iShoes · the meeting

A sixty-minute call · 22 August 2026 · screen shared

iShoes · the ninety days, and how it runs.

This page is the hour. It sits beside the plan you already hold, and every figure on it appears in the proposal that follows this call by email today.

1Findings recap 2The plan 3How it runs — AWS, Git, agents 4What it returns 5What it costs 6What we need from you
iShoes × RAAVA

Findings recap · from the report you already hold

Six things that are wrong, the fix for each, and the check you run yourself.

Each row opens. The left cell is what we found, the middle is the named fix, the right is a check anyone in this room can run without us.

ProblemSolutionExpected result — your check
01Pages describing the wrong pageopen
ProblemThe writing on your biggest brand's pages is about a different page, and your shoe polish page tells people about women's sandals.
SolutionThe search and page fixes: each block of writing pointed at the page it belongs to, done in one pass on a copy of your shop.
Expected resultOpen the pages. What each page says matches what it sells.
02The Wilde tile leads nowhere usefulopen
ProblemOn a phone, the Wilde tile on your homepage lands on Panda slippers, not on the brand that sells close to a third of your website.
SolutionThe tile pointed at a proper Wilde brand page, and the brand page built to earn the visit.
Expected resultOpen your shop on your phone and tap the Wilde tile.
03Phone numbers you cannot tapopen
ProblemYour phone numbers are text. On a phone they cannot be tapped.
SolutionTap to call, on the 1300 number and all four stores.
Expected resultOpen the site on your phone and tap a number. It dials.
04The size chart trapped in a pop-upopen
ProblemYou keep a size chart up to date. It only exists inside a pop-up, so it never reaches the page.
SolutionThe chart on the page, in words, where a customer and a search engine can both read it.
Expected resultOpen any product page and read the sizing without clicking anything.
05Every email goes to everyoneopen
ProblemEvery campaign goes to the whole list, including the people who have not bought in six months.
SolutionCampaigns built on the customer groups already sitting in your own Omnisend account, run by hand first.
Expected resultOpen Omnisend and look at the audience on the next draft. It is a group, not the list.
06The articles stopped publishingopen
ProblemA hundred and twenty finished articles sit switched off. Your blog on a phone shows three headings and nothing underneath.
SolutionThe blog reactivated and the twenty-nine usable articles published on a weekly rhythm, each read by a person first.
Expected resultOpen your blog, weekly. Something new is on it.

How it runs · the part that is new

Why a server, why Git, why a master database.

Three pieces of plumbing carry the whole plan. None of them needs a technical person in this room to make sense.

AWS + background agents

A server that is always on, with agents that are always working.

Trending articles, product and collection enrichment, SEO updates all have a right time. Short-handed, those windows get missed. A server that is always on, with agents that are always working, stops the missing.

Today · windows missed
hollow = missed
With the agents · windows caught
solid = caught
Git · version control

Google Drive for AI agents and for code.

Every member and every agent shares the same knowledge of every digital task. After a successful run, the templates, rules and examples update themselves for next time.

It also means every change to your theme is kept, so any change can be looked at, and any change can be stepped back.

Master database

A root record of every change.

A root database recording every change to every product, version-managed, so the AI can check its own work and step back if something is wrong.

Nothing ships on a guess. The record is what lets the system correct itself before a customer ever sees a mistake.

Why it matters

This plumbing is what makes doubling organic revenue reachable. The honest word is reachable, never promised — and the numbers behind that word are further down this page, unchanged from the proposal.

The plan · month one

Month 1 — Stabilise. The thirty days, week by week.

Month one is the iShoes website: the fixes, the search and AI-search work (SEO and GEO), the catalogue enriched, the UI/UX corrected — with the infrastructure underneath. Each band opens. Nothing in it is new scope: it is the plan you already hold, laid on a calendar.

W1The foundations stood upopen
  • Claude seats set up for the team — who holds which seat is in the costs frame below.
  • The AWS server stood up, in your own account.
  • Git and version control connected to the Shopify theme, so every change from here is kept.
  • The agents configured — their memory, their skills, and the guard-rails that stop them.
W1A requirements meeting with each team memberopen

A short session with each person on the team, which becomes a shared AI knowledge base of where information lives. We also ask each person for the list of manual work they repeat, because that list is what the system absorbs.

W1–W2The move off the old hostingopen

n8n and the workflows migrated off the old hosting onto the new server, and the master database layer created — the root record every later step writes into.

W2–W3The site, fixedopen
  • Pages, brands and landing pages corrected — the twenty-one defects closed.
  • Products and collections enriched: templates, metafields and meta objects filled.
  • Related collections attached to products, and the homepage reordered.
  • The Shopify sales channel completed.
W2–W4The articles, publishing againopen

The blog reactivated and articles publishing from the archive — twenty-nine usable of the hundred and twenty. And one in three articles is written for the season six months out, so January's school shoe month is being built for now, not in January.

W3–W4Lifestyle imagery, made with your teamopen

Lifestyle imagery generated with the Runway subscription you already hold plus fal.ai, referenced to how competitor sites use consistent multimedia rather than product shots alone. Every mockup is reviewed by your team before anything ships, and each approved look is saved as a reusable workflow and template, so what your team already makes gets a library to build on.

AlwaysThe standing ruleopen

Everything runs by hand first, is measured, and only then is automated — through GitHub workflows, n8n, or Shopify Flow, whichever fits the task. Nothing is automated on a guess.

