Availability & inventory
Out of stock by SKU × city × platform, days of cover, estimated lost sales, short-shipment detection. Highest ROI. Start here.
Kafal Business Consultancy · Dubai
Kafal is a Dubai consultancy creating Endcap for brands selling on Amazon, Noon, Talabat Mart, Careem Quik, Carrefour and their own D2C. It will watch every platform, find where money is leaking, explain why, and — inside limits you set — fix it.
Platforms Endcap will watch
01 — The problem we are solving
This is why Kafal is building Endcap. Fifteen to thirty percent commission before a dirham of ads. About two hundred dark stores across the UAE. A typical brand here is on Amazon, Noon, Talabat Mart, Careem Quik, Carrefour and their own site — with nobody watching all of it together.
Marketplace commission before you have spent a dirham on ads.
Dark stores in the UAE. Availability is local, not national.
Platforms a single brand is realistically selling on at once.
Each week spent assembling Monday numbers by hand.
What it actually costs
Revenue is tracked. The cost of earning that revenue, platform by platform and store by store, is not.
Nobody knows until Talabat has already lost the basket.
Not at the same moment, and not in the same store.
What it costs to sell that revenue, per SKU per platform, stays invisible.
So nothing is comparable without a week of manual work.
02 — The product Kafal is building
The difference is the arrow that goes back up. Endcap will not just tell you what is wrong — it will push the correction into the marketplace, inside guardrails you approve.
Amazon, Noon, Talabat, Careem, Carrefour, ERP, WMS and ad consoles — through APIs, authorised portals and crawlers.
ASIN, Noon SKU and quick-commerce IDs tied together. City → locality → dark store. AED and SAR in one metric layer.
Availability, content, search, price, ads, competitors, reviews, true SKU P&L and forecast.
Detect → diagnose → recommend with AED impact and confidence → act. Models compute; the language model only explains.
Bids, listing edits, price moves within a floor, PO signals. Suggest, then Approve, then Auto — earned per action, per brand.
“In a physical store, the endcap is the most valuable shelf in the aisle. Brands pay to own that corner. That is exactly what we sell on the digital shelf.”
Why the product is called Endcap
The morning screen
Four roles, four different first screens, one warehouse underneath. Click through the rail — this is the shape of the product we are building, on sample UAE data.
Console
How the Endcap agent will see
Endcap swallows sales, stock, price and ads from every platform a brand sells on — then turns the leak into an action ranked in dirhams. Choose a marketplace to see what it can actually read.
Six platforms in
Amazon.ae and Amazon.sa — SP-API and the Ads API. Orders, inventory, listings, search queries, bids and budgets. The richest source in the category, and the first place a write-back agent will run.
The agent in the middle
Classical models find the anomaly and size it. The language model explains it in a paragraph a category manager can act on in five seconds — and never invents a number.
Cause and dirham impact inline. Reply 1 to approve, 2 to snooze.
Category, KAM, supply and CFO each open a different first screen.
Nothing writes back until the brand has seen the rule be right.
03 — Nine modules on the roadmap
Each module answers one of those questions with a number, a cause, and a next action. We start with availability. True SKU profitability is the one that gets us into the CFO’s room.
Out of stock by SKU × city × platform, days of cover, estimated lost sales, short-shipment detection. Highest ROI. Start here.
Titles, images, bullets and A+ scored against a rulebook. Flags silent listing changes made by the platform.
Organic and sponsored rank by keyword by city. Share of the first screen versus competitors, and the keywords you are not on.
Your price against MOP and MRP, competitor index, discount depth, unauthorised-seller and price-erosion alerts.
Spend, ROAS, ACOS and bid pacing across Amazon Ads, Noon and quick-commerce consoles. Wasted-spend detection.
New launches, assortment gaps, share movement and private-label encroachment — at store level, not a national average.
Rating trend, theme extraction, early detection of a quality or packaging problem, drafted responses ready to send.
Contribution per SKU per platform after commission, fulfilment, ad spend, returns and damages. The module CFOs actually pay for.
Demand by city and platform with Ramadan and White Friday uplift built in, feeding a PO plan back to supply.
04 — The AI layer we will not skip
Competitors are strong at detection and thin at action. That gap is why Kafal is building Endcap this way.
Anomaly detection over sales, rank, price, stock and reviews — with Ramadan and White Friday calendars built in, so a festival does not fire a thousand false alarms. Plus a rules engine for the known failures: buy-box lost, listing suppressed, ad budget exhausted before noon.
Sales in Dubai Marina fell 22% because rank dropped, because four stores went out of stock, because a purchase order was short-shipped. One chain, four systems, thirty seconds. A signal on its own only creates work.
Raise the bid on “face wash” from AED 18 to 24 → an estimated +AED 21k of GMV over thirty days, at 82% confidence, based on fourteen comparable moves. Without the number, a recommendation is only an opinion.
Bounded write-back: bids, listings, price within a floor, PO signals. A deterministic workflow engine runs the action; the model only chooses which pre-approved action to run. Logged, reversible, kill-switch ready.
05 — Why Kafal, from Dubai
The global players in this category are real. Kafal is building Endcap the other way: hyperlocal availability at dark-store level, true SKU profitability, and backtest-and-holdout attribution on every claim.
Quick commerce is a local business. National averages hide everything that matters.
Where there is no seller API — Talabat, Careem — we say so, and the fix goes as a tracked note to the category manager.
Per SKU, per platform, after every fee. The number finance will actually sign.
A 90-day backtest before Auto, and a live holdout of SKUs the agent never touches.
06 — Who we are building it with
What changed overnight, what it is worth in dirhams, and which action to take before 10am.
The same alert queue, scoped to the platforms they own — with a note ready for the buyer.
Days of cover, forecast with Ramadan uplift, and PO suggestions that never write to the ERP unattended.
True contribution, realised impact versus estimate, and a Monday review deck that writes itself.
Design-partner build · Dubai
We are looking for design-partner brands in the UAE. A conversation with your category and KAM leads — then a narrow, working slice on your SKUs.