelerating the adoption of persistent, tool-using, self-improving agents. But this new capability comes with a hidden cost: these systems often inherit access to sensitive environments, creating systemic risk surfaces that didn’t previously exist.

In practice, exposure can include the following:
• Developer machines and local files
• Credentials, API keys, and cloud configurations
• Active browser sessions and authenticated contexts

These risks make standard vulnerabilities such as prompt injection, malicious skills, and unsafe execution paths significantly more dangerous.

Novita Sandbox is designed to eliminate this class of risk through strict runtime isolation. Each agent task runs inside a dedicated Firecracker microVM with its own kernel, isolated memory boundary, and ephemeral filesystem. By enforcing separation at the system level, it ensures that execution environments remain fully isolated with no shared process space, no credential leakage, and no cross-agent interference.

Crucially, this level of security does not come at the expense of performance.

Novita Sandbox delivers:
• Sub-200ms average startup time
• High-density scaling to thousands of parallel microVMs
• Consistent latency under concurrent workloads

This performance makes it suitable for real-time and production-scale agent systems, where both responsiveness and isolation are required.

For long-running workflows, Novita Sandbox also introduces stateful execution. Environments can be paused and resumed on demand with full runtime state (includes memory, filesystem, and active processes). All of this can be restored in approximately one second without compromising isolation guarantees.

“With this launch, Novita introduces a new category: Secure Agent Infrastructure. A trusted execution layer for deploying autonomous systems safely at scale without exposing local machines or sensitive runtime environments,” said Junyu Huang, COO of Novita AI.

Novita Sandbox is available immediately through managed NovitaClaw deployments, Sandbox Skills for OpenClaw, Hermes-compatible runtimes, and persistent node configurations for long-running workflows. These implementations provide a flexible path from secure prototyping to enterprise-scale automation.

As agents become the next interface for software, security is no longer optional. Novita AI is building the secure runtime layer that makes the emerging agent economy possible. For more information, visit novita.ai.

Novita AI is an AI and agent cloud platform helping developers and startups build, deploy, and scale models and agentic applications with high performance, reliability, and cost efficiency. The platform delivers fast inference across 120+ LLMs and multimodal models through a single API, alongside GPU Instances, Bare Metal, and Agent Sandbox infrastructure built for production AI.

View original content to download multimedia:https://www.prnewswire.com/news-releases/novita-ai-launches-sandbox-to-secure-openclaw-hermes-agent-and-autonomous-systems-302755870.html