THE ARBITRAGE MONEY PLAYBOOK — FINAL

The pivot stands

Selling cortextOS into chiropractic clinics is a labor trap. Every clinic is a custom integration, a support ticket, a HIPAA liability, a sales call, a churn risk — a job wearing a SaaS costume, inside the most compliance-hostile vertical in software. Reject it. That part of the draft was right and nothing here changes it.

The reframe that changes everything

The draft's organizing claim was: you own the inputs (spare compute, a 6M-view catalog, clinical authority) so the marginal cost of monetizing them is near zero. That is half-true and dangerously incomplete.

The scarce arbitrage input is not spare compute or model budget. It is Zach's trusted judgment. Model budget is effectively free and effectively infinite. The fleet can produce unlimited drafts, scripts, thumbnails, SKUs, and clips. But the market and the platforms only pay for outputs that are trusted, differentiated, and compliant — and every one of those gates routes through a decision only Zach can make or explicitly authorize:

• - medical/clinical claims ("can we say KST helps X?")
• - patient consent and footage redaction
• - affiliate truthfulness and FTC disclosure
• - positioning and brand voice
• - platform appeals when a video gets flagged or demonetized
• - support boundaries (what we will and won't answer for a buyer)
• - "is this good enough / on-brand enough to ship?"

Every play in this document — even the ones that look fully autonomous — still queues decisions for Zach. A fleet that produces infinite drafts simply produces an infinite decision queue. If the plan does not CAP that queue, it becomes the exact job Zach is trying to avoid. That is the whole game.

The new primary scoring axis: judgment-queue load

The draft scored plays on labor / autonomy / speed / realism. That weighting missed the real bottleneck. This version makes "judgment-queue load" — the number of non-delegable decisions per week a play puts on Zach, and whether that number stays flat or grows with revenue — the primary axis.

The litmus test for every play:

• - Good play: revenue scales with compute (free), human decisions stay flat and batched. Doubling revenue does NOT double Zach's decisions.
• - Job in disguise: human decisions scale with revenue/volume/buyers. Doubling revenue doubles the queue. Amputate the component causing it, every time.

For each play below you will see an explicit Judgment-queue cost line: decisions/week on Zach, and how that queue is capped, batched, or delegated-with-guardrails.

The real unfair advantages (honestly stated)

1. 1. A live 24/7 autonomous multi-agent fleet in production — a genuine content/labor multiplier AND a non-fakeable story. Real moat.
2. 2. A 6M-view aged YouTube channel with a decoded viral format — ~4.5M of ~6M lifetime views from ~25 legacy 2009-2013 KST condition videos. You have the answer key (real patient + named condition + visible KST adjustment + outcome in title). But: a historical format proves history, not current monetization. Whether this is a cash machine or a dormant trophy is unknown until verified (see §2).
3. 3. Clinical authority + two finished podcast catalogs + the fleet narrative itself — sunk assets, cheap to re-monetize.
4. 4. Near-zero marginal AI cost — real, but NOT the cost center. The cost center is Zach's review time, consent/QA workflow, claims judgment, and platform risk. Compute being free does not make any of those free.

The threat that hard-codes a rule

YouTube renamed "repetitious content" to "inauthentic content" on July 15, 2025, and explicitly makes repetitive/mass-produced content ineligible for monetization. (The draft's specific "January 2026: 16 channels, 4.7B views, $10M/yr erased" anecdote is unverified — Codex could not find official support. Treat it as color, not fact.) The durable, sourced rule that matters: mass-produced, templated, AI-assembled content — especially in a health/YMYL niche — is in the demonetization blast radius. Owning the footage helps rights clearance; it does NOT automatically make clipped/templated/AI-assembled output transformative enough to pass review. Never run pure-faceless/zero-human on a real channel.

---

Three bets, ranked by judgment-queue discipline and cleanness of asset exploitation. They stack: A pays soonest, B is the highest-margin product, C is the 12-24 month authority asset that sells B.

Before any of them: the one load-bearing unknown that gates the whole plan — VERIFY YPP STATUS OF THE LEGACY CHANNEL FIRST. It determines days-vs-months for the only play that pays now.

---

BET 1 — Legacy KST Catalog Revival (audit + re-monetize ONLY; small real-footage tests)

The most grounded first-dollar idea — but only the catalog half, and only after the unknowns clear.

