WilliamDavidThe 90-Day Engine

Day
Plan locked

Overview

Monthly Tally · Calendar Month
Commission earned
R0
vs Target
0% of R150,000
R150,000 to go
Files closed
0
Gross revenue
R0
Avg commission / file
R0
Daily run-rate needed
R0/day
Total Revenue · all time
R0
Units Sold · all time
0
Avg Sale Value · all time
R0

Sales by Month

The 30-60-91 Gates

Day 30 · Wed 17 Jun
R150,000Nett target · Month 1
R0 commission0%
Safe Floor Target
PENDING
Floor breakdown
Day 60 · Fri 17 Jul
R190,000Nett target · Month 2
R0 commission0%
Safe Floor Target
PENDING
Floor breakdown
Day 91 · Mon 17 Aug
R245,000Nett target · Month 3
R0 commission0%
Safe Floor Green Target
PENDING
Floor breakdown

Daily Signal

The five numbers that run the day. Sales Block 09:00–13:00, Off-Sprint Block 14:00–16:00. Mon–Fri only. Weekends = payments clear passively, no active work.

Signal Five
Sprint Day —
Today's Off-Sprint focus: —

Today

History · last 14 days

Slipped items reveal the real story. Don't carry them silently.

The Mission

The North Star

By 30 July 2026, WilliamDavid clears R150k+ surplus cash, ships the foundations of Ella, and one travel professional has paid real money for B2B access.

The Three Wins

Win 01
Cash
R150k+ surplus over 90 days from flash sales + R299 Plan-My-Trips.
Pays the floors. Sacred.
Win 02
Foundation
100 verified suppliers, R299 live, 4 destinations of Reasoning Extraction.
Builds the engine.
Win 03
Proof
One paid B2B pilot at any size.
Tells us Ella is real.
The Morning Read 3 min · read this daily

You are ninety days into building something formidable. By the end of it, the bills are paid, the engine is started, and one travel professional has paid you for what you know. That last sentence is the test of whether any of this is real.

The plan has three threads woven at the same time. Cash is the first — flash sales of custom Zanzibar packages, repeated and refined, sold to warm and cold audiences. This is what pays the rent, the tax man, and you. It is also the foundation the rest is built on, so it does not get dropped, ever, even when the more interesting work is calling. Foundation is the second — the supplier network, the Reasoning Extraction, the R299 Plan-My-Trip. These are not products, exactly. They are assets that compound when you stop touching them. They are the part that survives 2028. Proof is the third — one travel professional who pays you for B2B access to what you are building. That single yes tells you the bigger business is real.

The first two weeks are about building the rails. You build the Quote Builder Sheet, the master Zanzibar package architecture, the Trust Stack landing page, and the WhatsApp Business reply templates. You spend one full day on the package architecture, and after that every quote takes five minutes instead of three hours. The Sunday-night kill criterion at the end of fortnight one is two closes or you escalate the trust stack. That is the rule.

The next two weeks are about founding what becomes Ella. You recruit thirty suppliers from your existing rolodex onto the WilliamDavid Verified Africa Network — free entry, no listing fee ever; our buyers fund the network. You write the Zanzibar Reasoning Extraction over five hours of structured interview, putting your judgment into a document that becomes the system prompt for everything Ella will ever know about Zanzibar. You send the first three editorial pitches to a private bank, a card programme, and a magazine. You brief the contractor for the R299 Plan-My-Trip, and you run the third flash sale if the second worked.

The two weeks after that are tools. The R299 Plan-My-Trip ships live. You film the launch reel — your face, Cape Town, a real WhatsApp from a named lodge, the PDF that shows lodges anyone can Google. You give twenty Plan-My-Trips away in exchange for case studies. The second flash sale tests whether the offer architecture works on a different destination. The supplier network reaches sixty.

Then come pitches and the closer. Six more editorial pitches go out. The supplier network reaches one hundred. You write the Mauritius Reasoning Extraction. Closer interviews open. By Day 60 you make a pass / extend / pivot call on the cumulative numbers and decide whether to continue, hold, or rebuild the offer.

The last six weeks are about compounding. The closer is hired and onboarded. AVN Verified Plus opt-in opens for the suppliers who want amplification. The first B2B pitch goes to a mid-sized SA tour operator. The R299 starts running paid traffic. By the end of week twelve, the closer is running 70% of the inbound, and you are doing supplier work, B2B pitches, and a handful of high-value closes. One travel professional has paid you. The bills are paid, the engine is started, and your hands are off the inbox enough that you can think.

The shape of the day, daily, is this. You reply to warm and hot leads in two windows — early morning and late afternoon — using the templates and the Quote Builder. You spend an hour on supplier acquisition or the editorial pitch of the week. You spend an hour on Reasoning Extraction when it is that destination's turn. Everything else is execution that the system or the closer takes off your plate. You are not the bottleneck. You are the brain that decides.

What you are building, in one sentence, is Africa's verified travel desk. The supplier graph, the reasoning, the trust layer — accessible eventually by anyone over WhatsApp. The Zanzibar flash sale is the wedge that funds the build. The closer is the human who scales the close. The R299 is the funnel that surfaces demand. The Reasoning Extraction is the moat that no engineer can cold-build. Ella is the brand under which all of this finds its name.

Eight and a half years built the moat. Ninety days reveal the business. You are exactly where you need to be.

What does "surplus" actually mean? Tap to expand

Surplus = nett earnings above the monthly cost of running the business + paying yourself. Tracked monthly so the target lives in the cadence you actually operate in.

Monthly Targets · Stage by Stage

BASELINE COSTS (every month) Carmen take-home R60,000 Tech stack (all-in) R15,000 Services (legal + acc.) R10,000 Tax provision (~25%) R25,000 ───────────────────────────────── Baseline burn: R110,000 MONTH 1 · solo Nett earnings target: R150,000 Surplus above baseline: R40,000 MONTH 2 · solo, leaning Nett earnings target: R190,000 Surplus above baseline: R80,000 MONTH 3 · with Closer + VA Closer (base + comm) R25,000 VA / coordinator R10,000 ───────────────────────────────── Total burn: R145,000 Nett earnings target: R245,000 Surplus above burn: R100,000 MONTH 4–6 · scaling Tech stack scales 2× to R30,000 Possible R35k senior hire Nett target rising toward R300k MONTH 6+ · technical hire Add platform/Azure consult Streamline + secure Ella infra

The 90-day Win 1 (R150k surplus) is the cumulative version of this monthly compounding: ~R40k + ~R80k + ~R100k = R220k over 3 months at the high end, R150k at the realistic floor. The dashboard tracks against the floor (R150k) so we always have margin.

What "surplus" is NOT: not gross revenue, not commission, not retained earnings before tax. It's the cash you can take to the bank or reinvest after Carmen is paid, the lights are on, and SARS is provisioned for.

Right Now
Pre-launch
Read the plan. Approve or revise. Mon 4 May is Day 1.

