Blog
Notes from the machine room.
Engineering notes, product updates and lessons from running fleets of autonomous machines.
Engineering
Snapshots as a safety net for autonomous code
Letting a model run code it wrote is only reckless if you can’t undo it. We made forking a running machine a sub‑second operation so that every risky step has a checkpoint.
8 min read
Product
Headful browsers, and why headless is not enough
Sites behave differently when no one is watching. A real display, a real GPU and a real window manager make agents dramatically more reliable on the web.
5 min read
Security
Egress filtering for agents that browse
An agent with unrestricted network access is a liability. Allow‑lists at the VM boundary keep prompt injection from turning into data exfiltration.
4 min read
Engineering
Booting a Linux VM in 400 milliseconds
Firecracker, a trimmed kernel and a lot of pre‑warming. A tour of what happens between the API call and the first shell prompt.
9 min read
Company
What we learned running 10,000 agents for a week
We stress‑tested the platform with a fleet of browsing agents. Some numbers, some surprises, and the changes we shipped because of it.
7 min read
