Multi-site reporting, stocktake variance, rostering against forecast trade, and one honest stock figure across online and in store.
Search for custom software for retail operations in Australia and you get rostering apps and POS platforms. That is the whole first page. Those products are good at what they do, and if one store on Shopify with Deputy and Xero behind it covers your week, you do not need us. The gap opens once you run several sites, because the questions an area manager asks at 8am on a Monday are not ones any single product was built to answer.
Drawn AI was started in 2024 by Liam and Jacob — one out of retail operations management, one out of mining operations. The retail half of this is not a vertical we picked off a list. It is stocktake weekends, wage percentage, the Monday trade meeting, the store that is always short on Fridays, and the delivery that was signed for and never scanned in.
Your POS tells you what sold. It will not tell you whether a store was rostered correctly for the trade it got, whether its variance is theft or a receipting error, whether last week's promotion moved units or only moved margin, or whether store 6 is genuinely up or up because store 9 was shut for a refit. Those answers sit across the POS, the rostering system, the accounting file and somebody's spreadsheet, and get assembled by hand every week by the same person.
We build the layer over the top — connected to Shopify, Cin7, Lightspeed, Xero, MYOB and Deputy rather than replacing them, so nothing changes at store level.
One measure set calculated identically in every store: sales, units, average transaction value, items per sale, wage cost to sales, margin after markdown. Like-for-like handled properly, so a refit, an hours change or a store that opened in March does not quietly break the comparison. Managers see their own site, area managers see their group ranked, and nobody waits for a file. It is dashboard and reporting work on a retail measure set.
Counts done on a phone or scanner in the aisle, reconciled against the POS while the count is fresh, with variance by department and by SKU. The useful part is tracing a variance back to the delivery that was short-shipped, the transfer sent and never received, the write-off nobody recorded, or the override at 6pm on a Thursday.
Trade forecast by day and by hour from your own history, published next to the roster so a manager sees the wage percentage they are committing to before the fortnight is worked rather than after it is paid. Approved hours still flow to Deputy or payroll. We are not replacing award interpretation, only putting the trade curve next to it.
Click-and-collect and ship-from-store fail on the same thing: the number the website believes and the number on the shelf are different, and the gap gets filled by someone ringing a store. We build the reconciliation layer — buffers by store, negative stock flagged rather than hidden, ageing of unpicked orders, and a clear view of which sites can fulfil.
Sell-through by promotion and by supplier with the cost side in: rebates, deals, freight and markdown taken. Short-shipped and late deliveries tracked against the purchase order, so the supplier conversation is a record rather than a recollection. Ageing stock surfaced by store before it becomes a clearance decision.
Opening and closing checks, safe drops and banking, cleaning logs where they apply, promo bay photos with the date stamped on them, and the area manager's store visit report written in the store rather than in the car afterwards. Head office chases exceptions, not everyone.
A paid discovery session with the people who use the numbers — usually an area manager and whoever assembles the weekly pack. We work out which decisions the reporting is meant to support, and say plainly if a product you can buy would do it better.
We map where every figure comes from and how each is calculated, including definitions that disagree between stores and head office: what counts as a sale, when a transfer moves, whether staff purchases are in. Those get settled before anything is built. Where the real problem is process rather than reporting, process analysis is the better entry point.
We agree the views by role. A store manager, an area manager and a CFO need three different screens, not one dense report with filters. You see the layouts before any code exists.
We build it and connect it to what you run — Shopify, Cin7, Lightspeed, Deputy, Xero, MYOB, or the POS database directly. Where the reporting currently lives in a workbook that has outgrown itself, this is the same work as replacing spreadsheets with custom software. Client data we host sits in Australian regions, in Azure Melbourne and AWS Bedrock Melbourne.
Retail changes every quarter — a new range, a new site, a new fulfilment model. Under the subscription we keep adjusting it, and once the reporting is steady it is a short step to analysis of what is driving trade and to customer retention systems built on your own transaction history rather than a loyalty vendor's.
Drawn AI charges a monthly subscription rather than a fixed project fee — minimum three months, billed monthly, 30 days' notice to cancel. What moves the figure is the number of source systems, the number of sites and how many role-specific views you need. You get the monthly number after discovery, before any build starts.
No, and we would usually advise against it. POS is the hardest thing in a retail business to change and the least likely to be the actual problem. We read from Shopify, Cin7, Lightspeed or your POS database and build above it, so nothing changes at the register.
Yes, and it is common in groups that grew by acquisition or run franchised and corporate sites side by side. We map each source to one shared set of definitions, so a sale means the same thing everywhere. Sites on weaker systems support fewer measures, and we tell you which up front.
Most of the work happens away from your stores. The heaviest demand is at the start: a few hours of interviews and access to your systems. We pilot in one or two stores before a group rollout, and avoid cutting over across peak trade or a stocktake. Store staff see a new screen, not a new process.
Possibly, and we will say so. Three sites where one person still holds the detail in their head are usually better served by tightening what you have. The case gets strong when the weekly pack takes half a day, or when a second person needs the same numbers and cannot get them.
A first conversation costs nothing and usually ends with a clearer idea of what is worth building — sometimes that answer is “not yet”, and we will say so.
No lock-in templates. Built around you.