The 90-Day Map

#DatesThemeHeadline
F119 May – 1 JunFoundation+13-day salvage sprint. Trust stack live. Daily Signal Five running.
F22–15 JunFounding30 verified suppliers. RE Zanzibar finalised. R299 contractor brief approved.
F316–29 JunToolsR299 live (22 Jun). Round 4 flash sale. 60 verified suppliers.
F430 Jun – 13 JulPitch9 editorial pitches in market. 100 suppliers. Closer interviews open.
F514–27 JulHireCloser onboarded. Verified Plus opt-in opens 1 Aug. First B2B pitch.
F628 Jul – 10 AugCompoundFirst B2B paid pilot. Day-91 review Mon 17 Aug.

This Fortnight · Who Am I Helping?

One person, one act, no transaction. Keeps the soul oxygenated. (86 it later if it ever feels forced.)

The Plan — 30 Steps

Tap the box to mark done. Tap "+ notes" to log progress. Both save on this device.

F1 · Foundation+ 19 May – 1 Jun
F2 · Founding 2–15 Jun
F3 · Tools 16–29 Jun
F4 · Pitch 30 Jun – 13 Jul
F5 · Hire 14–27 Jul
F6 · Compound 28 Jul – 10 Aug

14-Day Checkpoints

Every fortnight, Sunday 8pm, score against this. Tap to expand. Add achievements at the bottom of each.

30-Day Gates

Day 30 · Wed 17 Jun · "Is the cash machine working?"

PASS: ≥R40k cash. ≥3 bookings. ≥30 verified suppliers. RE Zanzibar in progress.

FAIL: <R40k OR <2 bookings. Stop building. All energy into trust stack + warm-list re-engagement.

Day 60 · Fri 17 Jul · "Are we ready to scale?"

PASS: ≥R75k cash, R299 live + case studies, ≥60 verified suppliers, Round 4 flash sale shipped, closer interviews booked.

FAIL: Extend Route A 30 days OR pivot to high-ticket private-list bookings.

Day 91 · Mon 17 Aug · "Is Route B alive?"

PASS (must hit ALL): ≥R150k surplus · ≥R80k MRR · Closer running 70%+ inbound · ≥10 paying suppliers · One travel professional has paid for B2B access.

PASS → Route B begins.

Miss Win 1 → pivot. Cash machine is broken.

The Sunday-Night Scoreboard

R0
Cumulative cash
0
Founding suppliers
0
RE hours
This week — cash (R)
Suppliers added
RE hours

Archive on Sunday night. Cumulative ledger keeps the running total.

How It Works

Everything you need to remember about Ella, the engine, the audiences, the pricing, the trust architecture, the R299, the supplier rates, and the endgame. Always here.

Definitions — the only words that matter

Ella
A travel brain on WhatsApp, trained on Carmen's reasoning and connected to a verified network of African suppliers. One product, one engine, three audiences. The endgame business.
The Engine
Carmen's reasoning + the supplier graph. The single backend Ella runs on. Built once, used three ways.
The Reasoning Extraction
Carmen's documented decision-making for each destination. The actual moat. (First Principles: engineers can copy a supplier list in six months; nobody can copy how Carmen decides where to send a honeymoon couple in shoulder season when the wind is wrong.)
WilliamDavid Verified (public name TBC — internal shorthand: WAVN)
The supplier graph, named for the world. Vetted lodges, DMCs, transfers, activities — with seasonal rates, response-time signal, Carmen's verification. Suppliers join free at entry — not for a limited period, free. Our buyers fund the network. AVN Verified Plus at R499/mo is opt-in amplification only, available from 1 August onwards for suppliers who choose it. Doctrine: "Suppliers don't pay to join the network. Buyers pay to use it."
Round 2
The new Zanzibar offer relaunching mid-May. Four custom-built packages, supplier-aligned deposits, trust stack on the deposit page, deposit funded directly into the supplier's account.
The R299 Plan-My-Trip
Landing page → 12 questions → R299 → personalised PDF travel plan generated by Ella. Lead-magnet that builds your warm list and proves the AI works.
The Trust Stack
What appears on the deposit page that makes a stranger comfortable wiring R8–15k: testimonials with full names, ASATA membership number, supplier logos, the supplier's own banking details visible, 90-second founder video, "speak to Carmen on WhatsApp before paying" button.
The Day-91 Condition
The only test that matters: did one travel professional pay real money for B2B access by 30 July 2026? Everything else is vanity.

Ella, WhatsApp AI, and the supplier graph — they are one thing

Ella is one product with one engine and three different audiences. The engine is a single brain that has two pieces of knowledge: (a) your reasoning — your taste, your judgment, the way you decide what to recommend; and (b) a verified supplier graph — a structured database of African travel suppliers (lodges, DMCs, transfers, activity operators) with their seasonal rates, response-time reliability, and your personal verification stamp.

That single engine — your reasoning + your supplier graph — gets exposed to the world through different surfaces, and each surface is a different product to a different buyer. WhatsApp is one of those surfaces. The web chat at hey.williamdavid.co.za is another. A B2B login for tour operators is another. They are not separate products. They are the same product wearing different clothes.

Why this matters: you only build one engine. You sell access to that engine in different shapes to different people. That's what makes it formidable rather than fragmented. Eight products is unfocused. One engine, three audiences, is a platform.

What does Ella actually do, and how?

In one sentence: Ella is a travel brain on WhatsApp, trained on your taste and connected to a verified network of African suppliers, that anyone — a holidaymaker, a tour operator, a partner brand — can talk to and get a real, opinionated answer with a real, bookable price.

Step by step: a user sends a message in WhatsApp ("I want to take my family to Zanzibar in October, R200k budget, hate beach bars"). The message lands in your backend. The backend hands it to the AI brain (Claude). The brain reads it, looks up your reasoning rules, then queries the supplier graph. The brain composes a reply in your voice: "Stay at X, not Y. Here's why. Want me to send the link to confirm?" Total time: under ten seconds.

The three audiences and what each pays for

Audience 01
End users
Holidaymakers. Get the WhatsApp brain.

Taste: Free, 10 messages/mo.
Companion: R149/mo, unlimited + memory. Recommended launch tier.
Insider: R499/mo, adds human handoff + full WhatsApp booking.

Cost to serve ~R20–32. Gross margin ~80%.
Audience 03
Tour operators / B2B
Plug your engine into THEIR ops.

Embedded API access: R15k–R25k/mo.

White-label or embed Ella inside their WhatsApp / website / call centre. 60–90 day cycle, starts Q3.

This is where the formidable money is.

The supplier graph — rates, verification, the real-world plumbing

You asked the right question. Most suppliers don't have a live API. NightsBridge / ResRequest / SiteMinder are common; many lodges still send rates on a PDF. Here's how the graph actually works, the rates we hold, and how the discount mechanic flows.

Do we need live rates?

No, not for MVP. Probably not for v2 either. The premium African travel market runs on seasonal rate cards, not real-time inventory. A lodge sends you peak / shoulder / off-peak rates valid for a season (typically 3 months ahead, refreshed quarterly). Those are the numbers we hold. Availability is confirmed by a quick WhatsApp to the supplier when a real booking comes in — that takes 5 minutes, not 5 seconds, and that's fine because the booking is human-handled at close.