The arbitrage: Re-monetize ~6M views you already own. The marginal compute cost of a fleet audit pass is near zero; the views already exist.

Numbered first-revenue steps:

1. 1. VERIFY YPP STATUS + account health of the legacy channel — FIRST, before anything else. Capture: current subscribers, trailing 12-mo watch hours, YPP membership status, whether monetization is enabled/paused/disabled, copyright claims, strikes, geography of viewers, current monthly views, and RPM by video. This single block of facts determines whether revenue is days or months away.
2. 2. Capture the analytics baseline + current RPM (you cannot estimate revenue or detect a regression without it). This is a deliverable, not a guess.
3. 3. If YPP-active and clean: re-enable monetization on un-monetized legacy videos. (Note: re-enable is NOT guaranteed — it is an account-owner action subject to review, not "fleet does it autonomously.")
4. 4. Fleet audit pass over the ~25 proven winners: titles, thumbnails (nano-banana), end-screens, playlists, affiliate + lead-magnet cards — all drafted by the fleet, Zach approves in one batch.
5. 5. Add mandatory AI-content disclosure to any video with synthetic thumbnails/edits.
6. 6. Small real-footage tests only (Dr. Saylor incidental clinic capture → fleet post-production → batched publish behind the clinical-claims gate). Do NOT unleash a faceless condition factory on this channel (see Kills, §5).

Judgment-queue cost (explicit): Catalog audit = ~1 batched approval session, one-time. Ongoing: one weekly batched gate covering medical-claims review + title/thumbnail approval for any new real-footage upload (~3-5 decisions/week, realistically). Codex is right that the draft understated this — at scale, claims review + consent + comment moderation + capture can hit 3-5 hrs/week if not disciplined. Cap: one weekly review window; if the queue exceeds it, stop publishing net-new and let the catalog ride passively. The catalog half is genuinely near-passive; the net-new half is where the queue lives, so it is the thing you throttle.

Load-bearing unknowns (lead with these):

• - Is the channel actually in YPP and clean? (gates everything)
• - Current monthly views — 100-200k or 10-30k? (10x revenue difference)
• - Is chiro/KST content limited-ads/yellow-dollar? (likely yes — textbook limited-ads magnet)
• - Does the 2009-2013 format transfer to 2026 algorithms/audiences at all?

Grounded revenue (honest): If current traffic is high (100-200k views/mo) AND RPM is clean ($8-15), catalog alone could be ~$800-3k/mo. But if traffic is 10-30k views/mo OR content is limited-ads (likely), the honest near-term range is $50-600/mo. The draft's $3-10k/mo 12-month target is wishful unless a new upload hits. State the conservative range until the §2.1 facts are captured.

---

BET 2 — Branded Agent Templates / "Clone My Fleet" (as-is files, no support, no community)

The only play where revenue can scale without Zach touching each buyer — IF support is refused by design.

The arbitrage: The product is the fleet's own battle-tested artifacts. Near-100% margin. Updates happen anyway as cortextOS evolves. "Proven in production" is a real moat over commodity prompt-packs.

Numbered first-revenue steps:

1. 1. Mandatory sanitization, with an automated secret scanner + redaction checklist. Configs carry HIPAA-adjacent clinic context, 1Password references, secrets paths, agent identities. Human-reviewed scrub before ANY release. Reputational/secret-leak risk is the real downside here, not coaching.
2. 2. Install smoke-test on a clean machine — files that don't run create the support burden you're trying to avoid. The fleet can run this test; it must pass before shipping.
3. 3. Package tiered downloads (agent templates, skill files, orchestrator CLAUDE.md patterns, cron/daemon configs).
4. 4. Write the labor-cap into the product: explicit license terms, "as-is / no support" copy, a refund policy. This is the single most important step — it is what makes the play bettable.
5. 5. List on Gumroad + own site, fully automated delivery. Tiers $29 / $79 / $199 (price is plausible; demand is the unknown).
6. 6. Wire distribution to Bet 3 — cold marketplace discovery is thin; the channel is the traffic.

Judgment-queue cost (explicit): Near-zero by design. One-time sanitization approval (high-stakes, do it carefully). Ongoing: occasional template-update approval that happens anyway. The queue is capped by a hard rule: no support, no community, no coaching. The instant you add support, you've rebuilt a buyer-facing job. Refusing support IS the labor cap.

