Retail and e-commerce
15 released templates. Each passed a check on test data; it is switched on in the panel by the director, and a person decides on every proposal. Example results are illustrative: the names and companies in them are made up.
Kaspi store
The problem it removes: Kaspi listings and replies are neglected.
More detail
What it does: Uses the Kaspi seller account export to suggest what to do with price, stock and buyer questions.
What a person decides: Prices.
For whom: Seller.
How to measure: Kaspi search positions; sales.
Product finder
The problem it removes: Shoppers don't know what to choose.
More detail
What it does: Recommends a product by asking questions.
What a person decides: The final decision on the outcome.
For whom: Shopper.
How to measure: Consultation-to-purchase conversion.
Abandoned cart
The problem it removes: Shoppers abandon the purchase at checkout.
More detail
What it does: A reminder, sent with the customer's consent.
What a person decides: The text.
For whom: Shopper.
How to measure: Recovered carts.
Example result: “Aidos, you left Nike Pegasus sneakers in size 42 in your cart — that size is still in stock. If you're unsure about fit or delivery, message me and I'll help. You can check out using the same link.”
Product descriptions
The problem it removes: Descriptions are missing or copied.
More detail
What it does: Descriptions and specifications.
What a person decides: Publishing.
For whom: Staff member.
How to measure: Time per product listing.
Example result: “Listing: DeLonghi Magnifica S coffee machine. Title: ‘Coffee machine with milk frother — fresh bean-to-cup coffee in a minute.’ Description: 3 paragraphs — who it's for, what it does, how to care for it. Specs from supplier data: 15 bar pressure, 1.8 L tank, 1,450 W. Kazakh version — same meaning. Not in the data: warranty period — check with the supplier.”
Product assortment
The problem it removes: Slow-moving stock takes up the warehouse.
More detail
What it does: Sales analysis and recommendations.
What a person decides: Decisions.
For whom: Manager.
How to measure: Inventory turnover.
Marketplaces
The problem it removes: Each marketplace is managed separately.
More detail
What it does: Combines orders and buyer questions from marketplace exports into one list and says what to do first.
What a person decides: Replies.
For whom: Seller.
How to measure: Unanswered orders and questions.
Example result: “Action: answer 3 buyer questions by 18:00 today — 2 on Kaspi and 1 on Wildberries have waited over a day, which is already hurting the store rating; draft replies below. 14 orders today: Kaspi — 9, Wildberries — 5, all being packed. Impact: the rating holds and buyers don't go to other sellers. Risk: on Wildberries there's 1 pair of headphones in stock but 2 orders for them, so one will have to be canceled. Fallback: if a delivery arrives before shipping, no cancellation needed.”
Promotions and discounts
The problem it removes: Discounts are given without doing the math.
More detail
What it does: Calculates a promotion's impact before launch.
What a person decides: The launch.
For whom: Manager.
How to measure: Promotion profit.
Example result: “Action: offer 10% off, not 20%. Why: category margin is 50% of price; at 20% off, profit holds only if sales grow by two-thirds, while past promotions gave +30–35%. At +30% sales, 10% off yields slightly more profit than usual, while 20% off yields about a fifth less. Impact: the promotion brings in buyers without eating profit. Risk: +30% growth is an assumption from past promotions, not a guarantee. Fallback: compare sales with past ones after a week and end the promotion early if the effect is weak.”
Installments and financing
The problem it removes: A shopper asks about installments, and while the clerk looks up the terms, they head next door.
More detail
What it does: Tells shoppers whether an item can be bought in installments or on credit: how many months, from what amount, with or without interest, and how to apply — per the store’s payment terms.
What a person decides: Processing and exceptions to payment terms.
For whom: Shopper.
How to measure: Installment purchases from chat.
Clothing and shoe sizing
The problem it removes: Shoppers guess their size — and one in three clothing orders comes back.
More detail
What it does: Picks a size from the shopper’s measurements and the model’s size chart, and explains what to do if it doesn’t fit.
What a person decides: Exchanges and returns.
For whom: Shopper.
How to measure: Returns due to size.