What Ella actually queries: "Is the supplier in our verified graph? What's their rate band for these dates? Do they typically have space at this time of year? What's their response-time SLA?" The bot replies with a confident estimate ("R23,500–R26,000pp for that week, exact rate confirmed within 24 hours") and Carmen / the closer confirms the precise number with the supplier on a WhatsApp ping. The bot doesn't lie; it qualifies.

Live API integration becomes worth it once we have 200+ suppliers AND volume to justify it (probably 2027). For now: structured rate cards in Notion or Airtable, quarterly refresh, supplier-confirms-on-booking.

The "WilliamDavid rate" — how the discount actually works

You're right that we need a real value exchange. Here's how it lands:

The verification process — what each supplier goes through

Five steps. Carmen-led for the first 100, delegated to the closer thereafter.

  1. Eligibility check: Carmen has either personally booked or stayed at the supplier OR has direct evidence of recent verified guest experience (review chain, supplier-shared booking confirmations, GM intro). No verification without this.
  2. Onboarding pack sent: a one-page invite explaining the network, expected commercials, response-time SLA, the data-rights consent, and the free-entry / buyer-funded model. PDF + WhatsApp message. Deliverable they need to send back: rate card, contact details, banking details, booking T&Cs, refund policy, response-time commitment, signed data-rights consent. Canonical pack: LLM-Council-Setup/strategy/suppliers/onboarding/v2_buyer-funded/.
  3. Onboarding form: a Typeform or Google Form that captures: lodge name, region, type, capacity, low-season / shoulder / peak rates, single-supp policy, child rates, transfer add-ons, payment terms, deposit policy, cancellation policy, banking details (for direct-to-supplier deposits), GM contact, reservations contact, WhatsApp number for confirmations.
  4. Test booking: Carmen sends a real or simulated booking enquiry. Measures response time, accuracy, completeness, professionalism. If >48 hr response or vague answer, supplier goes on a watch-list, not the verified graph. Response-time SLA is what makes Ella reliable to her users.
  5. Listing + Verified badge: supplier is added to the graph as AVN Verified (free). Verified Plus opt-in offered separately, never required. Quarterly rate refresh diary entry created. Verified badge issued (a graphic they can use on their own marketing).

Where this lives: strategy/supplier-graph.notion (or Airtable). Each supplier = a row with rate-card, response SLA, contact, T&Cs, last-verified date. See the Build Guides tab for the full Notion schema and the onboarding form template.

Bloomberg for African Travel

Bloomberg sells verified, structured data + relationships + a terminal professionals can't work without. Ella maps onto that exactly: verified African supplier data + 8.5 years of relationships + a WhatsApp terminal. Real money came from Bloomberg's professionals, not consumers — same here. R149 consumer is brand-builder; R15k+ B2B is where the floor clears.

Use carefully. "Bloomberg for African Travel" is for your head + sophisticated B2B pitches once you have something to log into. Until then, externally call it "a verified supplier desk on WhatsApp" or "the Michelin Guide for African travel trade." Earn Bloomberg in 2028.

The trust architecture — AI gives advice, human closes

If trust is the binding constraint, why would anyone book a R20,000 trip from a ten-second AI reply? They won't, and Ella shouldn't try to make them. Ella does the advice layer (free, fast, builds reputation). A human does the booking layer (deposit, supplier confirmation, contract).

When a user says "OK, I want to book it," Ella does not send a Yoco link. Ella replies: "Lovely — I'm handing you to a real person on our team to confirm dates and lock supplier rates. They'll be with you in the next two hours." The human picks up, confirms supplier availability, sends the formal quote, sends the deposit link. The handoff is the trust layer.

This is the deck's R499 Insider tier — "Hand-off to a WDT consultant when needed." That's not a feature; it's the architecture.

Cap inbound while solo. The handoff is the trap, not the safety net — the moment Ella works, every booking-intent lands on you. Until the closer is hired (Day 60-ish), Companion tier sells from a waitlist, not a live signup.

The R299 Plan-My-Trip — how to demo without looking like a scam

Without context, "R299 to plan your trip" looks like every fake travel scam on Facebook. You don't sell it cold. You demo it openly first.

Demo One — Film yourself using it

60–90 second reel. On screen (NOT in voiceover): Carmen's face, Cape Town backdrop, a real WhatsApp from a named lodge, the PDF showing supplier names a viewer can Google. Slow it down. The "90 seconds" brag actively hurts you — it screams ChatGPT wrapper.

Demo Two — Give 20 away in public

"Send me your trip brief and I'll generate your Plan-My-Trip free this week. In exchange, I'll publish your plan as a case study on socials and tag you." 20 published case studies = the real trust signal. The 21st person who lands on the page sees the proof.

Demo Three — Show the inside, not the outside

Show three real plans, opened, scrollable, with annotations. Scammers can't show the work because they don't have any.

Why not ChatGPT or Flight Centre?

Against ChatGPT: "ChatGPT can talk about Zanzibar all day, but it cannot tell you which lodge actually has rooms 14–18 October at the real net rate, with a five-hour response time, that I have personally stayed in." Verified supply data + personal taste.

Against Flight Centre: "Flight Centre will sell you whatever's commissioned highest. I'll tell you what's actually right for you, even when it costs me money."

Headline options

"ChatGPT will send you to a lodge that closed in 2023. I won't."
"The travel agent who tells you what's right, not what pays best."

Price-anchor

"The average South African couple spends 14 hours researching their honeymoon and still books the wrong lodge. R299 buys you a plan from someone who has stayed at the right one."

Deposit & commission math (corrected)

Commission on supplier bookings is 8–12%, not R1k pp:

The cash machine works. 13 bookings/month at R19,000pp is genuinely possible.

Deposit: keep it supplier-aligned (flights paid in full + 10% of land). Don't lower the ask — strengthen the trust stack around it.

Custom-package workload offset

One full day to architect the master Zanzibar package (3 lodges × 3 price tiers, transfer partner, flight pattern, single-supp pre-baked). After that, every booking is 5 minutes via the Quote Builder Sheet + WhatsApp template. Less work per booking than current bespoke-from-scratch.

The R299 Typeform — how it sets up

Carrd page → Typeform (12 questions) → Yoco payment → Make.com webhook → Claude API → Google Docs PDF → emailed to client with WhatsApp CTA. R2k/mo tooling. 10–15 hrs solo, or R3k contractor (1 week). Best use of Carmen's time = the prompt + the PDF template. Contractor wires plumbing; you write the brain.

Editorial licensing — in plain words

You write to a magazine, bank, or card programme and offer to sell them six finished travel articles. Each one a polished, photographed Blueprint. They publish under their brand or co-branded. They pay R40k–R80k flat for the pack. Inside each article, a soft mention of WilliamDavid + a way to talk to Ella on WhatsApp.

One-page email + sample Blueprint PDF. One named editor at one publication. 2 pitches/week from Day 15. One yes pays the WD floor for that month.

Blueprint protection — offsetting the "selling my packages" worry

The Blueprint is a fixed PDF. It is the cheapest version of what Ella does. Anyone who buys one and gets value will eventually want a customised, dynamic, conversational version — Ella on WhatsApp at R149/mo. The Blueprint is the trailer; Ella is the film.