The plan · month two

Month 2 — Replicate. If month one earns it.

Month two is not asked for today. It is what the proven work does next, and the day-25 review is where it is decided. It carries the same weight as month one — about 84.5 hours — because every channel opens with its own discovery session and its own requirements research.

What replicates
  • Each channel is discovered and researched on its own terms — what a Pinterest board needs is not what the blog or the email programme needs, and the templates, metafields and meta objects for each are designed before anything publishes.
  • The proven iShoes workflows run on Just Bee and Panda — the same catalogue work, the same content programme, the same email plan.
  • The organic channel content pipeline starts, with engagement as the measure, not volume.
  • A monthly plan per channel, so each channel knows what its month holds.
What connects
  • Loyalty and abandoned-basket automations connected — after their manual test, never before.
  • Each workflow arrives on the other stores already measured, because month one measured it on iShoes.
The honest framingMonth two is convincing, and it is optional. It runs if month one earns it, and the day-25 review is where that call is made.

The plan · month three

Month 3 — Systematise. The system you keep.

Month three turns what worked by hand into the written method, and hands it over — at the same weight again, because development is work: the Raava OS app is built to your own customisations, not unwrapped from a box.

What you hold at day ninety
  • The written method — what worked by hand becomes the system you keep.
  • The Raava OS app, developed to your customisations — the workflows run by hand in months one and two, running under your approval.
  • A person with entry-level software skills can maintain and improve it.
  • The workflows run under your approval, on your accounts, in your name.
What becomes possible only now
  • The team-rate option becomes available — only now, because the method now exists. More hands cannot follow a method that is not written down.

What it returns · every figure from the proposal, unchanged

What a year of this brings back.

Hosting, off your bill

$1,421–$1,526 a year

About $1,474. The $2,292 is what you pay now — it is what the saving comes out of, not the saving itself. Year one is smaller, because moving takes hours and those hours are real.

Repeated manual work

Absorbed, then measured

Week one collects the list of manual work each person repeats. The system absorbs that list task by task, and each one is measured before it is automated — so the hours saved are counted, not claimed.

Publishing windows

Caught, not missed

Trending pieces, seasonal articles and catalogue updates land in their window because the agents are always on. You check it on your own blog and your own feeds.

The sales expectation · money from search · four positions

Where you check thisGoogle Analytics, Reports, Acquisition — what organic search brought in over the last twelve months. These four positions are printed in the proposal with their workings, and nothing here changes a single one of them.

What it costs

Month one, priced. Months two and three, scoped.

A$5,000 A$6,500 MONTH 1 · STABILISE

Months two and three are shown above as scope, and are priced at the day-25 review — after month one has shown its work, not before.

The equivalent-hours device

Every month is about 84.5 hours of deciding on the AI layer — work that would take a traditional team 200 to 300 hours, worth $20,000 to $30,000 at about $100 an hour.

And the real hours are printed honestly beside the multiplier. The months weigh the same because month two’s channel discovery and research, and month three’s development of the Raava OS app to your customisations, are work of the same order as month one’s build — about 253 hours across the ninety days:

84.5 Month one · your website: fixes, SEO and GEO, enrichment, UI/UX · itemised
~84.5 Month two · your channels: discovery, research, organic channel content, the sibling stores · estimated
~84.5 Month three · your system: automations, the written method, the Raava OS app built to you · estimated
Claude for the team

The seats

The team plan · proposed

1 Claude Max + 4 Pro

One Max seat for the lead operator, four Pro seats for the working team, set up before onboarding begins. Every seat is a person who can run the workflows themselves.

The minimum start

1 Max only

Enough for the pilot if the team is smaller to start. The lesser path: one operator, and the team watches rather than works.

AI tooling budget

The budget, and what you already hold

A$350 for the pilot phase, spent on the tools we use to generate, render and test by hand, tracked against each workflow so the business case is real.

Then A$200–300 a month steady state.

Plus access to the Runway subscription you already hold — costs nothing new, and is asked for as access, not budget.

A$350 pilot · then A$200–300 a month

Who does the work

The record, and the team.

For a Victorian government agency, Demi built an AI software platform single-handed; three people now run it. The same client's Salesforce app development went from months per feature to two or three days.

Lead Demi Runs the engagement, builds the system, holds the day-25 review.
Graphic design · multimedia JB The imagery, the templates, and everything a customer sees.
Creative director Malinda Has worked with Fujian before. Holds the creative line across the brands.
Full-stack developer Isuru The server, the database, and the plumbing underneath the agents.

What we need from you

Eight things, none of them a build.

Tick them off on this screen as we talk. Each one is access or an hour of time, and the plan starts when they land.

The Claude seats1 Max + 4 Pro as the team plan, or 1 Max as the minimum start.
The pilot AI budgetA$350, tracked against each workflow.
Runway accessAccess to the subscription you already hold. Costs nothing new.
Read access to Google AdsRead only, so the measuring can start.
Omnisend accessSo drafts can be staged in your own account. A person at iShoes presses send.
A requirements meeting with each team memberShort sessions, week one. They become the shared knowledge base.
The list of repeated manual workFrom each person: the day-to-day tasks they repeat. That list is what the system absorbs.
A seat in the marketing meetingSo the work follows your calendar, not a guess at it.

The next step

The day-25 review decides month two.

After this call

The proposal follows this call by email today. Month one is priced on this page; months two and three are priced at the day-25 review, after the work has shown itself. Nothing here expires, and nothing is agreed until it is signed.