Example result: “Madina, with your measurements — bust 92 cm, waist 74 cm — size M in this dress will fit. The fabric doesn’t stretch, so if you’re between M and L, go with L. If it doesn’t fit, we’ll exchange it within 14 days — just keep the tag on.”
Back-in-stock alerts
The problem it removes: A shopper asked to be told when an item arrived, and when it did, everyone forgot.
More detail
What it does: Matches the waitlist with incoming stock, finds who to notify, and drafts each a short message: what arrived, how long it’s held, and how to get it.
What a person decides: Holding items and sending messages.
For whom: Shopper.
How to measure: Sales from the waitlist.
Example result: “Good afternoon, Aigerim! The New Balance 574 sneakers in size 38 you were waiting for just came in. We’ll hold a pair until 20:00 tomorrow — pick them up at 150 Abay Ave. or we can arrange delivery. Which works better for you?”
Gift cards
The problem it removes: Before the holidays there are dozens of gift card questions, and every clerk explains the terms differently.
More detail
What it does: Answers shoppers about gift cards: denominations, validity, how to buy and redeem — per store rules.
What a person decides: Extending a gift card or refunding it.
For whom: Shopper.
How to measure: Gift cards sold.
Business purchases: invoice and e-invoice
The problem it removes: A company accountant is ready to pay by bank transfer, but the clerk doesn’t know which details are needed or when the e-invoice will come.
More detail
What it does: Explains to business buyers how to pay by bank transfer: which company details to send, when we invoice and issue the e-invoice (ESF), and how to collect the goods — per store rules.
What a person decides: Invoicing, issuing the ESF, and payment deferrals.
For whom: Business buyer.
How to measure: Business orders without delays.
Example result: “Yerlan, yes, we sell to companies by bank transfer. Send us your LLP’s BIN, legal address, and bank details, and we’ll invoice within an hour. After payment we release the goods against a power of attorney and issue the e-invoice on the shipping date. If you need a supply contract, we’ll prepare one.”
Store end-of-day summary
The problem it removes: The owner only hears how the day went from the clerk’s evening call — and not all of it.
More detail
What it does: Builds the owner a summary from the day’s POS export: revenue and receipts, top sellers, returns and discounts, what’s running low, and what to decide tomorrow.
What a person decides: Purchasing, pricing, and staff decisions.
For whom: Owner, store manager.
How to measure: Owner time reviewing the day.
Inter-store transfers
The problem it removes: One store has stock sitting idle, another has customers asking for it — and more gets ordered from the supplier.
More detail
What it does: Uses each location’s stock and sales to suggest what to move between stores before ordering from the supplier.
What a person decides: The transfer and supplier orders.
For whom: Store manager, buyer.
How to measure: Supplier orders avoided.
Example result: “Action: move 6 pairs of Timberland boots in size 42 from the Keruen mall to the Abay Ave. store — Abay sold 5 pairs this week and has one left, while Keruen sold none with 8 on hand. Hold off on ordering this model from the supplier. Impact: sales don’t stall, and no money goes into an extra order. Risk: the model may pick up at Keruen over the weekend — leave 2 pairs there. Fallback: move them back if they don’t sell.”
Kaspi order cancellations
The problem it removes: Every Kaspi cancellation hurts the store’s rating, and nobody looks into why orders get canceled.
More detail
What it does: Shows from the Kaspi cancellation export why orders are canceled — out of stock, not confirmed in time, shopper changed their mind — and what to fix first.
What a person decides: Stock levels, prices, and order handling.
For whom: Kaspi seller.
How to measure: Share of canceled Kaspi orders.
Example result: “Action: set the real stock level for the Tefal KO-200 kettle in the Kaspi dashboard — 5 of this week’s 9 cancellations were ‘out of stock’ on it. Confirm orders on weekends too: 2 more cancellations weren’t confirmed in time. Impact: the rating stops dropping. Risk: with zero stock, the kettle drops out of search. Fallback: restore the stock level when the delivery arrives.”
Basis: product documentation; the platform code. This page describes what exists in this build, not what is planned.