Flagship product family

Retail Loss & Ordering Platform

Scan the invoice, catch the price creep, save the expiring stock, and reorder before the shelf is empty.

A margin-protection and ordering system for independent shops, mini-markets and building-material stores. It reads supplier invoices and receipts, matches products to a catalogue, tracks expiry and stock, flags abnormal price increases, and proposes markdowns and reorders — all under the owner’s approval — so the two biggest leaks, waste and price creep, become visible and managed.

Product concept

The Retail Loss & Ordering Platform is a complete margin-and-ordering product for independent retailers whose profit is quietly eroded by two invisible leaks: goods that expire or spoil before they sell, and supplier prices that creep up unnoticed between deliveries. It suits mini-markets, grocers, ethnic and neighbourhood shops, and building-material stores.

It is not just an OCR scanner. Reading the invoice is the entry point; the value is what happens next — matching each line to the shop’s products, comparing the new price against history, updating stock and expiry, detecting what is about to run out or spoil, and proposing a concrete reorder or markdown that the owner approves in seconds.

The unifying insight is that small retailers already have the data to protect their margin — it just arrives on paper, in a language of SKUs and abbreviations, at the back door, and is never captured. The platform turns that paper into a live picture of cost, stock and expiry, and closes the loop with owner-approved actions rather than dashboards nobody reads.

Customer problem

Today a shop owner receives a supplier delivery with a paper invoice full of abbreviated product names, units and prices. It is filed in a drawer, if at all. The prices are never checked against last time.

Because price increases are invisible, the owner keeps selling at the old margin — or worse, below cost — until an accountant notices months later. There is no per-product price history.

Perishable and dated stock expires at the back of the shelf. Nobody knows what is close to expiry until it is waste, so markdowns happen too late or not at all.

Stockouts of the fast-movers lose sales quietly; reordering is from memory and habit, not from what actually sold and what is left. Comparing two suppliers’ prices means digging through paper.

The economic consequence is a shop that works hard on thin margins while silently losing money to waste, price creep and missed sales — none of it measured, all of it fixable.

Before and after

5Today

1. Invoices filed in a drawer, never checked
2. Price increases sold at the old margin for months
3. Dated stock expires unnoticed at the back
4. Reordering from memory; fast-movers run out
5. No number for waste or price creep

Thin margins leaking quietly in every direction.

5With the platform

1. Every invoice scanned and matched at delivery
2. Abnormal price increases flagged the same day
3. Near-expiry stock marked down above a cost floor
4. Reorders proposed from real stock and sell-through
5. A weekly margin, waste and price-creep report

Same shop — the leaks are finally visible and managed.

	Today	With the platform
Supplier price creepillustrative	unnoticed for months	flagged at delivery
Near-expiry stockillustrative	often binned	marked down in time
Reorderingillustrative	from memory	from real stock

Text alternative (accessible description)

1. Today: Invoices filed in a drawer, never checked → Price increases sold at the old margin for months → Dated stock expires unnoticed at the back → Reordering from memory; fast-movers run out → No number for waste or price creep
2. With the platform: Every invoice scanned and matched at delivery → Abnormal price increases flagged the same day → Near-expiry stock marked down above a cost floor → Reorders proposed from real stock and sell-through → A weekly margin, waste and price-creep report

How it fits together

1 / 6

ChannelCustomerAIData storeBusiness rulesBusiness systemStaffExternal service

Stage 1: Paper comes in — An invoice, receipt or order arrives.

Text alternative (accessible description)

1. Paper comes in — An invoice, receipt or order arrives.
2. Read & match — OCR extracts lines; products are matched.
3. Compare to history — New prices meet the catalogue and price history.
4. Compute actions — Rules turn data into reorder and markdown proposals.
5. Owner approves — A human confirms prices, markdowns and reorders.
6. Act & report — Reorders, shelf prices and the weekly report update.

End-to-end workflow

1 / 8

1. Supplier invoice photographedChannel

   Paper from the delivery.

2. OCR extracts every lineAI

   Supplier, product, unit, qty, date, price.

3. Match products to catalogueAI

   Abbreviations resolved; uncertain flagged.

4. Compare price historyBusiness rules

   A staple has risen 12%.

5. Flag abnormal increaseBusiness rules

   Margin protection triggered.

6. Update stock & expiryBusiness system

   Delivery booked in.

