• Inference Systems

Sail is building the compute stack for background agents, from silicon to API. We'll discuss the unique infra challenges of agents that run for hours without human intervention, and why Sail is building a new stack here. The focus will start on token processing, and what tradeoffs we can make between throughput and latency. Then we'll move to the sandbox world, and explore our options to give every agent a cloud computer. Finally, we'll close with some examples of what customers do with massively parallel baclground agents on Sail.