The Crew

The agents that run the place

This is the part people don't believe until they see it. Every agent below runs on a schedule or on demand, in production, with a human review gate where it counts.

📋

The Morning Briefing Agent

Every morning it reads my calendars and inboxes, checks each project's status, and texts me a single structured briefing: family schedule first, then what each business needs today. My day starts with a report I didn't write.

🧪

The Synthetic Test Squad

Five scheduled tester personas exercise Legacy Loop like real users: they sign up, upload, journal, invite, and report what broke. Agent-written code is never trusted until agent-run tests and a human review say so. This is the governance layer most AI pilots skip, and it's why mine ship.

✍️

The Content Pipelines

QuickAZQuote's blog researches, writes, and publishes itself three times a week. Legacy Loop's weekly social package builds itself every Sunday. UncMode clips get packaged with captions and titles per platform. I review; they produce.

🛡️

The Guarded Harness

The whole crew runs inside a hardened Claude Code setup: deterministic hooks block dangerous operations (nobody touches git identity, nobody edits dead code paths), permissions are scoped per project, and the crew once ran a full multi-agent audit of its own configuration and fixed what it found.

The operating principle: production is cheap, trust is expensive. Agents draft everything; nothing reaches a customer, a stakeholder, or a git push without a gate it has to pass. I write about the wins and the failures in the build log.