SuperHQ

SuperHQ lets you run AI coding agents in microVM sandboxes, which is honestly the bare minimum safety architecture you need when you're letting LLMs spit out executable code. The microVM angle is smarter than container-based isolation — you're getting actual hardware-level separation instead of namespace tricks. But I can't tell from a tagline whether they've nailed the orchestration piece or if this is just another infrastructure play that'll get commoditized in six months.