For: Paul Lunghis · From: Anthony Booth, Articulate AI · Dubai · 17 May 2026
Reach me: - WhatsApp · +971 50 508 4451 - Email · anthonybooth71@gmail.com - Live online · https://articulate-ai.work/client/Paul
For Paul Lunghis · Articulate AI · 17 May 2026
Twenty years of work sits in your phone. Most of it dormant.
This is a one-off, two-week project. I take your WhatsApp, enrich every contact with public Dubai property and life-event data, score against fifteen real-estate segments, and hand you a list of fifty contacts worth calling this month. Each one with full context and a drafted opener in their language.
One-off project · Two weeks · 50 callable leads
A contact isn't a name. It's a future commission on a different clock.
Past buyers cycle every 5–7 years. Tenants renew every 12 months. Off-plan investors transact every 2–3 years. Landlords rotate portfolios. Brokers and conveyancers refer both ways. Enquiry-only contacts wait for a life event.
The book you've built is fifteen different relationships behaving fifteen different ways. The job is to know which clock each contact is on, then prompt you at the right time with the right context, in the right language.
Right now that intelligence lives in your head. The proposal is to externalise it, enrich it, and surface it back to you one prompt at a time.
Fifteen segments. Each behaves differently. Each pays differently.
| # | Segment | Cycle / Trigger | Expected commission |
|---|---|---|---|
| 01 | Past buyers | 5–7 yr upgrade window | AED 50k–500k+ |
| 02 | Past sellers | ~60% rebuy in 18 months | AED 50k–500k+ |
| 03 | Past tenants | 12 mo renewal | AED 10k–60k |
| 04 | Past landlords | Portfolio rotation, 3–5 yrs | AED 50k–500k+ recurring |
| 05 | Off-plan investors | 2–3 yr cycle, repeat | AED 50k–200k/unit |
| 06 | Enquiry-only (never closed) | Life-event triggered | AED 15k–500k+ |
| 07 | Property management clients | Recurring | AED 6–25k/yr/unit |
| 08 | Holiday let / Airbnb owners | Seasonal | 15–20% AUM |
| 09 | Agent partners | Continuous, co-broke | AED 50k+ per deal |
| 10 | Mortgage brokers | Referral exchange | Both ways |
| 11 | Conveyancers / lawyers | Referral exchange | Indirect |
| 12 | Developer sales teams | Continuous | Inventory access |
| 13 | Contractors / suppliers | Reciprocal referrals | Indirect |
| 14 | Social / lifestyle | Network compounding | Indirect |
| 15 | Dead / wrong number | Skip | 0 |
Segments 1–8 are direct revenue. Segments 9–14 are the referral engine. Segment 15 gets archived. The scoring model weights all three differently.
Full segment-by-segment detail in section The fifteen segments below.
Nothing new to buy. Everything you already have, surfaced and joined.
Your stack already holds most of the picture. WhatsApp threads, Gmail history, Calendar cadence, Goyzer notes. The pilot adds three free public layers on top. No new SaaS bills. No data subscriptions.
Free public sources:
Already paid sources (in your stack):
AI runs over the joined dataset to classify each contact for language preference (Greek / French / Italian / English), life stage, life events, and relationship temperature. Every signal feeds the score.
Paid sources (Property Monitor, LinkedIn Sales Navigator, Truecaller, Reidin) are deliberately excluded from the pilot. If a specific one earns its keep after the run, we add it then. Not before.
How a single contact lights up. Illustrative composite, not a real contact.
Sarah H. — past buyer, Umm Suqeim
Drafted opener — copy-paste ready:
Hi Sarah, hope all well. Kids must be ready for Year 7 already. Market's moved a fair bit since 2019, worth a coffee to take stock?
In the deliverable, Sarah's page also includes a copy-paste-ready WhatsApp opener. Every one of the 50 in the final list is delivered in this format. You read, decide, send from your phone, mark sent. Under 90 seconds per contact to review and act.
Full sample brief with the maths in section A sample brief below.
One ranked list of fifty. Each contact, one page.
The output is a single document — your top fifty contacts to call this month, ranked by expected revenue, ordered by ease of re-engage. Each contact gets its own page with the full brief above.
Per-contact page: Segment, history with you, DLD ownership record, comps on their street, life signals from public sources, last spoke, who went silent, expected commission, probability score, drafted opener in their language.
Format: Delivered as a single PDF you can keep on your phone, plus a CSV for Goyzer import. The drafted openers come as separate WhatsApp-ready text snippets you can copy-paste straight from the file.
Goyzer integration: Every contact's enrichment data gets written back to Goyzer as a structured note. Goyzer stays your CRM of record. This is a one-time data injection, not an ongoing sync.