7. Owner approves actionsStaff

   Reorder + shelf-price change.

8. Weekly margin reportBusiness system

   Waste and price creep now visible.

ChannelAIBusiness rulesBusiness systemStaff

Step 1: Supplier invoice photographed — Paper from the delivery.

Text alternative (accessible description)

1. Supplier invoice photographed — Paper from the delivery.
2. OCR extracts every line — Supplier, product, unit, qty, date, price.
3. Match products to catalogue — Abbreviations resolved; uncertain flagged.
4. Compare price history — A staple has risen 12%.
5. Flag abnormal increase — Margin protection triggered.
6. Update stock & expiry — Delivery booked in.
7. Owner approves actions — Reorder + shelf-price change.
8. Weekly margin report — Waste and price creep now visible.

Detailed real-world examples

Happy path

A photographed supplier invoice catches a hidden price increase

A neighbourhood mini-market in Vienna run by a diaspora family. — A delivery arrives with a paper invoice; the owner photographs it at the back door.

1. OCR extracts the supplier, each product line, unit, quantity, date and price from the photo.
2. Each line is matched to the shop’s catalogue; two abbreviated names are low-confidence and set aside for review.
3. The new prices are compared to history: one staple has risen 12% since the last delivery — flagged as an abnormal increase.
4. Stock levels and, for dated goods, expiry are updated from the delivery.
5. The system proposes actions: raise the shelf price of the risen staple to protect margin, and reorder two fast-movers now near their reorder point.
6. The owner reviews the flagged price and the two uncertain matches, confirms them, and approves the reorder and the shelf-price change.

Outcome: A price increase that would have quietly eaten the margin is caught the day it arrives, and reordering follows real stock — with the owner approving every change.

Exception path

An uncertain OCR match is verified, not guessed into the catalogue

The same mini-market. — An invoice line reads “PARAD.TOMAT 400G x24” — an abbreviation the system has not seen.

1. OCR reads the line but the product match is low-confidence; the system does not assign it to a random SKU.
2. It proposes the two most likely catalogue products and shows the raw text and price for context.
3. The owner picks the correct product; the mapping is saved so the same abbreviation is recognised next time.
4. Only after confirmation does the line update stock, price history and cost.
5. The learned mapping quietly improves matching accuracy for future invoices from that supplier.

Outcome: Uncertainty is surfaced and resolved once by a human, and the system gets better — instead of silently corrupting the catalogue with a wrong match.

Different market

Expiry tracking turns near-waste into a timely markdown at a butcher

A butcher-deli in Graz with dated fresh stock. — The morning expiry check runs against the current stock.

1. The system lists products within two days of their use-by date and their quantities.
2. For each, it proposes a markdown that clears the stock while staying above a cost floor the owner set.
3. Business rules prevent marking anything below cost and cap the discount depth.
4. The owner approves the markdowns for three items; shelf labels and the POS price update.
5. Sell-through is tracked so tomorrow’s ordering reflects what the markdown actually moved.

Outcome: Stock that was heading for the bin is sold at a controlled discount that protects the cost floor — turning waste into recovered cash with one approval.

Who is responsible at each step

1 / 9

Delivery / paperAIBusiness rulesOwnerSystems

Step 1 — Delivery / paper: Invoice photographed — At the back door.

Text alternative (accessible description)

1. Delivery / paper: Invoice photographed — At the back door.
2. AI: OCR extracts lines — Supplier, product, price.
3. AI: Match to catalogue — Two lines low-confidence.
4. Business rules: Compare to price history — One staple up 12%.
5. Business rules: Flag abnormal increase — Margin at risk.
6. Owner: Confirm matches & price — Owner verifies.
7. Owner: Approve reorder & markup — One tap each.
8. Systems: Stock & price updated — Shelf and POS in sync.
9. Systems: Weekly report updates — Waste and creep visible.

Product modules

Invoice & receipt scanningmvp

Reads paper supplier invoices, delivery notes and receipts with OCR into structured lines.

Product matchingmvp

Matches each line to the shop catalogue, learning supplier abbreviations and flagging uncertain matches.

Price history & variance monitorcore

Keeps a per-product price history and flags abnormal supplier price increases.

Stock & expiry trackingcore

Updates stock from deliveries and tracks expiry/use-by dates for perishable and dated goods.

Stockout & reorder proposalscore

Detects items near their reorder point and proposes owner-approved reorders based on real sell-through.

