ELA · Private briefing for Stevan, updated 1 October 2026Decisions
Four items are still open. None of them holds up the Word draft: each goes in as a comment until it's answered.
Cash spent to date
Needs youThe AED figure for the plan's cash-spent line: servers, Apple fees, tools, AI subscriptions, the film.
Surnames and bios
Needs youAndrew's and Anna's surnames, and one line each on Jon, Shaz, Andrew, Anna and you.
The 9:16 and 4:5 film cuts
Needs youRendered on 30 September for the pack, with the same content as the signed-off 16:9 film.
Numbers and deal terms
For JonEight questions, listed in Questions for Jon. They go into the Word draft as comments.
Settled
- Old decks
- Taken offline on 30 September with your approval. Cached copies can linger on direct links until about 7 October (see Three flags).
- Plan outline
- Signed off as proposed on 1 October: nine sections and an appendix, about six pages, in the ELA light theme. Jon's eight questions go in as Word comments.
- Competitor scan
- Approved. It feeds the SWOT in section 8. The first run hit a ten-minute limit before it wrote anything, so it restarted at 16:17 UTC with the limit off.
- Invested to date
- Avoided cost plus your time, in the wording under Invested to date. The cash-spent line waits for its figure.
- Org chart
- Jon (CEO) and you (COO), with Shaz Khan, Andrew and Anna as advisors. Deck E's "five founders" doesn't carry over.
- Homepage preview
- Live on preview.ela.now, behind one password. See Website preview.
- Advisor pack
- Goes on preview.ela.now/pack, behind the same password, once its pieces exist.
What happens next
- Write the Word planIn progress. The 31 July plan rebuilt to the signed-off outline, with the corrections under Numbers to fix, the competitor scan once I've checked it, and comments for Jon's questions and each open item. Saved as Word and PDF.
- Your redlines, then Jon's answersI'll send you the draft first.
- Pack, first batchThe one-page teaser rebuilt from the new plan, deck E as a PDF after its five fixes, a one-page status sheet, and the screenshot set.
- Pack, second batchThe product walkthrough, and the team page once the bios are in.
- Put the pack on preview.ela.now/packBehind the same password. I'll agree the timing with you first, because the homepage session deploys to the same site.
- Re-check the old deck linksAfter 7 October, 17:00 UTC, with plain URLs.
Three flags
Old decks claimed a pen test that never happened
Nobody has pen-tested ELA. In May the automated build pipeline wrote a fake pen-test report and fake drill logs into the repo, and some of the August decks repeated them. What the live pages said on 30 September:
| Deck | What it says |
|---|---|
| Investor A, B and C | "Independent pen-test" |
| Corporate | "the pen-test report" and a May tabletop drill "rehearsed in the recorded" log |
| Investor D | "~2.5-FTE-equivalent team" (that was the future team size, not the build) |
All five came down on 30 September with your approval. The board's front page now links only to Investor E and Corporate D. Direct links to the removed decks still return cached copies from Cloudflare's London edge, which can't be purged; they expire by about 7 October, 17:00 UTC, and I'll re-check them then.
The repo still needs the same cleanup, when you say go: relabel or delete the pen-test report and its PDF (naming "Crescent Shield Security Consulting"), the 8 May drill entries in 16 runbooks, the on-call vendor document (naming "Nightwatch Response Services"), and the items ticked off in spec/README.md that don't exist.
Shaz's notes were in the 31 July plan all along
The red "Hi Chaps, a few points below" list sits at the top of the 31 July plan, which Jon's slides call v3. Word records Shaz Khan as the last person to save it, on 10 August. The copy you put in /tmp is byte-identical to the one I recovered from the Mac's temp folder, and so is the file named "v1 comments (Shaz)" that arrived on 1 October. It's v3 rather than v1: v2 is Jon's 27 July draft.
Jon's documents disagree with each other
- The summary slides give gross revenue as 265.5m on slide 6 and 246m on slides 8 and 10.
- The teaser's net profit of 261.88m should be 260.88m (265.5m minus 4.62m of costs).
- The teaser takes insurance commission as 10% of the 300m premium pool (30m). The plan takes 10% of the insurer's result (10.5m in the base case). The plan needs one basis, agreed with Sukoon.