Inside each PDF: CTA to Ella, unique tracking code, partner-only voucher (R100 off first Ella month). Licensing agreement says: "Editorial use only, all conversion to bookings via WilliamDavid systems." They get content; you keep distribution + customer relationship.

The Closer — Compensation

You flagged that R8k base from the original Council was too low. Agreed — we're not paying minimum wage and we're not asking the closer to bring intent without a fair share of the upside. Here's how the SA market structures it for travel ops + closing roles, and what to offer.

Three structures, ranked by fit

  1. Hybrid (recommended for first hire): R15,000 base + 20% of WD's commission per closed booking. At average performance (13 bookings/mo × R1,900 comm) the closer earns R15k + ~R5k = R20k. At good performance (20 bookings) they earn ~R23k. At excellent (30 bookings) ~R26k. Skin in the game, predictable floor. This is what your Month-3 R25k allocation buys.
  2. Pure commission (test-fit only): 35–40% of WD's commission per booking, no base. Closer earns R0 if they close R0; ~R10k for 13 closes. Best for someone you want to test for 30 days before committing salary. Risky for them — uses signal up front.
  3. Pure base (avoid for first hire): R20–25k flat. Predictable cost, no incentive alignment. Use only when the role is fully ops-managed and closing is one of many things they do.

What the role actually does

Where to find them

SA travel-industry LinkedIn, the boutique-DMC network, ASATA member associations, ex-Flight Centre / ex-Travelstart who went out on their own. Cape Town remote works fine. Ideal candidate: 3–5 years in travel sales, ASATA-aware, comfortable on WhatsApp Business, gets why supplier banking is on the deposit page (i.e. has the integrity instinct).

Onboarding (Day 31 if conditions met)

The Endgame — October 2026

If this works: 100+ verified suppliers in the network (R7–15k MRR realistically, R50k stretch), 200 paying Companion users (~R30k MRR), two licensing deals at R50k each (~R8k/mo amortised), one B2B pilot at R15k/mo. ~R60–80k MRR honest, ~R100–120k stretch. Plus flash sales R80–150k net underneath.

Most of that revenue does not require Carmen to be on every reply. About 25 hours a week, on the things you actually want to do. The autonomous offering. The endgame.

Build Guides

Click into any build to see the full step-by-step, the Claude Code agents you should call, where to store the output, and the realistic time. Everything you need is in here — no document-hopping.

Convention. Output files live in /Users/carmendavids/WilliamDavid/LLM-Council-Setup/ in the named subfolder. Customer-facing assets (landing page, reel, Typeform) live in their respective tools. Suppliers + leads live in this dashboard's local storage AND a Notion mirror you set up in Build #6.

01 · Quote Builder Sheet
Mon 4 May PM · Defend Wednesday begins here
60 min

What it is

A Google Sheet that takes 4 inputs (package, pax, kids, single-supp count) and outputs full quote line items: per-pax price, single-supp surcharge, child rate, total, deposit due (= flights 100% + 10% land), balance due date, supplier bank, commission earned. The single point of failure for the entire 90 days.

Step-by-step

  1. Open Google Sheets. New blank sheet. Rename "WD Quote Builder v1".
  2. Tab 1 — Packages. Columns: ID · Name · Lodge · Nights · Land cost pp · Flight cost pp · Transfer pp · Inclusions · Single-supp R · Child rate R · Supplier · Supplier bank · Comm % · Notes. Pre-fill 4 Zanzibar packages.
  3. Tab 2 — Quote. Top: dropdown for Package ID (data validation pulling from Packages tab). Inputs: Adults, Kids, Single-supp count, Travel start, Travel end. Outputs (formula-driven): Sub-total, Single-supp total, Child total, Grand total pp, Grand total all, Deposit (flights × pax + 10% × land × pax), Balance, Balance due (travel-start − 45 days), WD commission.
  4. Tab 3 — Quote text. A single cell with a =CONCATENATE or =TEXTJOIN formula that produces a paste-ready WhatsApp quote in plain English, populated from the Quote tab.
  5. Lock the formula columns. Share with no one. Pin the file.
Claude agents to use
general-purpose — paste the column list above and ask: "Write the Sheets formulas for the Quote tab. Each formula should reference the Packages tab via the Package ID dropdown. Output the formula text I can paste cell-by-cell." 15 minutes saves 45 of yak-shaving.
Where it lives
Google Drive · WilliamDavid folder · file name: WD Quote Builder v1. Bookmark the URL. Add to phone home screen as a Sheets shortcut.
02 · Master Zanzibar Architecture
Tue 5 May · 8 hrs
8 hrs

What it is

The one-time architecture document that turns custom packaging from a 3-hr-per-quote slog into a 5-min fill-in. 3 lodges × 3 price tiers, transfer partner negotiated, flight pattern locked, single-supp pre-baked, inclusions mapped.

Step-by-step

  1. 0900–1000: Pick 3 lodges. One 3-star (R8–10k pp land), one 4-star (R14–18k pp), one 5-star (R22–28k pp). Carmen has stayed at all 3 OR has direct GM contact.
  2. 1000–1100: WhatsApp each GM: "Hi [name], I'm building a featured Zanzibar package for WilliamDavid. Can you confirm your peak/shoulder/off-peak rates Sept 2026 – Mar 2027, single-supp, child rate, plus a WilliamDavid inclusion (transfer / dinner / upgrade)? I'll have 6–10 bookings/month, supplier-aligned deposits, your banking details on every quote."
  3. 1100–1300: Lock flight pattern. Cape Town → Zanzibar via Joburg (Mango/Airlink/Kenya Airways). Pick the consistent option. Note ticketing partner.
  4. 1300–1400: Lunch.
  5. 1400–1500: Negotiate transfer partner. One named operator on the island; daily rates locked.
  6. 1500–1700: Build the architecture doc (Google Doc in strategy/packages/): per-package landing copy, inclusions list, exclusions list, fine print, single-supp policy, child policy, payment terms, cancellation terms, supplier banking line.
  7. 1700–1730: Populate the Quote Builder Sheet with the real numbers from the day.
Claude agents to use
general-purpose — to draft the package landing copy from the rate cards (saves 30 min). council-outsider — once draft copy exists, ask: "Read this Zanzibar package as a stranger on Instagram. What's confusing, what's missing, what feels scammy?" 10-minute review.
Where it lives
strategy/packages/zanzibar-master-v1.md + Google Doc copy. Each package = one section. Reference IDs match the Quote Builder Sheet's Packages tab.
03 · Trust Stack Landing Page
Wed 6 May · 2 hrs
2 hrs

What it is

A single Carrd page that the Yoco payment links route through. Designed to make a stranger comfortable wiring R8–15k pp. Carrying the trust stack (testimonials, ASATA, supplier logos, supplier banking line, founder video, "speak to Carmen first" button).