Markdown recommendationscore

Proposes timely markdowns for near-expiry stock, kept above a configurable cost floor.

Supplier price comparisonextension

Compares the same product across suppliers so the owner buys at the best real price.

Customer & contractor ordersextension

Takes WhatsApp/voice customer or contractor orders and turns them into picking lists and baskets.

Owner reportingextension

A weekly margin, waste and price-creep report that makes the invisible losses visible.

Inputs and integrations

• Photographed supplier invoices and delivery notes
• Receipts and paper price lists
• Supplier catalogues and price sheets
• POS sales data (where available)
• WhatsApp/voice customer and contractor orders
• The shop’s product catalogue and cost data

Users and buyer

Daily user

The owner or a shop assistant who scans invoices, reviews flagged prices and approves reorders and markdowns.

Process owner

The owner, who sets margin targets, markdown floors and reorder points.

Economic buyer

The owner — the person whose money is lost to waste and price creep.

Technical administrator

Usually us (managed) or a family member; building the initial product catalogue from past invoices is the main setup.

Final decision-maker

The owner. Short cycle — the leak and the buyer are the same person.

AI capabilities

• OCR of printed and semi-structured invoices and receipts
• Matching abbreviated supplier product names to catalogue SKUs
• Extracting supplier, product, unit, quantity, date and price
• Detecting abnormal price changes and anomalies
• Understanding voice/WhatsApp customer orders across DE/EN/SQ
• Drafting reorder and markdown proposals and owner reports

Deterministic capabilities

• Prices, costs and totals are recorded facts from the invoice, not inferred
• Markdown floors and maximum discount depth are hard rules — never sell below the cost floor
• Reorder points and quantities follow explicit thresholds and real stock
• Price history and variance thresholds are computed deterministically
• A confirmed product mapping is authoritative for future matching

Object lifecycle

1 / 6

State 1: Scanned — Invoice captured by camera.

Text alternative (accessible description)

1. 1. Scanned — Invoice captured by camera. (→ extract)
2. 2. Extracted — Lines and prices read. (→ match)
3. 3. Matched — Products matched; uncertain flagged. (→ review)
4. 4. In review — Owner confirms and approves. (→ update)
5. 5. Updated — Stock, price and expiry current. (→ action)
6. 6. Actioned — Reorders and markdowns applied.

Human responsibilities

• The owner sets margin targets, markdown floors and reorder points.
• A human confirms low-confidence OCR matches before they update the catalogue.
• The owner approves every reorder, markdown and shelf-price change.
• Final pricing and supplier decisions stay with the owner.

Economic value

• Protected margin because supplier price creep is caught the day it arrives.
• Less waste because near-expiry stock is marked down in time, above the cost floor.
• Fewer lost sales because fast-movers are reordered before they run out.
• Better buying because the same product’s price is compared across suppliers.
• For the first time, a weekly number for waste, price creep and margin — turning invisible losses into managed ones.

No market statistics or financial promises are implied. Any figures in the visuals above are illustrative examples, not measured results.

Risks, limitations and failure cases

• A wrong OCR match corrupts stock and price data — hence low-confidence lines are always human-verified.
• An auto-applied markdown below cost would destroy margin; a hard cost floor and human approval prevent it.
• Faded, crumpled or handwritten invoices reduce OCR accuracy; the system asks to confirm rather than assume.
• Over-ordering on a noisy signal wastes cash and shelf space; reorder proposals are approved, not automatic.
• Building the initial catalogue takes effort; onboarding from past invoices must be part of the managed setup.

Product evolution

Smallest credible first version

An invoice-scanning and price-variance tool: photograph an invoice, get structured lines and a flag when a supplier price jumps, with the owner confirming matches.

Professional product

Full loop — scanning, matching, price history, stock and expiry, reorder and markdown proposals and an owner report, all under approval.

Optional extensions

• Supplier price comparison
• WhatsApp customer/contractor ordering
• End-of-day surplus marketplace
• Deeper POS/accounting integration

Long-term platform

A shared retail-margin layer across many shops that also powers supplier-price intelligence and group buying for independents.

Shared guidance

Generally applicable method — how to validate, pilot, price and keep humans accountable — lives in the shared playbook so these pages stay specific:

• AI vs deterministic automation
• Human approval & accountability
• Measuring economic value
• Evaluation & evidence
• Validate an idea