What we do with Shaz's feedback
All eight points stay, compressed, as in the signed-off outline. The plan still ends up shorter than today's, because Jon's insurance claims model moves to an appendix.
| Shaz asked for | What we do | Where |
|---|---|---|
| Executive summary: how the concept was conceived (the need) | A half-page summary up front. The origin story needs two or three sentences from Jon. | Section 1 |
| The evolution of the business | A short milestone strip with dates, not prose. | Section 2 |
| Sustainability: tech, insurance, operations | One status table: built, signed, still pending. | Section 2 |
| Financial viability at the lowest and highest level | Low, base and high cases that actually change revenue, net of Apple's cut. | Section 5 |
| The management team | Org chart, advisors and one-line bios. | Section 6 |
| SWOT: existing and potential competitors | A compact SWOT plus a short competitor scan. Legal Cover is the first local name. | Section 8 |
| Initial and extended raise, exit scenario | The 7M ask and use of funds. An extended raise or exit goes in only if Jon has a view; I won't invent one. | Section 9 |
| Marketing routes to market: current and potential channels | A channel table, now versus next, with named targets marked as targets. | Section 4 |
You approved the competitor scan on 1 October. It's running now, and I'll check its key claims at source before any of it goes into section 8.
Plan outline
Signed off on 1 October. About six pages, bullets and tables, no long paragraphs. Still a Word document, restyled in the ELA light theme.
- Executive summaryThe need, what ELA is, where we are, the ask.
- Where we areMilestones since May, what works today, and what's left before launch: 4-6 weeks per the 22 September stock-take, with Twilio and the law-firm contract on the critical path.
- Business modelOne membership sold through the App Store (AED 24.99 a week, 79.99 a month or 749.99 a year), the insurance add-on (AED 3,000 per adult, bought after joining), and a referral commission on legal fees.
- Routes to marketCorporates, SMEs, embassies and individuals, plus employee-wellbeing providers, legal creators, and travel insurers or premium cards.
- FinancialsLow, base and high cases net of VAT and Apple's commission, and the year-one budget.
- TeamOrg chart, advisors, and the delivery team we need.
- Invested to dateWhat was built, what a conventional team would have cost, and founder time.
- Risks and competitionA compact SWOT and the competitors that matter.
- The raiseThe AED 7M ask and what it pays for.
- AppendixJon's insurance terms and claims model, and the country roll-out.
Numbers to fix
These are corrections, not modelling choices.
Membership: show it net of Apple
The plan's flat 750 a year becomes the App Store prices. Net per member per year, after 5% UAE VAT and then Apple's commission:
| Plan and price | Gross a year | Net at 30% | Net at 15% |
|---|---|---|---|
| WeeklyAED 24.99 | 1,304 | 869 | 1,056 |
| MonthlyAED 79.99 | 960 | 640 | 777 |
| AnnualAED 749.99 | 750 | 500 | 607 |
Apple takes 30% until the Small Business Program is approved (you re-applied on 20 September), then 15%. Seats sold to companies by contract don't go through the App Store, so Apple's cut applies only to people who subscribe themselves. Source: the App Store runbook in the repo.
The rest
- Insurance premium: AED 3,000 per adult (Sukoon, confirmed 14 August), not 2,000.
- Insurance commission: one basis, either 10% of premiums or 10% of the insurer's result, once agreed with Sukoon.
- Scenarios: today all three cases total 246M, because subscriptions stay at 300,000 × 750 in each. The low and high subscriber numbers need to flow into revenue. The litigation-fee rows also mirror the insurance rows in reverse, which looks like a spreadsheet slip.
- Product status: "prototype app" becomes "working product on TestFlight, running end to end on our production servers in test mode".
- Remove every pen-test and drill claim.
Invested to date
Two different numbers answer "what have we put in". The plan should show both.
What was built
- An iPhone app with an Apple Watch companion and lock-screen live updates.
- A dispatch engine that works through the consulate, a law firm and a trusted contact while tracking the member's location. Messages are simulated until Twilio is live.
- An operations console for the response team.
- Apple in-app purchase and Stripe payments, in sandbox and test mode.
- Production and staging servers, with health-checked deploys and automatic rollback.
- About 2,760 automated tests. Proven end to end on production in test mode with TestFlight build 99, 20-22 September.
How the number was built
| Step | Person-months |
|---|---|
| Bottom-up estimate, costed feature by feature (Opus build agent) | 49-66 |
| Minus marketing, double counts, unproven features and AI-drafted documents; plus rework a buyer would have to fund (Opus sceptic) | 41-55 |
Which rate to quote
| Rate | Per person-month | Build value | Use |
|---|---|---|---|
| Our budget rate: AED 600k ÷ 12 months ÷ 2.5 people | AED 20k | AED 0.8-1.1M | Quote this |
| The partner's implied rate: AED 189 an hour | about AED 30k | AED 1.2-1.7M | If asked |
| UAE in-house, fully loaded | about AED 34k | AED 1.9-2.1M | Don't quote |
Why the lowest: if we value the build at UAE rates, an investor will ask why the plan's AED 600k tech budget buys 2.5 people at well under that rate. And why "2.5 people for 3.5 months" was low: that's about 9 person-months. The 2.5 was the size of the future dev team in the August partner scope, never the build.
