// blog
Notes from the browser layer.
Engineering deep dives, product launches and field reports on running agents against the real web.
Extract API v2: schema-guided agents with self-healing selectors
How we rebuilt extraction around a planning model, so your data keeps flowing after a site redesigns, with no code changes on your side.
Hana SatoHead of Research
Live view handoff: let a human take over any agent session
Approvals, 2FA and judgment calls without breaking the run.
How we cut cold starts to 184 ms with microVM snapshots
Restoring a booted Chromium from memory, safely, ten thousand times a minute.
Hello São Paulo, Mumbai and Johannesburg
Three new regions and what we learned launching them.
A practical guide to evaluating browser agents
Task suites, trace scoring and the metrics that actually predict production success.
Inside our fingerprint lab: 1,400 bot tests a day
How we keep stealth honest against every major anti-bot vendor.
Migrating from self-hosted Puppeteer in an afternoon
A step-by-step checklist from teams who've done it.
No posts in this category yet.
One email, every other Thursday.
The best of the blog and changelog. No spam, unsubscribe anytime.