Load-bearing unknowns (lead with these):

• - Is there ANY traffic to sell to? (without Bet 3's audience, first-quarter reality is realistically $0-500, not $500-1,500/mo)
• - Do the files actually run on a clean machine for a confused buyer?
• - Did the scrub catch every secret/clinic reference? (this is a release-blocker, not a nice-to-have)

Grounded revenue (honest): Shipping in days is real. Selling $500-1,500/mo per product is NOT grounded without traffic, proof, screenshots, demos. First-quarter reality: $0-500 unless Bet 3 already has an audience. With a content channel feeding it, a 4-6 product catalog → $2-6k/mo at 6-12 months is plausible. State the $0-500 cold-start number honestly.

---

BET 3 — AI-Fleet-Builder Channel ("I run a 24/7 autonomous agent OS from my phone")

The least-replicable asset you own. A 12-24 month authority asset, NOT quick cash.

The arbitrage: B2B/AI-automation is a top-RPM niche, and the source material — real git history, real activity logs, a real proprietary OS with a real business attached — cannot be recreated by anyone in an afternoon. It survives the inauthentic-content purge for exactly that reason. 99% of "AI agent" creators demo toys; you operate one.

Numbered first-revenue steps:

1. 1. Stand up a self-documenting pipeline with a built-in redaction stage: a cron-driven content agent that drafts a script + screen-recording shot-list from real git history / activity logs (repurposing real work, not net-new labor) — then blurs/redacts secrets and clinic context automatically before it reaches Zach.
2. 2. Set a named persona + transformative POV from day one. Never pure-faceless — that is the demonetization kill-zone. Zach's script voice minimum; face/voice for the sponsorship tier.
3. 3. Publish long-form "how I built X with autonomous agents" + a Shorts network sliced off the same material.
4. 4. Build the affiliate matrix — but VERIFY each program first. Confirm availability, commission rate, approval criteria, payout threshold, and whether medical/automation claims restrict promotion (Anthropic, automation SaaS, hosting). Do not assume "first dollars in weeks" until at least one program is confirmed live.
5. 5. Open the sponsorship lane only after a track record + media kit exist ($500-10k/integration). Two or three videos do NOT create this.
6. 6. Funnel to Bet 2 as the channel's owned product. The channel IS the marketing for the templates.

Judgment-queue cost (explicit): Higher than the draft claimed. The authenticity moat forces a recurring human layer: script approval + redaction spot-check + occasional voice/face recording + "no overclaiming" judgment. Realistically 3-6 hrs/week, not 1-3 (Codex's correction — Zach must narrate, approve, protect secrets, blur logs, avoid overclaiming). Cap: a fixed weekly publishing cadence (e.g., one high-quality story/week) with a hard rule — if the fleet cannot produce one publishable story from real operations without exposing secrets, the play isn't autonomous enough and you slow the cadence rather than lower the bar.

Load-bearing unknowns (lead with these):

• - Do the affiliate programs actually pay, and are you eligible? (gates "fast money")
• - Can the redaction pipeline reliably keep secrets/clinic context out of public video? (release-blocker)
• - Can the fleet produce one genuinely good weekly story without Zach rewriting it? (determines real labor)

Grounded revenue (honest): Affiliate possible in weeks only if a program is verified and a video converts — otherwise months. AdSense 4-6 months (possible, not plan-worthy). Sponsorships 6-12 months. $500-2k/mo affiliate-led at 6 months → $1-10k/mo at 12-24. Treat as a compounding asset; pair with Bet 1 for near-term cash so you're not 18 months in with nothing.

---

THIS WEEK (resolve the unknowns that gate everything)

• - VERIFY YPP status + full account-health snapshot of the legacy channel. Single most important action in the playbook. (Bet 1, step 1.)
• - Capture the analytics baseline + current RPM by video. No revenue estimate is honest without it.
• - If YPP-active + clean: fleet drafts the catalog audit (titles/thumbnails/cards/AI-disclosure) for one batched Zach approval.
• - Start the one-time sanitization scrub + automated secret scanner for Bet 2. Release-blocker; front-load it.
• - Spec the Bet 3 self-documenting pipeline WITH the redaction stage. No publishing yet.
• - Verify ONE affiliate program end-to-end (eligibility, commission, payout). Resolves the biggest "fast money" unknown for Bet 3.