What needs to be on the page

  1. Above the fold: hero image (your face + Cape Town backdrop), one-line proposition: "Real Zanzibar packages, real supplier rates, deposit goes straight to the lodge." ASATA membership number visible. Founder video (60-sec Loom).
  2. Three testimonials with full names + company / Instagram handle. Photo each. Real, not stock.
  3. Supplier logos row. The actual lodges + airlines you're booking with. Visible proof of network.
  4. "How payment works": single line — "Your deposit pays your flights in full + 10% of your accommodation, paid directly to [Supplier Name]. WilliamDavid is paid commission on confirmation. Your money is never held by us against unsecured inventory."
  5. Three Yoco payment buttons (per package tier). Each labelled "Confirm with [Supplier Name]" not "Pay WilliamDavid".
  6. "Speak to Carmen first" button — direct WhatsApp link.
  7. Footer: ASATA registration, IATA if applicable, T&Cs link, refund policy link.
Claude agents to use
general-purpose — to draft the page copy in your voice (paste your existing Instagram captions as voice samples). council-outsider — once the page is live, send the URL: "Visit this as a cold stranger. Would you wire R12,000? Why or why not?" The friction list it gives you = the priority fix list.
Where it lives
Carrd.co. Custom domain: book.williamdavid.co.za (or use Carrd's free subdomain for v1). Asset folder: strategy/landing/ — keep copies of testimonials, supplier logos, founder-video master.
04 · WhatsApp Reply Templates × 6
Wed 6 May · 60 min
60 min

The six templates you need

  1. Price-ask: First reply when someone asks "how much?". Includes Quote Builder output + supplier banking line + valid-until date.
  2. Single-traveler: Single-supp policy, "I'm happy to roommate-match if you'd prefer."
  3. Family: Child rate, family room policy, kid-friendly inclusions.
  4. Dates-flex: "If you can move ±2 weeks, I can save you R2k per pax — here's why."
  5. Deposit-objection: The honest scripted answer. "This holds [Supplier]'s allocation in your name. It's paid directly to them, not to me. Allocation releases in 72hrs without it. I'm happy to jump on a call before you decide."
  6. Ghost-revival: 5-day silent → "Hey [name], wanted to check in — is the [date] window still open for you, or has the trip shifted? No pressure either way."

Setup

WhatsApp Business app → Settings → Business tools → Quick replies. Each template = one shortcut (e.g. /price, /solo, /family, /flex, /dep, /ghost). Six total. Use labels HOT / WARM / COLD / QUOTED / CLOSED on every chat.

Claude agents to use
general-purpose — give it your last 5 real WhatsApp threads with clients (paste, names redacted). Ask: "Write 6 reply templates in this exact voice for these 6 scenarios. Keep each under 80 words." Output goes straight into Quick Replies.
Where it lives
WhatsApp Business app on Carmen's phone. Master copies in strategy/templates/whatsapp-replies-v1.md for version control + future closer onboarding.
05 · Founding Supplier Outreach Pack
Mon 18 May · 1 hr setup, then ongoing
1 hr setup

What's in the pack

  1. One-page invite (PDF) — canonical text from Ella Doctrine v1.1 §10 / LLM-Council-Setup/strategy/suppliers/onboarding/v2_buyer-funded/one-pager.md. Free entry, no listing fee ever, buyer-funded. R499 founding-member framing killed 2026-05-26.
  2. Onboarding form (Typeform or Google Form) — 15 questions: lodge name, region, type, capacity, peak / shoulder / off-peak rates, single-supp, child rate, transfer partners, payment terms, deposit policy, cancellation, banking details, GM contact, reservations contact, response-time SLA.
  3. WhatsApp pitch script — 3 lines for the cold WhatsApp / LinkedIn DM.
  4. Test-booking script — what you ask the supplier as your verification booking.

Verification process (5 steps — see How It Works tab for full)

  1. Eligibility: Carmen has booked or stayed there.
  2. Onboarding pack sent.
  3. Onboarding form completed.
  4. Test booking — measure response time + accuracy.
  5. Listing + tier assignment in the supplier graph.
Claude agents to use
Use doctrine, not regeneration. Canonical pack already drafted at LLM-Council-Setup/strategy/suppliers/onboarding/v2_buyer-funded/ — one-pager, Typeform 16 questions (incl. data-rights consent), 3-line WhatsApp pitch + 2 variants, "what's the catch?" handler. If a refresh is needed, use the doctrine sentences verbatim from doctrine/ella-doctrine-v1.md §8, §10, §11; do not paraphrase.
Where it lives
Pack PDFs in strategy/suppliers/onboarding/. Typeform under your account. Founding-supplier list in this dashboard's Leads tab AND mirrored to Notion (WD-Verified-Network database) for backup + closer access. Notion schema: Name · Region · Type · Tier · Status (invited / onboarding / verified / churned) · Rate-card-URL · GM contact · Last-verified date · Bookings-sent count.
06 · Reasoning Extraction (per destination)
5 hrs per destination · 4 destinations across 90 days
5 hrs

What it is

The single most important deliverable in the 90 days. The moat. A structured document of how Carmen actually decides — what makes her recommend X over Y, what red flags she watches, what dates she avoids, what trade-offs she accepts. Once written, this becomes Ella's system prompt for that destination.

The 5-hour structure (one destination)

  1. Hour 1 — The map. List every lodge / DMC / transfer / activity in the destination you'd consider for a client. For each: who it's right for, who it's wrong for, your one-line opinion.
  2. Hour 2 — The decision rules. "When a couple asks for X, I default to Y because Z." 8–12 such rules per destination. Include the contrarian ones ("everyone recommends X but I send my clients to Y because Z").
  3. Hour 3 — The seasons. Month-by-month: weather, prices, crowds, events, what to avoid, what's secret. The thing only someone who's been many times knows.
  4. Hour 4 — The traps. Bookings ChatGPT or Booking.com would suggest that you'd never. The wrong lodge for a honeymoon. The lodge that closed in 2023. The transfer operator that's gone downhill. Why each is wrong.
  5. Hour 5 — The voice + the format. 8–12 example client questions and your exact reply. This becomes the few-shot prompt for Ella. Voice is warm, opinionated, Cape Town-honest.
Claude agents to use
The most useful agent here is me directly. Open a Claude conversation. Tell me which destination. I'll run a structured interview for hour 1, hour 2, hour 3 etc. — asking the questions, capturing your answers, organising the output. 2.5 hrs of my time = 5 hrs of structured output because you talk and I write. Alternative: general-purpose agent can do the same with a slightly weaker structure if you prefer to drive.
Where it lives
strategy/reasoning-extraction/zanzibar.md, mauritius.md, etc. Each one becomes a system-prompt section for Ella. Version-controlled. This is the asset. Back up to a non-Drive location (laptop + USB) — it's literally the moat.
07 · Editorial Pitch Email
From Mon 18 May · 90 min per pitch
90 min × 9

What it is

One-page email + sample Blueprint PDF, sent to one named editor at one publication. The pitch is for a 6-pack of WilliamDavid African itineraries, white-labelled or co-branded, R40k–R80k flat. 30–60 day cycle. One yes pays the WD floor that month.

