The Brightdata integration exposes 5 actions — 4 read,1 write — that any agent, team or AI employee can use once Brightdata is connected at the organization level. Every write can be gated behindhuman approval, and every call lands in theaudit trail.
Safe to run freely — they observe, fetch and search.
They change things. Gate the sensitive ones with approval policies; approvals are single-use.
Browse templates and solution packs that work with Brightdata, or start from the problem you're trying to fix.