Handover call: One hour at the end. We walk the top ten together, you tell me which to start with this week, I help you tune the openers if needed.
One hour of you, on the sofa, tapping yes.
The thing only you know is who-is-who. The system can read your WhatsApp, your Gmail, your Calendar, the DLD. It cannot read the twenty years inside your head. The pilot is designed so that uploading what's in your head costs you one hour total, not weeks of data entry.
The bulk-confirm pattern:
I cluster the 2,000 contacts into ~40 likely-similar groups before you see anything. Each group has a name-pattern, a chat-content signature, a phone-prefix tell. You see one card per group:
Group 14 of 38 · likely Past tenants — Villa 40 series Chat names contain "Villa 40" · all sent/received lease-related attachments · 47 distinct contacts Sample names: Belal Villa 40 2025 · Marcus Villa 40 · Xavier Villa 40 2024 · Daniela Villa 40 2025 · +43 more Suggested segment: 03 · Past tenants Action: Tap one button. All 47 confirmed in one go.
Forty groups. Each takes 10–60 seconds. Whole 2,000-contact book classified in under an hour. No spreadsheet. No CRM data entry. No typing names.
The voice-note exception: For the top 50 contacts the model surfaces, you optionally voice-note 30 seconds of extra context per contact before I draft the openers. "She's the daughter of George Andreou, family's based in Athens, looking at Palm for the school proximity." That voice note transcribes, extracts the entities, attaches them to her record, and gets folded into her opener. Skip is fine — the model still produces a usable draft from the public signals.
Outside that — zero you. No spreadsheets to fill. No CRM screens to click. No tagging UI to learn. Your existing WhatsApp, your existing Gmail, your existing Goyzer.
Two weeks, end to end.
| When | What |
|---|---|
| Day 1 | WhatsApp export via iMazing. 30 minutes of your time. You hand over the CSV. |
| Days 2–4 | Ingest. Auto-cluster the 2,000 into ~40 segment groups. DLD enrichment pass on the top 500 by raw score. |
| Day 5 | You sit on the sofa for one hour and bulk-confirm the segment groups. Whole book classified in one session. |
| Days 6–9 | Voice profile built from your outbound messages. Scoring model calibrated. Top 50 ranked. Openers drafted in Greek, French, Italian, English. |
| Day 10 | Optional voice-notes from you on the top 50 (under 30 mins). Openers refined with your head-context. |
| Day 11 | PDF + CSV delivered. Goyzer notes written. One-hour handover call, walk the top ten together. |
| Week 8 (optional) | Single follow-up call. What closed, what was useful, what was noise. |
Yardstick: three deals booked from this list inside sixty days that wouldn't have closed otherwise.
Full iMazing path in section How the export works below.
One pilot fee. Nothing new to subscribe to.
| Two-week project · 50 callable leads | AED 12,000 one-off |
| External data costs | AED 0 · free sources only |
| Retainer | unchanged · separate |
One fee. One deliverable. The fifty-contact list on day 11, the optional 30-minute follow-up at week 8 to track what closed. Nothing ongoing. If the list earns its keep, we talk about a second round in three months.
Nothing here asks you to leave Goyzer, change CRM, or buy new SaaS. No new monthly bills.
Full privacy + data posture in section Privacy and data below.
One 30-minute call. Three answers:
I'll have one of those five contacts modelled end-to-end before the call ends. Proves the system live, on your real data.
Ping me:
WhatsApp is fastest — I see those before email.
Two weeks. One deliverable. Fifty contacts in your phone you can act on this month. No new subscriptions. No CRM to learn. Goyzer untouched. Your retainer untouched. One fixed fee. Done in eleven days.
Read it. Push back. Pick.
Anthony Articulate · Dubai +971 50 508 4451 · anthonybooth71@gmail.com Live version: articulate-ai.work/client/Paul
Remaining sections are deeper detail — read whichever earns the time.
01 · One contact, end-to-end
This is what one of the fifty looks like in the deliverable. Composite, illustrative, not a real client. Every contact in your final list gets its own page in roughly this shape.