THIS MONTH (ship the capped-labor product + stand up the engines)

• - Ship Bet 2 as as-is files on Gumroad ($29/$79/$199) — only after the scrub passes and the install smoke-test passes on a clean machine. No support, no community, by design.
• - Run small real-footage tests on Bet 1 behind the clinical-claims gate (not a factory — tests).
• - Launch Bet 3 with a named persona + the first 2-3 build-in-public videos + a Shorts network sliced off them.
• - Establish ONE batched weekly review session covering every publish/claims/affiliate gate across all plays. (Honest caveat: Codex calls "one session clears everything" the biggest operational fantasy in the draft — so treat the session as a cap, not a guarantee. If the queue overflows the session, you cut publishing volume, not skip review.)

THIS QUARTER (compound — only what's attached to real audience)

• - Layer the podcast clip network (§4.1) as a pure amplifier on the back catalog, driving a defined destination (long-form/leads/product) — never as a standalone line.
• - Add the digital-product cross-sell (§4.4) ONLY once Bet 3 or an email list exists. No standalone cold-start.
• - Open affiliate + sponsorship lanes on Bet 3 once a track record + media kit exist.
• - 30/60/90-day kill review: any channel/SKU below threshold gets retired. This is how the portfolio stays high-margin and the queue stays bounded.

---

> Reframed around judgment-queue load. Plays the draft kept that Codex demolished are demoted or folded in honestly.

4.1 Two-Podcast Clip & Repurpose Network — KEEP (amplifier only)

• - Arbitrage: You own the source (two finished podcast catalogs — no sourcing grind, no licensing/strike risk) and the fleet runs the distribution tooling (late.dev/blotato).
• - How it works: Finished episodes → fleet transcribes → hook-detects → cuts → captions → schedules. Human glance on the publish gate only.
• - Monetization: Direct clip-rev is thin (Creator Rewards $0.40-1.00/1k US; Shorts ad RPM $0.01-0.05/1k). Real value is indirect: inflating sponsor rates + funneling to a defined paid destination.
• - Honest labor: 15-30 min/week brand-safety review.
• - Judgment-queue cost: Low but NOT zero, and it creeps: every clip can need context repair + guest-sensitivity review + clinical-claim review. Cap: batch the brand-safety review weekly; auto-publishing raw is reckless when clinical authority is part of the brand.
• - Autonomy: Highest in the set. Speed: Launch day-one (back-catalog = inventory); revenue slow.
• - Grounded revenue: Standalone thin (~$2k/mo cap is itself ungrounded without a current follower base + proven clip performance). Real value is attribution-dependent.
• - Discipline: Amplifier stacked on a paid offer with a hard-defined destination. Never standalone.

4.2 Legacy KST Catalog Revival — KEEP (Bet 1)

See §2, Bet 1. Catalog audit + re-monetize + small real-footage tests. Not a faceless factory.

• - Judgment-queue cost: One batched weekly claims/title gate (~3-5 decisions/wk); throttle net-new if it overflows. Catalog half is passive.
• - Grounded revenue: $50-600/mo if traffic/RPM low; $800-3k/mo only if traffic high AND RPM clean. 12-mo upside needs a new hit.

4.3 Branded Agent Templates / "Clone My Fleet" — KEEP (Bet 2)

See §2, Bet 2. As-is files, no support, no community, automated delivery.

• - Judgment-queue cost: Near-zero, capped by refusing support. One-time high-stakes sanitization approval.
• - Grounded revenue: $0-500 first quarter cold; $2-6k/mo at 6-12 months only if paired with a content channel.

4.4 Niche AI Digital-Product Vaults — KEEP (downstream cross-sell ONLY, demoted)

• - Arbitrage: 95-99% margin; the agent/automation angle (cron libraries, Claude Code starter vaults, n8n bundles) is credible to you and hard to fake.
• - Codex's correction (accepted): The margin math is true; the distribution math is not. Gumroad/Etsy provide no free demand for commodity AI bundles, and marketplaces are increasingly hostile to low-effort AI products. The draft's "$500-1.5k/mo in 2-3 months" is wishful cold-start.
• - How it works: Fleet generates niched products + listings + mockups, bundled $19-49. Lists where an existing audience/list can be pointed at them.
• - Judgment-queue cost: Shows up as SKU sprawl — broken templates, stale instructions, refund requests, marketplace messages, listing optimization, bookkeeping. Cap with a hard rule (Codex): no SKU ships unless Zach has personally used the artifact OR the fleet can test it. Retire dead SKUs ruthlessly.
• - Grounded revenue: Only meaningful as a cross-sell once Bet 3 / an email list exists. Do NOT run standalone first.

4.5 AI-Fleet-Builder Channel — KEEP (Bet 3) — and this is the ONE canonical entry

See §2, Bet 3. The draft listed this concept three times (as Play C, 4.7, and 4.9-with-community). Codex is right that the duplicates hid portfolio complexity. Collapsed into one canonical play here.

• - Judgment-queue cost: 3-6 hrs/week (script approval + redaction spot-check + voice/face + no-overclaim judgment). Capped by a fixed weekly cadence; slow the cadence rather than lower the bar.
• - Grounded revenue: $500-2k/mo affiliate-led at 6 mo (only with a verified affiliate program); $1-10k/mo at 12-24. Authority asset, not quick cash.
• - Hard rules: secrets-redaction pipeline mandatory; "no fake case studies" rule; never pure-faceless.

4.6 Legacy Catalog → Shorts + Affiliate — KEEP (traffic recycling, NOT a revenue play)

• - Codex's reframe (accepted): This is traffic recycling, not a standalone revenue line. Shorts RPM is structurally thin and YouTube excludes non-original/advertiser-unfriendly Shorts views from revenue sharing. Owning old footage reduces copyright risk but does NOT solve transformation/repetition/medical-claims risk.
• - How it works: Fleet auto-clips legacy long-form into Shorts, each driving to long-form watch time, affiliate clicks, or booked consults.
• - Judgment-queue cost: Clip QA + caption review — and a bad 25-second medical claim creates MORE risk than a long-form educational video. Cap: weekly batched caption/claims review; affiliate disclosure in every description.
• - Grounded revenue: $1-5k/mo is wishful unless it drives long-form/affiliate/consults. Must be measured by what it drives, not by Shorts RPM. Keep the archive-repurposing arm only; drop the camera-dependent format-recreation arm (job-shaped).

---

The draft kept everything ("no outright kills"). That was too soft. Codex is right — three things must die, and a clear standalone must be folded in.

KILL 1 — The faceless KST condition-video NETWORK (mass-produced health content)

Why: Generating 100+ scripts + faceless AI-VO/B-roll assemblies in a health/YMYL niche is precisely the mass-produced profile YouTube's inauthentic-content policy targets, stacked with medical-misinfo and advertiser-suitability risk. The "$1-3k by month 4-6, $5-10k by 9-12" numbers are wishful without proof the 2009-2013 format transfers to 2026 and survives review. Worse, running it on the same channel as the real 6M-view catalog risks the whole channel. Salvage: keep ONLY human-led, real-footage experiments under Bet 1. Kill the standalone network.

KILL 2 — High-RPM Faceless Channel Network (generic finance/AI/SaaS)

Why: It violates the playbook's own moat test. In commodity niches Zach has no unfair advantage over thousands of AI operators, and mass-produced channels are a correlated account-risk bet under the inauthentic-content policy. The "$10-40k/mo at 18-36 months" is survivorship-bias math. Decisively: "fleet cost near-zero" is irrelevant — Zach's judgment queue grows linearly with every channel and every upload (review, appeals, analytics, thumbnails, affiliate checks, policy monitoring). That is the definition of the job he's escaping. Kill it outright (not "background bet").

KILL 3 — The build-in-public CHANNEL + paid COMMUNITY as a separate play

Why: It is a duplicate of Bet 3 + Bet 2 with a labor-trap community bolted on. The "$3-15k/mo with the community" is a job-shaped fantasy: moderation, churn, member questions, refunds, live expectations, content cadence. Fold the channel into Bet 3 and the productized pack into Bet 2. A newsletter can come later, after the channel has consistent viewership. The paid live community is killed; if community ever happens it is 100% async (recorded library + agent-moderated forum, zero live coaching) and treated as a soft labor floor, not a growth engine.

DEMOTE (not kill) — Digital-product vaults to downstream cross-sell

No free marketplace demand at cold-start. Only fires once an audience/list exists. (Detailed in §4.4.)

REFRAME (not kill) — Legacy→Shorts/affiliate to traffic recycling

Must drive long-form/leads/products; not a standalone revenue line. (Detailed in §4.6.)

The throughline: every kill/demotion targets a component where Zach's decisions scale with revenue or volume. That is a job. The plays that survive share the opposite property — decisions stay flat and batched while revenue scales with free compute.

---

The draft had none of this. These are the load-bearing controls that keep the judgment-queue bounded and the account risk capped. Build these once; every play uses them.

Portfolio capacity model (the queue cap, in numbers)

• - Max Zach hours/week: set a hard ceiling (recommend 4-5 hrs/week total across ALL plays).
• - Max publish gates/week, max channels live, max products live: explicit caps. New play cannot launch if it would breach the gate budget.
• - Kill thresholds: 30/60/90-day review; anything below target is retired. The portfolio stays high-margin only by pruning.
• - Core metric: "Zach-minutes per dollar." Track it per play. Any play whose Zach-minutes/dollar is rising over time is becoming a job — amputate or kill.

Compliance kit (build once, reuse everywhere)

• - Patient release / consent form + workflow — required before ANY clinic footage is captured or published.
• - Medical-claims checklist + forbidden condition/title list — what KST/chiro content may and may not claim; YMYL-safe phrasing. Non-removable gate.
• - FTC affiliate disclosure standard — disclosure snippets + an affiliate registry (program, rate, terms, payout, restrictions). Material connections disclosed, claims truthful.
• - AI-content disclosure policy — applied to every synthetic thumbnail/edit; repeated nondisclosure can trigger YPP penalties.
• - Redaction checklist + automated secret scanner — for templates (Bet 2) and screen recordings (Bet 3). Release-blocker.
• - Account-isolation plan — never let an experimental/risky channel share an account with the real 6M-view asset.

Revenue instrumentation (you cannot manage what you don't measure)

• - Capture, per play: views, RPM, yellow-dollar rate, CTR, affiliate EPC, refunds, support tickets, and Zach-minutes/dollar.
• - Capture the analytics baseline + current RPM FIRST (Bet 1) — every revenue estimate in this document is provisional until this exists.

QA / refund harness (for any product)

• - Install smoke-test on a clean machine before any file ships.
• - Refund policy + "as-is / no support" license terms written into the product.
• - The shipping rule (the most important single control): nothing ships unless Zach has personally used it OR the fleet can test it. No untested SKU, no unverified affiliate claim, no unredacted asset, no unproven case study.

---

Reframe first: stop thinking of spare compute as the arbitrage. It isn't scarce. Zach's trusted judgment is the scarce input, and the entire job of this playbook is to cap the decision-queue that judgment must service. Every play is scored by judgment-queue load; any component where Zach's decisions grow with revenue is a job and gets amputated.

Do this, in order:

1. 1. This week — resolve the unknowns. Verify YPP status + account health + analytics baseline of the legacy channel. This gates everything and turns guesses into numbers. Verify one affiliate program. Start the secret-scrub.
2. 2. Bet 1 (catalog revival, audit-only) for near-term cash — honestly $50-600/mo if traffic/RPM are low, more only if verified high and clean. Catalog half is passive; throttle net-new to keep the claims-gate inside one weekly session.
3. 3. Bet 2 (as-is template files) as the highest-margin product — capped to near-zero queue by refusing support/community. $0-500 cold, $2-6k/mo only once a channel feeds it.
4. 4. Bet 3 (fleet-builder channel) as the 12-24 month authority asset that sells Bet 2 — budget 3-6 hrs/week honestly, fixed cadence, redaction mandatory.
5. 5. Layer amplifiers/cross-sells (4.1, 4.4, 4.6) only onto real audience. Never cold, never standalone.
6. 6. Kill the faceless KST network, the generic faceless channel network, and the paid community — they all make Zach's queue grow with revenue.
7. 7. Build the system-level compliance/QA kit once and route every play through it.

The honest bottom line: This is not a passive-income machine you switch on. It is a small, disciplined portfolio where free compute does the production and a strictly bounded slice of Zach's judgment does the gating. The moment any play starts demanding attention proportional to its income, cut the component causing it. Hold that line and this stays arbitrage. Drop it and you've rebuilt the clinic job in a new costume.