Retail twin · home and trade · independent assembl concept

A big-box hardware store, from the six o'clock utes to the Saturday queue

One fictional store, simulated customer by customer for a year: trade at the desk and the timber yard before nine, DIY through the weekend, which aisles have gaps, whether click-and-collect orders are ready when promised, and what each change would be worth.

The pharmacy mode the research plan names (the fastest-growing large store segment, +11.7%) is not built here. Hardware is the only mode on this page.

live feeds: waiting for the lab
SIMULATED

Today

$0sales so far today
0trade visits
0DIY visits
0in the store now

Trade desk and yard

0waiting at the desk
0utes waiting at the yard gate

Shelves

–on-shelf availability today
0bins empty on the floor now

Checkouts

0waiting at staffed lanes
0waiting for a self-checkout

Click and collectSimulated

0ready for collection
0past their two-hour promise today
01
The year

Twelve months against the hardware segment

Sales by month, trade and DIY

Against the segment

Month by monthexcl. GST

Show the table sales, trade share, availability, lost sales, hours and sales per labour hour by month

02
Friction

Where the store loses sales it had

The six o'clock yard

Empty shelves, by class

Lost sales, where they go

03
What if

Changes to try

All PROPOSED. Switch any on and the whole year runs again on the same customers. The figure on each is that change on its own against the store as it is, as store contribution: gross margin less wages and on-costs, shrink, stock holding, tool costs and write-offs.

The settingsEstimates

Hours by role, a week

The Saturday checkouts

04
Proof ledger

What the changes are worth

Store contribution a year, each change on its ownProposed · simulated

What the twin would trackas it is → all changes

Show the tables each change, and the measures as it is and with every change
    05
    AI and automation

    Where an agent helps, and how far it may act

    Five levels of authority, set per job by the store. Nothing here is running in any store: every row is a PROPOSED assembl concept. The evidence column says who measured a result and whether anyone checked it independently. Where nothing was found, it says so.

    Seven jobs, how far each may actPROPOSED

    Show the table what each job observes, drafts and may do
    06
    Connect your data

    Switch the twin from simulated to yours

    Start with exports the store already has: POS line items, an inventory snapshot, click-and-collect orders, the roster, and a door counter if there is one. Files are read inside this browser tab. Nothing is uploaded or sent anywhere. Customer names, emails and phone numbers are ignored; trade account ids are hashed before anything is kept.

    No files yet. Every figure is simulated, a source or an estimate.
    Sample CSVs to copy (made-up numbers; the personal columns are fake on purpose)

    Generated in this page from one simulated week, so a demo never touches real data. Column names differ from the contract on purpose. The POS file has two broken rows and fake account-holder name, customer name, email and phone columns; the roster names staff. All of those are ignored on load, and the trade account numbers are hashed.

    The data contract: five tables
    07
    Sources

    Figures used, and how far to trust them

    Show the table figure, value, label and where each comes from