# P13 — Pilot Roadmap (Weeks P1–P8)

**Updated:** 30 August 2026 · Owner: founder only. No build slot, no devs. ~6–10 founder hours/week.
Calendar anchor: P1 = week of 1 Sep 2026 → P8 gate ≈ week of 19–25 Oct 2026.

**Binding frame (from the portfolio):** 3 pilots from the agency network · ~2 weeks delivery per pilot · success = **3 pilots → 2 retained at full price** · kill criterion verbatim: **delivery time >10h/client/mo (service margin death) → drop or productize config into templates.**

---

| Week | Focus | Work (tasks in `TASKS.md`) | Acceptance — week is DONE when |
|---|---|---|---|
| **P1** | Platform + first sound | `F-01` platform account (Retell primary per D1) + US test number · `F-02` measure real $/min on our stack with 10 test calls · `F-03` shortlist 3 pilot clients (O1) + run missed-call audits · `F-04` record demo assets (before: their voicemail; after: our assistant) | You call the test number and a natural voice answers, books a dummy slot into a test calendar; real measured $/min logged in `PROGRESS.md`; 3 audit recordings exist |
| **P2** | Restaurant template v1 | `C-01` restaurant prompt pack · `C-02` call flows (reservation, hours/menu, message, escalation) · `C-03` calendar tool config (Cal.com/Google) · `C-04` lead/summary delivery via webhook → WhatsApp/email · Bangla/Arabic feasibility test calls (D10, log honestly in `COMPATIBILITY.md`) | Template passes the full 20-call QA script (`TEST_PLAN.md`) against a fictional restaurant; setup-hours counter started |
| **P3** | Pilot #1 live | `C-05` pilot 1 onboarding per `DEPLOYMENT.md` SOP (restaurant-system client preferred) · pitch with audit recording (`SALES_PLAYBOOK.md`) · signed pilot agreement + first invoice (O2) | Pilot 1 answering real calls; go-live gate (`RELEASE_CHECKLIST.md`) passed; forwarding rollback tested live |
| **P4** | Watch + template #2 | Daily transcript review pilot 1 (week 1 protocol, `TESTING.md`) · fix list applied · `C-06` clone template → vertical #2 (clinic or salon, per pilot 2's business) | Pilot 1 containment ≥ 70% of routine calls, zero unhandled dead-ends in week's transcripts; template #2 passes 20-call QA |
| **P5** | Pilot #2 live | `C-07` pilot 2 onboarding (target ≤ 15 setup hours — template payoff must show) | Pilot 2 live through the same gate; setup hours logged and compared to pilot 1 |
| **P6** | Pilot #3 live | `C-08` pilot 3 onboarding (third vertical if the network offers it) · weekly QA batch covers all live pilots in ≤ 2h total | Pilot 3 live; batch-QA rhythm holding at ≤ 2h/week for 3 clients |
| **P7** | Steady-state proof | Run all 3 · monthly per-client cost + hours report (margin floor check, `PRICING.md` §3) · client check-in calls · collect testimonial/referral asks | Every client: infra ≤ 40% of fee, labour on pace ≤ 2h/mo (annualized), owner has said in words whether they'd keep paying |
| **P8** | **Retention decision gate** | Renewal conversations at full price · write the gate memo in `PROGRESS.md` | **GATE (binding):** ≥ 2 of 3 retained at full price → CONTINUE: template library hardening, US client #1 via LLC, post-P8 setup fee per `PRICING.md` §4. Fewer than 2 → STOP selling; keep serving any retained client; write down why; re-evaluate only on new evidence |

---

## Mid-pilot tripwires (checked every week, not just at P8)

| Tripwire | Action |
|---|---|
| Any client's monthly hours trending >10h (the kill line) | Immediate: template/checklist the leak that week; if a second month breaches → **drop or productize config into templates** — verbatim, no third month |
| Infra >40% of fee for a client | Upsell/re-quote script (`SALES_PLAYBOOK.md` §6) within the week |
| Pilot candidate stalls >2 weeks on "thinking about it" | Replace from the network list — 3 live pilots by P6 outranks any single logo |
| Bangla/Arabic fails P2 feasibility | Sell English-only where acceptable; do NOT block the pilot count on language R&D |
| Platform pricing/quality shifts materially | Invoke fallback evaluation (Vapi) — decision D1 is re-opened only by this tripwire or the P8 gate |

## What this roadmap deliberately does NOT contain

No product build, no website beyond a one-pager (`MARKETING.md`), no white-label toolkit, no outbound-calling offers, no Gulf expansion before the P8 gate unless a partner brings a warm client at zero acquisition effort (O3).