Sarah H. · past buyer · Umm Suqeim Phase 3
| Field | Value |
|---|---|
| Segment | 01 · Past buyer · cycle window 5–7 yrs · hit now |
| Purchase (Gmail) | 4BR villa, Umm Suqeim Phase 3, July 2019 — offer letter PDF in your inbox |
| DLD public record | Sole owner. No subsequent purchases under her name. No mortgage refinance flagged. |
| Property Monitor comp | 4BR Phase 3 sold AED 14.8M last week (vs her Aug-19 purchase price AED 11.5M, +28.7%) |
| Last WhatsApp | 8 months ago. You went silent. Last message from her: "We must catch up properly when school's back." |
| Chat content scan | "kids starting Year 7" mentioned 3× in last 12 months of chat — typical move trigger window |
| LinkedIn public | Husband: role change · Regional MD · March 2026 (promotion signal) |
| Calendar history | 3 meetings 2019–2020 (sale, handover, follow-up) — none since |
| Language | English (UK) |
| Score | 0.34 · top-decile inside past-buyer segment |
| Expected commission band | AED 240k–340k if upgrade closes |
Past-buyer cycle window is hitting now (5–7 yr typical). Two strong move triggers in plain sight: kids reaching secondary-school age, husband's promotion. A directly-comparable property sold at +28.7% above her purchase price last week — she's sitting on real equity. You went silent eight months ago — that's the low-friction half of the population to re-engage. Your fault, not hers, easy to reset.
Copy-paste ready. Already in your tone, English (UK) register.
Hi Sarah, hope all well. Year 7 already, can't believe how fast that's gone. Market's moved a fair bit since 2019 — a 4BR on your street sold last week well north of what you paid. Worth a coffee to take stock?
If she replies, two follow-up options are ready in the deliverable.
Variant A — soft, info-only. Use if she's not ready: "Drop me a note if you ever want a comp valuation, no obligation."
Variant B — direct. Use if she replies warmly: "I've got two off-market Phase 1 villas coming in next month. Want a heads-up first look?"
Each contact's score is the product of four factors:
Score 0.34 puts her in the top decile of the past-buyer segment in your book.
Three taps on your phone, total time under 90 seconds:
Mark sent in the deliverable PDF if you want to track. The week-8 follow-up call covers what closed.
02 · How your book is partitioned
Your 2,000 contacts sit across fifteen behaviour patterns. Each one transacts on a different clock, at a different cheque size, with a different trigger. The model tags every contact into one of these and scores accordingly.
Cycle: 5–7 year upgrade window. Kids growing, school changes, divorce, relocation, second-home interest. Trigger to re-engage: life event (job change, kids' school stage), neighbourhood comp moving sharply, market shift. Expected commission band: AED 50k–500k+ depending on property class. Opener angle: equity update ("a 4BR on your street sold north of what you paid"), market check-in, comp valuation offer.
Cycle: roughly 60% bought again in Dubai within 18 months of selling. The rest left. Trigger: DLD shows them as current owner of a new address = they bought (sometimes through another agent). Re-engage to win back. Expected commission band: AED 50k–500k+. Opener angle: congratulations on the new place, offer to handle the next move.
Cycle: annual renewal window. Standard tenancy contract runs 12 months in Dubai. Trigger: 30–60 days before lease end (Ejari date), or family expansion / contraction signals. Expected commission band: AED 10k–60k per let, renewal commission often higher because lower acquisition cost. Opener angle: "renewal coming up?", offer rate-check against market, upsell to bigger unit, ownership offer.
Cycle: portfolio rotation. HNW landlords rebalance every 3–5 years. Trigger: new tenant placed (recurring fee), property vacant longer than 60 days, market shift on their building. Expected commission band: AED 50k–500k+ per disposal, recurring management fees on top. Opener angle: portfolio review, yield comparison, off-market acquisition opportunity.
Cycle: 2–3 year handover cycle, repeat-investor pattern. Trigger: developer launch, handover date approaching, secondary-market value moved on their unit. Expected commission band: AED 50k–200k per unit, repeat business is the gold here. Opener angle: new launch heads-up, handover-time decision (rent / sell / occupy), off-market resale interest.
Cycle: opportunistic, life-event triggered. Trigger: any life signal — promotion, baby, move-back-from-abroad, divorce, inheritance. Expected commission band: AED 15k–500k+ depending on what they ultimately buy. Opener angle: light-touch, "circumstances may have changed", warm market update.
Cycle: recurring. Renew, upsell, refer. Trigger: quarterly check-in cadence, end-of-contract anniversary, complaint or compliment. Expected commission band: AED 6k–25k per unit per year recurring + referral upside. Opener angle: performance review, yield update, additional service offer.
Cycle: seasonal. Q4 + Q1 is peak. Trigger: occupancy dip, ownership change, regulatory change (DET licensing). Expected commission band: 15–20% AUM if you co-host, or one-off setup fee. Opener angle: yield benchmark vs neighbouring units, licensing help, hand-off-to-pro offer.
Cycle: continuous, co-broke driven. Trigger: they have a buyer, you have inventory (or vice versa). Expected commission band: 50% of either side, AED 50k+ per deal typical. Opener angle: specific listing match, market intel exchange, dinner.
Cycle: continuous, referral exchange. Trigger: they have a financed buyer needing inventory; you have a buyer needing pre-approval. Expected commission band: referral fees both ways, indirect leverage. Opener angle: specific client referral, lender-rate intel, co-event.
Cycle: continuous, referral exchange. Trigger: transaction in motion needs legal; legal client needs broker. Expected commission band: indirect, deal-quality multiplier. Opener angle: transaction-specific intro, niche expertise (offshore structuring, family-trust ownership).
Cycle: continuous, supply-side leverage. Trigger: launch event, inventory clearance, exclusive allocation. Expected commission band: primary commissions 3–7%, scarce inventory access is the real prize. Opener angle: specific buyer fit, allocation request, partnership renewal.
Cycle: continuous, reciprocal. Trigger: client project pipeline, ad-hoc need, post-handover work. Expected commission band: indirect, referral economy. Opener angle: specific job intro, post-completion check-in.
Cycle: continuous, network compounding. Trigger: event-driven, not commercial. Expected commission band: indirect, but disproportionate referral generation. Opener angle: social, never transactional.
Cycle: none. Skip. Trigger: none. Expected commission band: zero. Action: removed from working pool, archived in Goyzer with reason.
Three signals combine:
After your pass, the segment is locked. The score builds on top of that lock.
Treat all 2,000 the same and you get a generic newsletter. Segment them properly and you get a different message strategy per group: lease-renewal reminders for tenants, comp-update for past buyers, launch alerts for off-plan investors, dinner invites for social. Same database, fifteen different campaigns.
The fifty in the deliverable are picked across segments, weighted toward where the model thinks your near-term revenue lives.
03 · 30 minutes on Day 1
The whole project starts with one CSV — a structured export of every WhatsApp chat on your iPhone. iMazing is the tool. Below is the exact path. I've already run this end-to-end on my own 1,455-chat history, so the friction points are known and pre-solved here.
This is the single trip-wire that breaks the whole flow. If you've ever enabled end-to-end encrypted backups in WhatsApp (separate from iCloud), iMazing cannot read your chats.
On your iPhone:
You can switch it back on after the export. Anthony's recommendation: leave it off until project end so we can re-pull if needed.
imazing.com — free download, free for what we need (just chat export). Mac or Windows both work.
Install. Open. Plug your iPhone into your Mac with a Lightning / USB-C cable. Trust the computer if prompted.
iMazing left panel → click your iPhone name → Back Up (top button).
This pulls everything onto your Mac. 5–30 minutes depending on iPhone storage used. You can use your phone normally during this.
This is the key trick — you do not export one chat at a time. iMazing supports a bulk operation that does all of them at once.
~/Desktop/whatsapp-export-paul/Progress bar runs. 10–60 minutes depending on chat count. Walk away.
Once done you'll have a folder with one CSV per chat (or one bulk CSV with all sessions), plus PDFs, plus media subfolders.
Three options to get it to me:
The data never leaves your control. I work on a copy stored locally on my Mac, encrypted at rest, deleted at project end. See privacy + data for the full posture.
This is rare but happens on very large chats (10,000+ messages).
I process whatever lands in the export folder. Partial extraction is fine — we just re-run the missing portion in a second pass.
Day 2 onwards, no input needed from you until Day 5 (the one-hour bulk-confirm session). Walked end-to-end in the main proposal and the build sequence.
Three things I hit on my own export that the SOP above pre-empts. Sharing for context — you shouldn't see any of these:
If you hit anything I haven't seen, WhatsApp me and we solve it on the spot.
04 · What we use, what we don't
Your contacts are the most sensitive asset in your business. The whole project is built around minimising exposure and keeping you in control end-to-end.
| Source | What we pull | Where it stays |
|---|---|---|
| Your WhatsApp CSV | Chat names, message text, dates, attachments metadata (not content) | My local Mac, encrypted folder |
| Your Gmail (optional) | Subject lines, snippets, sender/recipient. Full body only on threads you flag. | My local Mac, encrypted folder |
| Your Calendar (optional) | Meeting titles, dates, attendees | My local Mac, encrypted folder |
| Dubai Land Department open data | Public ownership records, transaction history | Cached locally, public source |
| Public LinkedIn profiles | Name, role, employer, public posts (no DMs, no contacts) | Cached locally |
| Public news / press | Mentions of contacts in articles, podcasts, awards | Cached locally |
| Goyzer notes (optional) | Existing notes you've already captured on contacts | Read-only |
Processing runs on the Claude API (Anthropic). Specifically:
Reference: Anthropic data usage policy — Section "Data Usage".
UAE Data Law (Federal Decree-Law No. 45 of 2021) — the project operates within the personal-data-processing exemption (data processed by the data subject's appointed agent for the subject's own legitimate business interests). You are the data controller. I am the processor. A short data-processing agreement (DPA) is part of the project paperwork — one page, plain English, signed at kickoff.
GDPR — relevant only for any European data subjects in your book (Greek, French, Italian HNW clients). Posture: same legitimate-interest basis under GDPR Art 6(1)(f), with the additional guard that no profiling decision is automated — every opener is reviewed by you before sending.
At any point, you can ask me to:
You own the data. You own the deliverable. You own the IP in any drafted message. Articulate's only retained right is to reference the existence of the engagement in case-study form, anonymised, with your written approval before publication.
WhatsApp me for anything not covered. The privacy posture is deliberately conservative — UAE HNW clients are sensitive about data exposure, and the project should not create new attack surface.
05 · Frequently asked
The five questions that come up every time I scope a project like this. Direct answers, no hedging.
Eleven days end-to-end. Day 1 you do the WhatsApp export (30 min). Days 2–4 I ingest and run the first enrichment pass. Day 5 you spend one hour on the bulk-confirm session. Days 6–9 I build the score and draft the openers. Day 10 we do the optional voice-notes pass on the top 50. Day 11 you get the PDF + CSV + handover call.
If anything slips, it slips on the exports side — iMazing crashes happen on very large chat histories. Pad to 14 days worst case.
AED 12,000 one-off. Fixed fee, no overages, no data subscriptions. That covers the full eleven-day project, the deliverable, the handover call, and the optional week-8 follow-up call. If the deliverable is useful enough that you want a second round in three months, we'd scope that separately — typical refresh project is shorter and cheaper (AED 6–8k).
The pilot is separate from your AED 7k/month retainer. It's a one-off project that fits inside the Database workstream (#7) of your engagement framework but is invoiced and delivered independently. Retainer doesn't change.
Goyzer is your CRM of record and stays that. What it doesn't do — and isn't designed to do — is fuse external data sources with your messaging history, score contacts by expected near-term revenue, and draft openers in your voice. The project is a one-off intelligence pass that lands enriched records back into Goyzer as structured notes. You're still working in Goyzer day-to-day. You're working in a Goyzer that knows more.
No. That's the design constraint.
You spend one hour at Day 5 (bulk-confirm), optionally another 30 minutes at Day 10 (voice-notes on top 50), and one hour at Day 11 (handover call). After that you send openers from your existing WhatsApp, on your existing phone. The work that the project does — segmenting, enriching, scoring, drafting — happens entirely outside your daily flow.
Three deals booked from the list inside sixty days that wouldn't have closed otherwise. That's the bar. If we hit it, the AED 12,000 fee is recovered roughly 25–50× over on a single mid-tier villa commission. If we don't hit it, we have a precise post-mortem on why (segments wrong, scoring wrong, openers wrong) and the next round is far sharper.
The model detects each contact's primary language from chat content + LinkedIn signals. Openers get drafted in that language, in your voice, by the time the deliverable lands. You'll see English versions alongside for your own review.
The project can run on WhatsApp alone. Gmail / Outlook / Calendar are optional enrichment layers. Each one you can include adds context (Gmail is the richest — lease PDFs, contract attachments, family references). If you're on Outlook 365, that's a connector swap, not a blocker.
Not in v1. The deliverable is built around your personal book of relationships and your voice. Scaling it to a team requires multi-tenant data plumbing and a per-agent voice profile per person. That's a separate, bigger project — let's see if the personal version earns its keep first, then talk about agency-wide.
Yes. WhatsApp me and I'll walk you through my own 1,455-chat run on a call. Same pipeline, same scoring approach, my own dormant list. Easier to assess on real output than in slides.
Two reasons.
One — Dubai's market is still ramping. Past buyers from 2018–2021 are sitting on 25–50% equity gains. The upgrade cycle is hitting now. Six months from now the easy wins are gone.
Two — the public data layer just became usable. Dubai Land Department's open-data portal, public LinkedIn, and large-language-model classification combined for the first time mean a project like this is achievable in days, not months. A year ago it wasn't.
If the deliverable is the right shape for you, the next 90 days is the moment.
WhatsApp the word "yes" to +971 50 508 4451 and we'll book the 30-minute call this week.
— Anthony
Articulate AI · Dubai · For Paul Lunghis · Confidential · 17 May 2026
Live online: articulate-ai.work/client/Paul