Ten store tasks we never do by hand

Comparison table

The facts in one place first. Figures as of July 2026.

The half day that disappears into clicking

Add up how much of a working day in a store goes on routine some time. Book in a delivery. Write descriptions for the new pieces. Set up a weekend promotion and switch it off again on Monday. Regenerate a feed. Issue invoices. Check where the revenue is even coming from. None of it is hard, there is simply too much of it and it never stops.

That is the layer we wanted to take off the plate. Not strategy, not creative work, but the repeated grind that eats half a day and nobody enjoys. So I wrote up ten tasks we no longer do by hand in a store on Behio, because Tom, the AI partner, handles them. For each one I say what he actually does, so it is not just a nice sounding promise.

Catalogue: descriptions, categories, prices

I will start with the catalogue, because that is where manual work piles up fastest. The first task is descriptions. You import three hundred products and half of them have an empty description field. Tom drafts text from the specifications that are in the catalogue, not out of thin air, and tells you outright to read what he wrote. How to approach AI descriptions with SEO in mind is covered separately in the article on generated product copy.

The second task is sorting into categories. New products belong somewhere and clicking through them one by one is a chore. Tell Tom to drop twelve new candles into the Candles category and it is done. The third task is analysing prices and margins. Tom goes through the price list, works out the margin from cost and selling price and shows you where money is leaking. He will never run a bulk price change on his own, though, only after your confirmation, and he shows you the old and new prices first. We wrote that dialogue up in the article on repricing a catalogue.

Selling: promotions, orders, invoices

The fourth task is promotions and discounts. A scheduled buy two get one free on a category, valid from Friday to Sunday, switching itself off afterwards. You state it once in a sentence and there is nothing to delete manually on Monday. The fifth task is processing orders. Tom moves states from received to shipped and can tag orders so a team can split the work.

The sixth task is invoicing. He issues an invoice or a proforma from an order and keeps an eye on whether payment arrived. That is work you would otherwise be retyping from one system into another, whereas here both sit over the same data. It sounds like small stuff, but it is precisely this small stuff that separates a calm day from an endless afternoon in the admin.

Stock, feeds and data

The seventh task is booking stock in. Forty shirts arrived, you tell Tom, and he records the goods receipt and updates availability, so the product shows as in stock on the site straight away. Stock in Behio is not a quantity field on a product, it is a full record of movements and receipts. The eighth task follows from it: the watchdog. On a restocked item Tom switches on notifications for customers who were waiting while it was sold out, and they get an email saying it is back.

The ninth task is comparison shopping feeds. In the Czech market that means Heureka and Zbozi.cz, the price comparison engines that dominate Czech ecommerce roughly the way Google Shopping does in English speaking countries. They want an XML feed with prices, availability and categories, and it has to stay current. Tom regenerates it on request, and the same applies to whichever feeds matter where you sell. The tenth task is data. Behio has its own cookieless analytics that counts every visitor rather than only those who clicked through a cookie banner, and Tom builds a dashboard over it from a single sentence, something like build me an overview of revenue by source. A dashboard is assembled from queries over metrics and breakdowns, so it is not a locked preset you cannot change.

And one extra: smart offers

I will admit ten is just a round number, because there are more than 1000 actions. But one extra task is worth a mention, because the competition does not have it. Smart offers, meaning automatic personal discounts based on a visitor's behaviour. Tom can not only read the rules but prepare them from an analysis of your catalogue and your data.

The way it works is that switching the feature on kicks off an AI analysis in the background which proposes rules tailored to you, with reasoning: a product with many views, returning visitors and zero purchases gets a proposed discount after the fifth visit. And always in silent watching mode first, where nothing is given away and it only counts how many people it would reach. How the whole mechanism works is described in the article on smart offers.

What stays with a human

So this does not come across as the owner merely sipping coffee. The ten tasks above are routine, not running a company. Deciding what range to carry, how to price it or which market to enter is not something AI does for you. It will not conjure product photography either. And the output is still worth checking, particularly copy and prices.

Count on AI operations consuming credits too, of which every plan has a set volume. Exhaust them and only the AI features stop, not the store. Here is how I see it: AI should take the repeated work, not the responsibility. Where exactly the line runs, and what AI will not do, I wrote up in a separate article on the limits of AI in ecommerce.

The verdict

Common questions