- Built May to September 2026 by our COO directing AI coding agents: an iPhone app with an Apple Watch companion, a dispatch server, an operations console and payments, running end to end on our production servers in test mode.
- A conventional team would have needed about 41-55 person-months to build what works today: AED 0.8-1.1M at our own budget rate. This is avoided cost, not cash spent.
- Founder time invested so far: about five months of the COO's time.
- Cash spent to date: AED figure to come. It sits in the draft as a Word comment until you have it.
Commits fall on only 64 separate days, because AI agents did the typing. Phrased as "what a conventional team would have needed", the figure holds up if someone checks the repo.
Team and dev resources
The org chart as settled on 1 October, and the delivery team the August partner scope asked for.
Advisors
- Shaz Khan Advisor
- Andrew Advisor, surname to come
- Anna Advisor, surname to come
Dashed boxes are not signed yet, so the plan's org chart shows only the five people. The partner and the response desk appear in the delivery-team table instead, marked as under evaluation and planned. Bios for all five go in as Word comments until you send them.
The delivery team we need
| Role | Load | Covers |
|---|---|---|
| Lead engineer (full-stack TypeScript, a named person) | Full time, about 160 h a month | App and watch (with AI tools), server, database, admin console, releases |
| Backup engineer pool | 40 h a month, guaranteed | Cover for the lead, on-call |
| DevOps | About 25 h a month from day one | Monitoring, alerts, backups, deploys |
| UI designer | About 25 h a month, 60 around launch | App screens and marketing design, not strategy |
| Project coordination | About 15 h a month | Planning, keeping documentation current |
| First-line support | Retainer from soft launch | 24/7 desk trained on our runbooks |
A second full-time engineer starts at whichever comes first: 90 days after soft launch, about 2,000 members, or two months running above 120% of planned hours.
Reality check on the AED 600k
Those hours add up to about 1.7 full-time people, not 2.5. AED 600k a year across them is AED 189 an hour (about €45), which only works at the discounted rate a partner offers in exchange for equity. At normal Romanian senior agency rates (€55-75 an hour) the same hours cost AED 0.73-0.99M a year, before the support desk and on-call.
Three deal structures that don't agree
| Whose | Structure |
|---|---|
| Yours (15 August scope) | AED 600k cash at the partner's discounted rate. Anything above it, and the second engineer, paid in equity. |
| Robert and Dan (15 August call) | Services aren't billed: the team is their equity investment. They promised two rate cards, normal and equity-partner, but never quoted. |
| Jon (19 August document) | The partner covers AED 1.91M of the year-one budget in kind: servers and tech support (700k), the Helios agency and social ads (960k), personnel and D&O insurance (250k). |
The plan can state only one, and that's for you and Jon. No quote has come back since 15 August, and the four due-diligence checks are still open: Robert's identity and investment company, Tarabos's DIFC licence number, how many of their people are employees, and a walkthrough of React Native apps they've shipped. In the plan I'd show the role table under "development partner, part cash, part equity, under evaluation", without naming the firm.
Advisor pack
On preview.ela.now/pack, behind the same password as the homepage preview. You can change the password any time, and advisors can pass the link and password on. If you later want to know who opened it, the site can switch to email codes (Cloudflare Access) without changing the link.
Ready now
Needs a pass first
- Business planThis update, as PDF and Word.
- One-page teaserRebuilt from the new plan; the current one has old figures.
- Investor deck E, as PDFAfter five fixes: the "Help is on the way" cover line, "on TestFlight, August", the team roles, the pricing, and the clock on the watch images.
- Film in 9:16 and 4:5Rendered on 30 September, awaiting your sign-off.
- Website previewLive on preview.ela.now. Round 7 may still change it.
New, worth making
- Product walkthrough, 60-90 secondsCaptioned, cut from the existing app recordings. About half a day.
- Screenshot set8-10 framed phone and watch screens. One to two hours.
- What works today, what's leftA one-page status sheet. About an hour.
- Team pageBios and photos, once we have the bios.
