The differentiator isn't six websites. It's the organization and pipeline that build, test, and operate them — mostly without human hands.
A multi-agent company structure — CEO, planner, builders, QA auditor — running work through sprint-style phase loops driven by cron. Roles have narrow charters: builders ship to staging only, QA signs off independently, promotion is a separate authority.
Every property gets a per-site staging server on an isolated development machine — noindex-enforced, keepalive-monitored, one port per site. Nothing reaches production unsigned: a build is only "done" when it has a staging URL and a passing QA verdict.
One batched production write per day, in a fixed window. Per site: snapshot → QA-PASS gate → promote → smoke check → report line. Rollback is a snapshot restore. No ad-hoc writes to production exist — by policy and by tooling.
An affiliate link proxy with click tracking, an A/B experiment platform serving variants and capturing metrics across sites, retry tooling that absorbs upstream model failures, and hourly health checks feeding the live status below.
Health of all six properties, checked hourly from independent infrastructure. This widget reads the same feed our operations dashboards use.
Feed updates hourly. If this panel shows stale data, the checker itself is down — which we would also know, because it pages us.
How a change reaches production: