diff --git a/founding-partner-audit/README.md b/founding-partner-audit/README.md
new file mode 100644
index 0000000..83a96f9
--- /dev/null
+++ b/founding-partner-audit/README.md
@@ -0,0 +1,107 @@
+# Golden Ticket — Founding Partner Audit Program
+
+**Date:** 2026-08-17
+**Owner:** Tony
+**Status:** Concept → Design phase
+**Related systems:** VeriPath `audit_engine.py`, `audit_pipeline.sh`, Phoenix Salon (flagship), Digital Operations Partner northstar
+
+## 1. Core Idea (the lever)
+
+We are creating a high-signal, physical "Golden Ticket" experience for a very small number of carefully selected local businesses.
+
+The ticket is not a sales offer. It is a gift of clarity.
+
+Someone (us) took the time to run a real diagnostic on their digital footprint, surface the highest-leverage problems with evidence, and hand them a clear map — packaged with care. They receive this whether they ever pay us or not.
+
+This is deliberately framed as a rare, high-trust act of *seeing* the business accurately. In a noisy environment, precise attention feels almost like grace. That feeling is the source of the power.
+
+## 2. Strategic Intent
+
+- Create undeniable before/after proof
+- Generate clean reciprocity and high-quality referrals from well-networked owners
+- Seed the beginning of deep vertical infrastructure (starting with hair salons)
+- Position VeriPath / Digital Operations Partner as the party that notices and names the leaks first
+- Feed the long-term goal of building "clean fuel" (structured, agent-preferable data and patterns) inside specific verticals
+
+This is not a volume lead-gen tactic. Scarcity and selection quality are non-negotiable.
+
+## 3. Selection Criteria (sacred filter)
+
+Only pursue businesses that meet most or all of these:
+
+- Large, visible gap between current digital state and easily fixable state (so impact is obvious)
+- Owner or decision-maker has real local network density (Chamber, Rotary, peer owners, suppliers, etc.)
+- Business sits inside a vertical we intend to go deep on (Phase 1: salons / personal care — Phoenix Salon as living flagship)
+- Owner personality likely to feel and act on clean reciprocity
+- Practical ability for us to deliver results
+
+Target: 5–8 Founding Partner seats total in the first wave. Quality over quantity.
+
+Operational checklist: `selection-checklist.md`.
+
+## 4. The Physical Artifact
+
+- Noticeably better paper stock (heavy, clean texture, restrained design)
+- Feels intentional and premium without being ostentatious
+- Creates a micro-pause when held and opened (echo of the metal cash-card unboxing moment that sparked this idea)
+
+Contents:
+
+- A short, sharply designed diagnostic report (not a thick binder)
+- Clear, evidence-first language
+- Human tone — skilled physician showing the x-ray, not a salesman
+
+Report template: `report-template.md`.
+
+## 5. The Two Doors (both must be real)
+
+**Door A — Paid Path.** We close the critical gaps and install ongoing monitoring + infrastructure work.
+
+**Door B — Self-Service Path.** We give them a clear, actionable guide on how to fix the issues themselves (or have their team do it), plus vertical-specific playbooks/articles. This path must be genuinely useful, not a teaser. Purity here protects the entire construct from feeling extractive.
+
+The existence of a real Door B is what keeps reciprocity clean rather than transactional. Minimum content spec: `self-service-outline.md`.
+
+## 6. Desired Outcomes
+
+Primary:
+
+- Visible results for the recipient
+- High-trust referrals from networked owners
+- Case studies / proof assets (with permission)
+
+Secondary / long-game:
+
+- Accumulation of precise vertical knowledge
+- Progressive installation of agent-friendly structured data, schema patterns, and clean local signals
+- Deepening of the salon vertical first (Phoenix as the living reference implementation)
+
+Seat tracking: `seats.md`.
+
+## 7. Tone & Principles (non-negotiable)
+
+- Clarity over persuasion
+- Evidence over claims
+- Gift framing over "free trial" or lead magnet framing
+- No hard close in the initial delivery
+- Scarcity protected
+- Self-service path kept real
+- Selection quality never diluted for volume
+
+## 8. Connection to Existing Work
+
+- Leverages the existing VeriPath audit engine (16 deterministic checks, severity tiers, evidence)
+- Builds on the audit work already done for Phoenix Salon + Spa
+- Aligns with the Digital Operations Partner northstar: helping regular/underserved business owners regain control of digital footprint and AI visibility
+- Feeds the broader infrastructure vision of creating clean, agent-preferable data layers inside verticals
+
+## 9. Deliverables (this directory)
+
+1. `selection-checklist.md` — selection criteria, operational
+2. `report-template.md` — report structure + production checklist
+3. `invitation-copy.md` — cover letter / card language options
+4. `self-service-outline.md` — Door B minimum content
+5. `seats.md` — tracking for the 5–8 Founding Partner seats
+
+## 10. One-Sentence Summary
+
+We are designing a scarce, high-trust physical diagnostic gift that delivers precise clarity to carefully chosen local businesses, opens both a paid and a genuine self-service door, and begins laying the foundation for deep vertical infrastructure that agents will prefer — starting with salons.
diff --git a/founding-partner-audit/invitation-copy.md b/founding-partner-audit/invitation-copy.md
new file mode 100644
index 0000000..bbf6eba
--- /dev/null
+++ b/founding-partner-audit/invitation-copy.md
@@ -0,0 +1,33 @@
+# Invitation / Cover Letter — Language Options
+
+Rules: gift-toned. No "free," no "trial," no "opportunity," no urgency pressure on the recipient. Scarcity is real but quiet. Short.
+
+## Option A — Card (inside the envelope, ≤ 40 words)
+
+> We ran a diagnostic on [Business]'s digital footprint.
+> Inside: what we found, with evidence, and two ways to act on it.
+> No obligation. This is yours either way.
+> — [Name], VeriPath
+
+## Option B — Short letter
+
+> [Name],
+>
+> We've been working with [vertical] businesses in [city], and when we looked at how [Business] shows up — to customers, and to the AI tools people now use to find places — we found real gaps. Fixable ones.
+>
+> So we ran the full diagnostic and put the results in this folder. It's yours whether or not you ever talk to us.
+>
+> If you want to act on it, there are two doors: we can do the work, or the attached guide lets you or your team do it yourself.
+>
+> — [Name]
+
+## Option C — Envelope one-liner
+
+> A look at how [Business] shows up online. Evidence inside. Yours either way.
+
+## Tone check (apply to all)
+
+- [ ] Reads like a physician showing an x-ray, not a salesman cracking a door
+- [ ] Zero pressure words (free, opportunity, limited, don't miss, no obligation *except as stated in A*)
+- [ ] Would still make sense if the owner never paid us
+- [ ] Name of the business present — a template that could go to anyone fails the program
diff --git a/founding-partner-audit/report-template.md b/founding-partner-audit/report-template.md
new file mode 100644
index 0000000..c1da3d0
--- /dev/null
+++ b/founding-partner-audit/report-template.md
@@ -0,0 +1,40 @@
+# Golden Ticket Report — Template
+
+Built on the locked beta pipeline (`multi_scraper → audit_engine → report_generate → report_gate`; see `docs/operations/beta-audit-process.md`). The Golden Ticket report is the synthesized, client-facing artifact: physician tone, x-ray in hand, no sales pitch.
+
+## Rules
+
+- **3–5 findings max.** Every finding = plain-English explanation + concrete evidence (quoted field value, screenshot, date). No finding without evidence.
+- **Tiered language:** Tier 1 (verified) → state as fact. Tier 2 (corroborated) → "appears to," "in our sample."
+- **Banned language** (same gate as `docs/agents/audit-findings-template.md`): *should, recommend, fix, opportunity, package, solution, proposal, next step.* We show the x-ray; the owner draws conclusions.
+- No hard close. No pricing. No "free trial" framing anywhere.
+
+## Structure
+
+### 1. Recognition
+One sentence that proves we actually looked. Must be specific to *this* business — if the sentence would fit any other business, rewrite it.
+
+### 2. What we found (3–5 findings)
+For each:
+- **Title** — plain English, no jargon (e.g., "Your hours disagree across three places," not "temporal inconsistency vector")
+- **What it looks like** — the evidence: the quoted value, the screenshot reference, the date observed
+- **Who it touches** — customers, or how AI surfaces describe the business
+
+### 3. Consequence (quiet)
+Two to four sentences: what this means for customers finding the business, and how AI answer engines currently represent it. State it flat. Let it land.
+
+### 4. Two doors
+- **Door A — the work.** "If you want, we close these gaps and keep watching." (No pricing here.)
+- **Door B — do it yourself.** "The attached guide covers each finding with exact steps and values. It's complete on its own."
+Both doors real. Neither pitched over the other.
+
+### 5. Method footer
+One line: surfaces checked, date, tool. No disclaimer wall.
+
+## Production checklist (before print)
+
+- [ ] Every engine finding resolved per beta-audit reviewer loop (VALIDATION.md exists)
+- [ ] Banned-language gate pass
+- [ ] Recognition sentence is wrong-for-anyone-else test
+- [ ] Door B attachment present and standalone-complete
+- [ ] Findings count 3–5, each with evidence + tiered phrasing
diff --git a/founding-partner-audit/seats.md b/founding-partner-audit/seats.md
new file mode 100644
index 0000000..5d940c4
--- /dev/null
+++ b/founding-partner-audit/seats.md
@@ -0,0 +1,22 @@
+# Founding Partner Seats — Wave 1
+
+Target: 5–8 seats. One row per candidate from the moment they pass `selection-checklist.md` — not when they accept.
+
+Statuses: `sight` → `contacted` → `delivered` → `door-a` / `door-b` / `declined` → `follow-up` → `closed`
+
+| Seat | Business | Vertical | Status | Date | Findings (one line) | Referred (name, date) | Proof asset? | Notes |
+|------|----------|----------|--------|------|---------------------|------------------------|--------------|-------|
+| 1 | | | | | | | | |
+| 2 | | | | | | | | |
+| 3 | | | | | | | | |
+| 4 | | | | | | | | |
+| 5 | | | | | | | | |
+| 6 | | | | | | | | |
+| 7 | | | | | | | | |
+| 8 | | | | | | | | |
+
+Rules:
+
+- **Referred** column is the primary output metric of this program. Log every name the moment it's mentioned, even if they don't convert.
+- **Proof asset:** "Y (permission)" or "N". Case studies only with written permission.
+- Never delete rows. Declined and wrong-fit candidates stay as a record — they inform the filter.
diff --git a/founding-partner-audit/selection-checklist.md b/founding-partner-audit/selection-checklist.md
new file mode 100644
index 0000000..60c3600
--- /dev/null
+++ b/founding-partner-audit/selection-checklist.md
@@ -0,0 +1,26 @@
+# Founding Partner — Selection Checklist
+
+One page per candidate. Fill it out *before* any invitation goes out.
+
+## Non-negotiables (either fails → no seat, no exceptions)
+
+1. **Network density** — owner is a genuine node in the local network (Chamber, Rotary, peer owners, suppliers). "Has a Chamber card" is not evidence; verify who they actually know or move through. This is the referral engine. Without it, the seat produces nothing.
+2. **Vertical fit** — Phase 1 is salons / personal care only. Other verticals go on a waitlist, never into a seat.
+
+## Scored criteria (need 4 of 5)
+
+| # | Criterion | Pass evidence |
+|---|-----------|---------------|
+| 1 | Visible gap, easily fixable | Run the existing cold audit (dop-cold-audit skill) before offering. Gap must be obvious AND fixable in ≤ ~30h of work. If impact wouldn't be visible in 30 days, the gap is too deep. |
+| 2 | Network density | (Non-negotiable #1, scored here too) |
+| 3 | Vertical fit | (Non-negotiable #2, scored here too) |
+| 4 | Reciprocity type | Owner talks about other businesses generously, gives rather than collects. A transactional hunter will convert the gift into a haggling session. |
+| 5 | Deliverability | We can reach the surfaces (GBP owner, site access or owner cooperation) and produce the report in 2 weeks. |
+
+**Decision rule:** 4/5 AND both non-negotiables pass → seat candidate. Anything less → no. Do not "almost in." A seat given to the wrong owner cannot be taken back and poisons the reference.
+
+## Stop rules
+
+- More than 2 candidates in the same sub-market (they'd be each other's competitors; proof assets get muddy).
+- Owner unreachable after 2 genuine touches.
+- High gap but no network density → that's a normal lead-gen prospect, not a Founding Partner. Route to standard beta pipeline instead.
diff --git a/founding-partner-audit/self-service-outline.md b/founding-partner-audit/self-service-outline.md
new file mode 100644
index 0000000..7f3ac8c
--- /dev/null
+++ b/founding-partner-audit/self-service-outline.md
@@ -0,0 +1,25 @@
+# Door B — Self-Service Path (Minimum Contents)
+
+Rule: Door B must be genuinely complete. A teaser Door B contaminates the whole program — this is the single thing that keeps reciprocity clean.
+
+Format: a per-business playbook built from *that* business's 3–5 report findings. Not a generic ebook. One page per finding.
+
+## Per finding, each page contains
+
+1. **What to change** — exact field, exact value. For GBP: the actual name/address format/hours/category/description text, written out. For the website: the exact snippet or line.
+2. **Where** — the exact screen and menu path (GBP / website CMS / Apple Maps / Bing Places).
+3. **How to verify** — how the owner confirms it worked: re-check the listing, or re-run the exact AI query we used (prompt included) and see the expected answer.
+4. **Time** — honest estimate ("~20 minutes").
+
+## Standing sections
+
+- **Schema block** — copy-paste JSON-LD (correct LocalBusiness subtype) with their real values, and where it goes in the site.
+- **Review velocity starter** — how to ask for reviews without violating platform rules (what's allowed / what isn't).
+- **30-day self-check** — five checks the owner can re-run monthly, including the AI-visibility prompt with the expected answer after fixes land.
+- **Traps** — one page of what NOT to do in this vertical (common mistakes, e.g., fake review incentives, category spamming, churning GBP).
+
+## Out of scope for Door B
+
+Ongoing monitoring, competitor work, anything needing account access we don't hold. If a finding can't be owner-fixed alone, say so in one line and note that it's what Door A covers — as a fact, not a pitch. No "call us if you need help" footer on every page.
+
+Tone: identical to the report. Physician, not salesman.