Per-pitch checklist

  1. Identify named editor. Travel editor / lifestyle editor / member-content lead. LinkedIn search → direct email (or via published masthead).
  2. Tailor 60 words. First paragraph = one specific reason this magazine / programme is a fit (a recent piece they ran, a member demographic they serve).
  3. One-page email body. The offer (6 articles), the value (you'd otherwise commission a freelancer at R15k each, this is better), the proof (your 8.5 yrs supplier depth + ASATA + named lodges), the ask (15-min call this week or next).
  4. Sample Blueprint PDF attached. Your strongest existing one. Polished. Photographed.
  5. Follow-up cadence. Day 4 nudge if no reply. Day 11 second nudge with a different angle. Day 25 final close-out ("happy to revisit later this year").

Target list

  • Investec Lifestyle content team
  • FNB Private Wealth Travel
  • Discovery Vitality Travel partnerships
  • AmEx Centurion editorial
  • Visi magazine
  • Wanted magazine
  • House & Garden SA
  • Sawubona (SAA inflight)
  • One independent SA travel publication of choice
Claude agents to use
general-purpose — for drafting the per-pitch tailored email. Brief: "Here's the editor's name and one piece they ran recently. Write a 200-word pitch for our 6-pack of WilliamDavid African itineraries. Voice: warm, professional, specific." council-outsider — read each draft as the editor would. Would you reply? What needs changing? council-expansionist — "is there a sharper angle for this specific publication?"
Where it lives
strategy/pitches/editorial/[publication]-[date].md for each draft. Sent log in this dashboard's Leads tab (under a "B2B Pitches" filter, or as a parallel sheet). Email sent from carmen@williamdavid.co.za. Reply log in Gmail.
08 · R299 Launch Reel
Wed 3 Jun · 5 hrs
5 hrs

The shot list (per Outsider's brief, do not skip)

  1. Carmen's face, opening to camera, Cape Town backdrop visible.
  2. Hand-held phone showing the Carrd landing page being filled in (NOT a screen recording — a real human typing on a real phone).
  3. The Yoco payment screen confirming R299 paid.
  4. A real WhatsApp message from a named lodge GM ("Hi Carmen, October 12-15 confirmed for the Smiths") — visible on screen, supplier name readable.
  5. The PDF arriving in the inbox, opened, scrolled. Supplier names a viewer can Google highlighted.
  6. Carmen back to camera, signing off: closing line.

The script (60–90 sec, slow tempo)

"Most travel sites lie about prices. ChatGPT will send you to a lodge that closed in 2023. I built this so neither of those things happen to you. Tell me about your trip, pay R299, and within 24 hours you get a plan from someone who's actually stayed at every lodge in it. Real names. Real rates. Real WhatsApp confirmations from real GMs. If it's not the right lodge, I'll tell you. If it's not the right month, I'll tell you. If you want me to actually book it for you afterwards, I'll do that too. The link is in the bio."

Claude agents to use
general-purpose — to draft 3 alternative scripts in different tones (warm / direct / story-led), so you have options on the day. council-outsider — review the final script: "Does this make me trust her, or does it make me suspicious?"
Where it lives
Master video file: strategy/marketing/r299-launch-reel-v1.mp4. Upload to: Instagram Reel (pinned), TikTok, YouTube Shorts, LinkedIn (with longer caption).
09 · R299 Plan-My-Trip Stack
Across F3 · 10–15 hrs solo OR R3k contractor
~12 hrs OR R3k

The architecture

Carrd page → Typeform (12 questions) → Yoco payment → Make.com webhook → Claude API → Google Docs PDF template → emailed to client.

The 12 Typeform questions

  1. Your name + WhatsApp + email.
  2. Where do you want to go?
  3. Approximate dates (or "flexible — show me my best month").
  4. How many adults? Kids? Ages?
  5. Total budget (R) — be honest, not aspirational.
  6. Vibe in 3 words (e.g. "barefoot, slow, romantic").
  7. Three deal-breakers ("no buffets, no busy beaches, no early flights").
  8. Dietary requirements / accessibility needs.
  9. Three holidays you've loved + why.
  10. Single supplement needed? (yes / no / pair me).
  11. Travel insurance required? (yes / I'll handle it).
  12. Anything else you want me to know?

The Claude prompt (the brain)

This is the part Carmen writes herself. It has 4 sections: (a) System role ("You are Carmen of WilliamDavid Travel. You write opinionated travel plans..."); (b) Reasoning extraction for the relevant destination (pulled in based on the user's "where"); (c) Format — the exact PDF structure expected; (d) Voice constraints — what to avoid, what to use, the rude-but-warm tone.

The Google Docs PDF template

One Google Doc set up with merge fields ({{name}}, {{destination}}, {{lodges_picked}}, {{rationale}}, {{flight_pattern}}, {{cost_estimate}}, {{cta}}). Make.com fills the merge fields, exports as PDF, emails the client.

Claude agents to use
general-purpose — for the Make.com configuration and Claude API integration code. Brief: "Build me a Make.com scenario that takes a Typeform webhook, sends the answers to Claude API with this prompt template, parses the response, fills a Google Docs template, exports PDF, sends via Gmail." I (in chat) — for writing the master Claude prompt. We pair-write it: you give voice samples, I structure. R3k contractor — best option per Executor: 1 week, you get the prompt + PDF template right and the contractor wires plumbing. Saves Carmen 12 hrs of yak-shaving.
Where it lives
Carrd page: plan.williamdavid.co.za (custom subdomain). Typeform: under your Typeform account, named "WD Plan-My-Trip v1". Make.com scenario: named "WD R299 Plan-My-Trip pipeline". Claude prompt master: strategy/r299/system-prompt-v1.md. PDF template: Google Docs in strategy/r299/template/. Customer purchase log: this dashboard's Sales log + Yoco dashboard.
10 · First B2B Pitch (Loom + Google Doc)
F5 · 90 min
90 min

What it is

Pitch to mid-sized SA tour operator (R20–80M revenue). Wedge: R15k setup + R5k/mo for the WhatsApp triage AI. 30-day pilot. Not a deck — a Loom video + a 2-page Google Doc.

The Loom (5 min)

  1. Open with a screenshot of their public WhatsApp Business chat (real). Empathy: "Here's what's eating your team's day."
  2. Live demo: same query → your prototype Ella → opinionated reply in 8 seconds with verified rates.
  3. The wedge offer: "R15k setup, R5k/mo, 30-day pilot, you keep the IP we tune for you."
  4. The trust line: "8.5 years of supplier trust, you don't have to build it. Embed in 2 weeks."
  5. The close: "Book a 20-min call this week — I'll show you it running on your inbox."

The Doc (2 pages)

Page 1: The wedge offer (numbers). Page 2: The 30-day pilot terms (what they get, what you get, exit clauses, success metric).

Target list (start with three)

  • One mid-sized SA tour operator Carmen has worked alongside.
  • One DMC drowning in inquiries.
  • One outbound boutique agency.
Claude agents to use
council-expansionist — to sharpen the upside framing for each specific buyer ("how do I make this irresistible to [name]?"). council-contrarian — to stress-test the offer ("what's the objection they'll have, and what's the counter?"). general-purpose — to write the 2-page Doc.
Where it lives
Loom in your Loom account, named "WD B2B pitch — [target name]". Doc in Google Drive, strategy/pitches/b2b/. Pitch tracker = this dashboard's Leads tab with status "B2B-pitched". Calendar follow-ups in Carmen's calendar with "WD-B2B" tag.

