Sail Research Launches Sailboxes, the First Cloud Environment Purpose-Built for Long-Horizon AI Agents

Sailboxes pair auto-sleep and elastic resource scaling with usage-based pricing, enabling enterprise agents to run efficiently at scale.

SAN FRANCISCO, July 14, 2026 /PRNewswire/ -- Sail Research, the infrastructure company for AI agents, today launched Sailboxes, the first cloud environment purpose-built for long-horizon agent work. Sailboxes give enterprise teams full machines with persistent state and auto-sleep, enabling them to deploy long-horizon agents more cost efficiently and at greater scale.

As enterprises expand their use of AI agents, sandbox costs are becoming a significant line item. Existing sandboxing products require developers to reserve fixed CPU and memory and pay for every second of that reservation. However, multi-turn agents rarely consume their full allocation, and idle time waiting for input, boot-up, or overprovisioned capacity drives costs higher as volume grows. Capped lifetimes and container constraints further limit scalability for agents that run multiple turns over long context windows.

"Our early inference customers, saving up to 90% on tokens with Sail, immediately wanted to run many more agents, on longer tasks," said Neil Movva, CEO and co-founder of Sail Research. "We built Sailboxes to encourage exactly that: unlimited lifetimes, easy forking for parallel scaling, and efficient scheduling to deliver the most compute per dollar."

Unlike generic sandboxes optimized for fast spin-up, Sailboxes focus on endurance and efficiency at scale. Sail achieves this through a novel architecture that migrates virtual machines over Sail's large compute fleet, based on live resource usage.

"Today's sandboxes make you guess how much memory, CPU, and disk your agents need, and then either overpay or risk crashing," said Samir Menon, CTO and co-founder of Sail Research. "Sailboxes let you run agents without thinking about any of that and pay only for what you actually use, freeing you to focus on the transformative tasks your agents can do instead."

Sailboxes also offer agentic access to full machines with independent disks, persistent state, Docker support, and local NVMe, without runtime limits. Agents running in Sailboxes behave the same way in the cloud as they would locally. As part of Sail's end-to-end agent infrastructure platform, Sailboxes can auto-sleep while agents wait on the result of a Sail inference call, maximizing efficiency.

Thousands of Sailboxes are already deployed to host long-horizon agents, run reinforcement learning rollouts, and accelerate software development through coding agents, asynchronous bug finding, and automated triage. Quadrillion Labs uses Sailboxes to host cloud agents for Qualia Cloud, its autonomous research platform, where researchers can ask a Qualia agent to test a hypothesis and run multiple experiments in parallel.

Sailboxes are generally available today. Developers can get started and view pricing at sailresearch.com.

About Sail Research
Sail Research is the infrastructure company purpose-built for long-horizon AI agents. Sail's platform combines two core components: an inference stack rebuilt from the ground up for throughput and efficiency, delivering up to 10x lower cost per token; and Sailboxes, a stateful sandbox environment designed to run for days rather than seconds. Together, they make it possible for enterprises to scale their use of agents across multi-turn, long-horizon workflows with efficient economics. Sail is backed by Kleiner Perkins, Sequoia, Redpoint Ventures, Theory Ventures, Vine Ventures, CRV, A*, and Abstract Ventures.

For more information, visit sailresearch.com.

SOURCE Sail Research