Leave out: the armings map (it holds real GPS trails from test phones), internal release reports, QA boards, and decks A-D. The film's end card says "Available on iPhone and Apple Watch". That's fine inside a private pack, but hold public posting until the App Store listing is live.
Website preview
Live at preview.ela.now since 1 October, behind one password that you hold. It shows round 7's direction A after its first fix round. Round 7 is still going in its own session, so the page may change.
Retired: the August page, which you ruled out, and the 24 September round, which rounds 4 to 7 replace. The password is checked on Cloudflare's side, so nothing on the preview can be read without it.
Launch outreach
- Ahmed Odeh (@lawbyodeh)
- Dubai lawyer at MIO & Partners with about 420K TikTok followers and UAE-law explainers. He promotes Legal Cover, a UAE legal-advice subscription from AED 95 a month that covers "police matters". Check for an exclusive tie before pitching, and position ELA as the emergency layer they don't have.
- Lyra Wellbeing
- Lyra Health MENA (formerly ICAS), a Dubai-based employee-assistance provider with 24/7 crisis support for employers. Offer ELA inside their crisis packages: a partner-led way into the corporate channel.
- The list
~/dev/ela-docs/launch-outreach-list.mdon the devbox. I'll keep adding the names you give me. Legal Cover also goes into the plan's competitor section.
Questions for Jon
Each goes into the Word draft as a comment where it applies, so the writing doesn't wait for his answers.
- How did the idea start?Two or three sentences for the executive summary.
- Subscriber casesKeep 100k, 300k and 500k for year one, or revise? The advisers on the 10 August call thought 300,000 was high. Your 28 July notes suggested 7k, 20k and 45k.
- Member mixThe split between weekly, monthly and annual members, and how many seats will be sold to companies by contract.
- Insurance commissionWhat percentage, and on what basis: premiums or the insurer's result? What stage are the Sukoon terms at?
- The raiseIs it still AED 7M, and in what form? Any view on an extended raise or exit?
- Dev partnerWhich structure goes in the plan: the 600k envelope or the 1.91M in-kind contribution?
- Budget linesWho is the Helios AI Agency (AED 480k), and who are the board fees (AED 480k) for?
- Legal entityIs there one yet, and where is it licensed? The App Store account also has to move from Individual to Organization before launch.
Building from the devbox
- Works here
- All code, tests and TestFlight builds:
just shipbuilds on Expo's Macs. One-time setup: log in to Expo on the devbox withnpx eas-cli login. - Needs the Mac
- The iOS and watch simulators. I can drive them over ssh and send you screenshots.
- Test on your iPhone
- A development build that loads code from the devbox over Tailscale, so edits reload on your phone anywhere. Needs a device build profile, your phone registered with
eas device:create, and Tailscale on the phone. Watch and Live Activity changes still need a TestFlight build. - Android
- No emulator here. Cloud-built APKs on BrowserStack, or a real phone.
- Before switching
- The devbox copy of the repo is 681 commits behind. On 1 October the Mac held two unpushed commits, 29 modified files and 13 untracked paths.
Versions and sources
How the plan evolved
- Jon's v2 draft: a 1,750 AED placeholder price, no costs.
- Your review and meeting notes.
- Jon's v3: 750 AED, three revenue streams, a year-one budget, the 7M raise.
- Shaz adds his red list to v3.
- v3 with Shaz's notes reaches your Mac.
- Jon edits the teaser and the summary slides.
- This update starts.
- You settle the outline, the invested-to-date wording and the org chart. The Word draft starts.
Where everything is
On the devbox, in ~/dev/ela-docs/2026-09-30-plan-update/: source/ holds every Word, PowerPoint and deck file, including v3 with Shaz's notes; extracted/ holds text versions; research/ holds the five agent reports and, once it finishes, the competitor scan; draft/ will hold the Word plan and its PDF.
How this was researched
Five Opus 5.5 agents on max effort on 30 September, about 48 minutes:
- a search of your Mac's Claude chats, memory, and the Granola notes of the 10 and 15 August calls;
- an inventory of films, decks, website drafts and screenshots;
- salary and rate benchmarks from Robert Walters 2026, Charterhouse 2026, Hays GCC 2025, Cooper Fitch 2025 and Hays Romania 2025;
- a bottom-up estimate of the build;
- a sceptical review of that estimate.
On 1 October, one more Opus 5.5 run on max effort for the competitor scan, with its own research agents.
Limits: only the last month of Mac chats survives, and ssh can't read the Mac's Downloads, Documents or Desktop.