Phone Access · This Dashboard

Add to your iPhone home screen

  1. Save dashboard.html to iCloud Drive (Files app) — or AirDrop / email it to yourself.
  2. In Files app → iCloud Drive, tap the file. It opens in Safari.
  3. Tap the Share button (square with arrow up) at the bottom.
  4. Scroll down → "Add to Home Screen" → Add.
  5. Name it "WD Engine". An icon appears on your home screen.
  6. Tap it. Opens full-screen, no Safari chrome. Checkboxes, scoreboard, leads, notes — all save locally on your phone.

Want a real URL like engine.williamdavid.co.za? Tell me — Netlify deploy, free, 5 minutes. Auto-updates whenever we tweak.

Leads & Sales

Every inbound, every quote, every closed booking. Tracked here. Cumulative totals at the bottom feed the Weekly Report and the Sunday-Night Scoreboard.

0
Total leads
0
Closed
R0
Commission earned

Packages

Edit, add, or rename. The Quote Builder pulls from this list. Numbers default to placeholders — overwrite when Tuesday architecture is done.

Add a Lead

All Leads

Sales Log

Auto-populated from CLOSED leads. Add manual entries for non-booking revenue (R299 Plan-My-Trips, supplier MRR, licensing, B2B pilots).

Weekly Report

Auto-generated from your data. Run it Sunday 8pm — copy or email yourself. The dashboard does the local analysis; the Claude /schedule agent (set up below) does the deeper meta-review.

Generate this week's report

Pulls live from: day count · steps ticked · scoreboard inputs · leads + sales · checkpoint notes · giving notes. Compares to plan targets. Surfaces pivots.

Tap "Generate Report" to produce this week's snapshot.

What the report tells you

The deeper meta-review (optional)

The dashboard generates the data report. Claude does the meta-analysis — what the council would say about your data this week, which rules you're closest to breaking, where the strategy needs adjusting.

To set this up:

  1. Run "Generate Report" Sunday night.
  2. Tap "Email to Me" — opens your Mail app with the report pre-filled.
  3. Forward that email to your Claude conversation, or paste it into a fresh chat with: "Run the council meta-review on this week's data."
  4. I read the data, compare to the council's rules + prior weeks, write back a verdict + this week's pivot list.

If you'd like the meta-review to run automatically every Sunday 8pm, ask for the /schedule setup once you've run a couple of weeks manually and we know the format works for you.

Past reports

Each generated report is archived locally on this device. Tap to view.

The Council

Four rounds of pressure-testing. Every rule and number in the plan is anchored here.

Round 4 · 26 May · Travel Desk as Main Business?

Trigger: a customer told Carmen "that can be your business right? you're a travel insider, not some software guy."

Verdict: Travel Desk is not the main business — it is the instrumentation underneath it. 4 of 5 advisors converge. Forcing it into the main-business slot in 2026 turns Carmen into a solo SaaS CEO serving a dying buyer (SA agencies), racing Anthropic's roadmap, and three years late to the Feb 2028 advisory prize. Door B (data substrate for the advisory practice) is the recommended path.

Three things locked:

  1. The R15–R25k/mo Lane C line is a DIFFERENT product — "Ella Embedded" (intelligence layer embedded in operator ops), not Travel Desk SaaS. Sells reasoning, not plumbing. Survives the Council critique.
  2. The unnamed KPI is rows of verified African travel transaction data per month, attached to Carmen's name and verifiable counterparties. Not MRR. Not customers. Not Replit cancellation.
  3. Data-rights TOS clause is the moat. Without it, every pilot is just software work. With it, every pilot becomes a brick in the 2028 advisory moat.
The one thing to do first: draft the Travel Desk pilot Terms of Service this week with the data-rights clause as the load-bearing element. Then run the R5k non-refundable deposit test on one friendly operator for a July pilot slot. No deposit = not yet a product.

Files: Ella Execution Core — the synthesised single page · Round 4 full council report · Round 4 full transcript

Round 1 · 28 Apr · Original Verdict

Recommendation: Route A (Flash Sale Machine 90 days) → Route B (AI Builder Day 91+). Three non-negotiables: VA Day 15, distribution partnership Day 30, floors written.

Files: Round 1 visual report · Round 1 full transcript

Round 2 · 29 Apr · 22-Question Pressure Test + Flash Sales Review

Six things changed:

  1. Anthropic R30k → R3k. (~10× inflated.)
  2. VA Day 15 is dead. First hire = closer Day 31, conditional.
  3. Holiday Factory rebadge math broken at 5.3% — corrected to 8–12% commission.
  4. Ella, WDT AI MVP, WhatsApp brain are ONE product at different layers.
  5. Carmen's identity has already moved.
  6. Round 2 ships with three changes (warm-list first, kill criterion, supplier banking on deposit page).

Files: Round 2 visual report · Round 2 full transcript

Round 3 · 30 Apr · Cross-Check Peer Review

Six things broke:

  1. The moat is the Reasoning Extraction, not the supplier graph.
  2. "Bloomberg for African Travel" is for your head, not outgoing emails.
  3. 100 suppliers in 7 days is aspirational; 25–40 is real. 100 by 1 July.
  4. Free→paid conversion is 2–5%, not 30%. Realistic supplier MRR: R7–10k. (Model since killed 2026-05-26 — suppliers free at entry, buyer-funded. See Doctrine §5. Retained as historical record.)
  5. The handoff is the trap, not the safety net. Closer Day 60, not 31.
  6. R299 reel needs Carmen's face + named-lodge WhatsApp + Google-able PDF (not "90 seconds" brag); deposit page needs supplier banking visible.
The single new condition that matters most: Day-91 graduation = did one travel professional pay real money for B2B access? Adopted as the load-bearing condition above all others. (First Principles.)

