Rule 11 · primary narrative

Skeleton-crew roadmap

The owner’s real goal is fewer keepers/day than today AND lower $/wk than today. This is the phased path there. Every phase is benchmarked against the real current baseline this build measured read-only from production Pulse: 48 animal-care keepers/day (avg this week, range 42–52) at $50,036/wk.

Read this before the table

Phase 1 (Rule 11 today) is MORE heads/day than today (61 vs 48), not fewer — stated plainly. It buys Rule-11 proof-of-life coverage at flat cost. Only Phases 2–3 (which require investment and are not built today) take heads/day below today. The $64,007 true-pool framing (Phase 1 is a −20.7% cut against it) stays visible on the Cost page, but it does not hide the heads-per-day truth here.

Today · keepers/day

avg CORE, measured 2026-07-13..19

Today · animal-care $/wk

$50,036

department-attributed

Phase 3 floor · keepers/day

≤ today AND lower cost

Phase 3 · $/wk

$39,944

requires investment

Phase table — heads/day and $/wk both vs the REAL current baseline

Phase	Keepers/day	vs today	$/wk	Rule 11?	Availability
Today (measured) Current animal-care schedule, as scheduled right now	48	same	$50,036	no	measured now
Phase 1 — Rule 11 now Zero new tech, Rule-11-compliant proof-of-life	61	+13 vs 48	$50,762	compliant	available today
Phase 2 — vision assist Vision model + ~61 cameras on low-risk multi-animal enclosures	50	+2 vs 48	$41,608	compliant	requires investment
Phase 3 — the skeleton crew Full stack + per-enclosure thermal + medical-acuity field	48	same	$39,944	compliant	requires investment
Absolute floor High-risk enclosures only — NOT Rule-11-compliant	35	-13 vs 48	—	no	boundary only

Phase notes: Today (measured): Real: avg 48 CORE keepers/day (range 42–52) this week; animal-care spend department-attributed. Baseline every phase is measured against. · Phase 1: MORE heads than today (61 vs 48) — stated plainly. Flat cost vs animal-care baseline, a −20.7% CUT vs the $64,007 true pool it actually draws from. Available today. · Phase 2: Requires investment / NOT built today (BehaviorOS = untrained config + mock dashboard; enclosure-vision = scaffold on mock data; cameras repo = 277-row sourcing spreadsheet). Issue #7. · Phase 3: Requires investment / NOT built today. Fewer heads/day than today AND lower $/wk — the owner's real goal. Needs medical-acuity field (#8, Dec 2) + ~$21–50k capex. · Absolute floor: Shown as a boundary, not a recommendation: cameras cannot carry proof-of-life for low-risk animals under Rule 11. Do not staff below this without dropping compliance.

Guardrail — what can NEVER be camera-only

287 animals · 614 weight

All Tier 3/4 animals + neonates — lemurs, primates, penguins, otters, sloths, parrots, hoofstock, crocodilians. Humans fully staff these even in the skeleton crew.

No camera ever proves proof-of-life

No camera proves fed / watered / undistressed for ANY animal — only 'alive' + 'zone temp OK'. Automation shortens a route; it never removes the human owner or the physical feed/water check.

What Phases 2–3 actually are

not built BehaviorOS — today an untrained config + a mock dashboard.
not built Enclosure-vision — a scaffold running on mock data.
not built Cameras — a 277-row sourcing spreadsheet (issue #7).
A build program of roughly $21–50k capex. Labeled “requires investment / not yet built” everywhere — never presented as available today.

Two levers we explicitly reject / separate

Shift-restructuring is not a headcount lever: proof-of-life is heads-per-day, and a dedicated check-round shift splits ownership — the exact Loki handoff failure. Front-desk / guest-services automationis a separate parallel workstream (issue #11’s $90k bucket), mentioned but not built here.

Provenance

Baseline source: real railway.com — Pulse Scheduling project (Kevin Haynie's Projects), Postgres DATABASE_PUBLIC_URL, READ-ONLY. Window: 2026-07-13 to 2026-07-19 (current week, contains today 2026-07-14). Cost engine validated by reproducing the $62,076.79 Houston week to within 0.034% ($62,097.81 computed). All 138 keepers have real measured shift-days (0 fallback); 485 keeper-days/wk available. Read-only; production Pulse never written.