One daemon.
Every business you run.
If you own more than one thing — a franchise, a service company, a side project, a real-estate LLC — you're the only person holding the whole map. Operator is the background process that watches all of it so you can sleep.
A self-hosted daemon watching N things at once.
Daily rituals
Morning briefing at 7am. Deploy check at 9. Weekly state-of-the-businesses email on Sundays. You pick the cadence; Operator keeps them running.
Every surface
Production URLs, customer lists, inbox volume, revenue dashboards, deploy status. Anything with an API or a health endpoint.
On your rules
Tier 1: notify only. Tier 2: draft a reply, wait for you to send. Tier 3: auto-merge, auto-email, auto-invoice. You choose per-workflow.
Running live on the author's portfolio right now.
- >Morning briefing — every workday, a digest of every project: what moved, what broke, what needs a human today.
- >Deploy health — every production URL HTTP-pinged every 15 minutes; pages land in Discord the second one flips red.
- >Sprint tracker — 90-minute work-block timer, commit-since-start heartbeat, auto-generated handoff docs at session end.
- >Auto-PR review — Claude reads every pull request, comments the concerns, flags the risky diffs, approves the boring ones.
- >Claude spend — running ledger of every agent invocation so the bill never surprises you.
- >Public broadcast — a read-only page the operator can share with partners, investors, or family so the entire portfolio is visible in one screen. See it live →
If you run more than one business, here's the shape.
Lead scoring across every location
- >Watch inbound forms, calls, and walk-ins at each store.
- >Score each lead: hot / warm / cold with a reason.
- >Draft the follow-up email in your voice. You approve with one click.
- >Roll up a weekly per-location scoreboard so you know who's actually closing.
Customer-health nudges
- >Daily sweep over your CRM for customers going cold.
- >Detects no-contact windows, missed renewals, stalled invoices.
- >Drafts the right nudge — service reminder, seasonal check-in, referral ask.
- >Flags the 3 accounts most likely to churn this week.
- >See a live mockup at /demo/pools.
State-of-the-businesses brief
- >Sunday evening, one page to your phone or inbox.
- >Revenue delta per business, deploys this week, new leads, new invoices.
- >Anything outside normal ranges flagged with a one-line story.
- >You read it in two minutes and you know exactly where Monday starts.
No vendor moat. Your data never leaves your hardware unless you tell it to.
Operator is alive 24/7. It notices things between your check-ins — that's the whole point.
Every recommendation quotes the source — the email, the row, the commit. You can always verify.
Operator ships with sensible defaults, but every task is a text file you can rewrite. It speaks your ops language.
Want to see the live one first?
/kruz is the author's actual portfolio — same daemon, same code, same Claude agents you'd run on your own businesses. /docs is how it gets onto your machine.