Voices — Key Advisor Quotes

Contrarian
"F-you money in solo SA boutique travel in 2026 is not realistic. Bespoke leisure concierge is a lifestyle business ceiling, not a formidable one. The only travel operators making real money are DMCs with supplier equity, volume corporate / MICE, or media / IP plays."
First Principles
"She's not choosing routes. She's choosing what to stop being. Route A isn't a rescue plan — it's liquidating a dying asset for runway. Calling A a route makes it sticky."
First Principles · on Ella
"African tourism has a coordination failure. Ella is the verified supply layer every African travel seller queries. The moat is supplier trust earned over 8.5 years — no engineer can cold-call a Zanzibar lodge and get real net rates. Internal tool first. Sell access later."
Outsider
"Three times, different words: I don't want to be a travel agent anymore. I don't want to sell Carmen anymore. There's zero value and zero impact in what I do. That's a person asking advisors for permission to stop. Cash machines don't require you to be Carmen-the-travel-agent. They just require execution."
Outsider · on the R299 reel
"The form-fill-PDF-arrives flow IS the scam aesthetic. What needs to be on screen, not in voiceover: Carmen's face, a recognisable Cape Town backdrop, a real WhatsApp from a named lodge, and the PDF showing supplier names a buyer can Google. The 90-seconds brag actively hurts you."
Executor
"The one step that breaks everything if it slips: the Wednesday quote builder + WhatsApp templates. Without it, every other workstream generates inbound Carmen cannot service fast enough, and the funnel clogs at quote stage. Defend Wednesday."
Executor · on hires
"No VA Day 15 — impossible. First hire = closer Day 31 if surplus, R8k + commission. A closer compounds revenue; a VA compounds admin. VA never. Carmen's bottleneck is closing, not admin."
Expansionist
"The 49 leads ARE the asset. Reposition the deposit policy publicly: 'We don't do R1k holds. Our suppliers don't gamble with your booking.' Brand moat. Compounding picture: licensing funds payroll, supplier directory becomes Ella, WhatsApp brain validates the B2B AI thesis, paywall feeds the email list."
Contrarian · on the handoff
"The handoff is the trap, not the safety net. The moment Ella works at scale, every qualified lead funnels into a human bottleneck called Carmen. The system either never gets enough volume to matter, or gets volume and Carmen becomes the constraint she was trying to escape."
First Principles · on Day-91
"The missing single condition is: did one tour operator pay real money for B2B access? Everything else is vanity. If no professional has paid, Route B is theatre regardless of supplier count or MRR."

Session Archive

Strategic conversations preserved verbatim — questions, answers, decisions locked, things deferred. Reference at any time.

2026-05-26 · Ella Execution Core

The only page that matters this quarter.

Synthesised from a full day of strategic work. Holds the four 10-year-old definitions (Claude / OpenAI / Ella / Surge), the six SKUs, the four canonical doctrine sentences, the Ella Embedded sales line, the Day-91 gate scoring, the six-phase execution plan, the daily three-action rhythm, this week's calendar, what we are NOT doing, what we have already benchmarked (don't redo), the five disciplines of formidable product, and the three credentialling artefacts. Doctrine is locked. From here it is execution. Earlier session pages from today have been synthesised into this and removed.

10-yr-old defs · 6 SKUs · 4 doctrine sentences · Day-91 gate · 6-phase plan · daily rhythm · this week · benchmarks log · Click to open
2026-05-18 · Strategic Replan

13-Day Salvage Sprint + 90-Day Slide

Carmen lost 11–17 May to a hardware failure. Original 84-day plan (4 May → 26 Jul) slid 15 days. New clock: 19 May → 16 Aug 2026, Day-91 review Mon 17 Aug. 13-day Salvage Sprint (19–31 May, 9 active working days) layered on top with separate exit gate. Daily Signal Five locked: 20–25 conversations / 6–8 proposals / ~2 closes per active day. Off-Sprint Block 14:00–16:00 Mon→Fri rotation (Travel Desk / RE / Phuket / Editorial+Suppliers / Catch-up). Commission scale corrected: R3–10k per close, not R15–80k — volume model not boutique. May calendar-month target stays R150k but real-win = R80–100k. R299 launch slid to Mon 22 Jun. AVN Verified Plus opt-in opens 1 Aug (R499 forced-conversion model killed 2026-05-26 — suppliers free at entry, opt-in only above the line).

Signal Five locked · Off-Sprint rotation locked · Click for full replan
2026-05-05 · ~9 hours

Travel Desk architecture

Migration off Replit decided. 21-step rebuild plan locked. Security baseline + Red/Blue weekly review + POPIA guardrails captured. The unified hub vision (Phase 2) and the simplification-as-moat thesis (Q3) added to open questions.

13 sections · 26-item audit · Q&A verbatim
How this works

Future strategic sessions will be archived here, one card per session, newest first. Each card links to a standalone HTML page with the full conversation styled for readability.

Ella Execution Core

The only page that matters this quarter — synthesised 2026-05-26, doctrine locked. Open full page ↗

The Rules of the Game

When in doubt, the tiebreaker. Ten rules, each anchored in a council finding.

01
Cash is sacred. Win 1 above all. If a build doesn't move you closer to a deposit landing in Yoco within 14 days, defer it.
02
Defend Wednesday. The Quote Builder + WhatsApp templates are the single point of failure. Executor.
03
The Reasoning Extraction is the moat. Two destinations done deeply > eight destinations done shallowly. First Principles.
04
Cap inbound when you're solo. The handoff is the trap, not the safety net. Until the closer is hired, Companion tier sells from a waitlist. Contrarian.
05
Pay SARS as you earn. Bookkeeper at R2–3k/mo from Day 30. Reinvesting tax money is borrowing from the state at criminal interest.
06
Show the supplier banking, not yours. Money goes to the supplier's account; you are paid commission on confirmation. Outsider.
07
One pitch per week minimum. Editorial. B2B. Either. Pitching does not stop because the inbox is busy.
08
The Day-91 condition is one paid B2B pilot. Vanity is supplier count. Truth is paid usage by a professional. First Principles.
09
No new products until the closer is hired. Build only what is in this plan. Executor + Contrarian.
10
Bloomberg for African Travel is internal. Don't put it in pitch emails. Earn it. Outsider + Contrarian.

If It Breaks

Pre-decided pivots, so you don't have to think when something fails.

If <2 closes by Sat 23 May (Sprint mid-point, after trust stack live + 3 days of cold spend). Pause cold Meta. Escalate trust stack: founder video, 3 named-client testimonials with company + LinkedIn link, ASATA membership number visible. Re-launch upgraded stack to warm WhatsApp database only. No more cold spend until trust gap is closed. If still 0 by Sun 31 May, drop Zanzibar — pivot to private high-ticket booking push to existing list.
If 0 verified suppliers by Sun 14 Jun (mid-F2). The pitch is wrong, not the audience. Use the canonical 3-line WhatsApp script (Doctrine v1.1 §11): free entry ever, qualified enterprise bookings, buyer-funded. Try the top 10 lodges Carmen has personally booked in the last two years on a phone call, not WhatsApp.
If R299 demo reel doesn't convert (≤5 sales by Sun 29 Jun, one week after R299 ships). The headline is wrong, not the price. Test "ChatGPT will send you to a lodge that closed in 2023" against three other openers. Run the 20-free-giveaway motion before any paid spend.
If burnout signals (Carmen >65 hrs/wk for 2 consecutive fortnights). Mandatory 48-hour pause. Drop one workstream from the next fortnight. Editorial pitches go first, supplier recruitment second, RE hours third. Cash work and inbox stay.

The Sunday Ritual (suggested)

  1. Open the dashboard.
  2. Tick any steps you completed this week.
  3. Add notes to the steps you did — what worked, what didn't.
  4. Update the giving section if it's a new fortnight.
  5. Update the Leads tab — close any closed, mark any lost.
  6. Enter this week's three numbers in the Scoreboard. Tap "Archive Week + Reset."
  7. Open the Weekly Report tab. Tap "Generate Report."
  8. Email it to yourself (or paste to Claude for the meta-review).
  9. Read it. Decide one thing to change next week. Write it as a step note.
  10. Close the laptop. The plan is the plan.