← Back to all posts Custom development & AI

AI in retail: only useful when your stock adds up

Three tills, a webshop, and on Saturday night three different stock figures. That is not an AI problem - that is an 'one-stock-figure' problem. Once you have that, AI can do a lot. Three patterns that work, and a few that do not yet.

Also in: Nederlands Deutsch

Saturday night half past eight. The shop closes. Three numbers do not match. The till says 47 bottles in stock. The webshop says 51. The warehouse counts 44. Someone has to spend half an hour working out which is right. Nobody wants that.

That is not an AI problem. That is an “one-stock-figure” problem. AI on three matching wrong numbers makes it wrong faster. But once till, webshop and warehouse run on one stock figure - which is what a well-set-up Odoo does for you - AI opens a few doors.

What AI does in a shop with a webshop

  • Forecast stock across all channels
  • Read supplier invoices and receipts
  • Personalise offers
  • Guide visitors to the right product

What AI delivers in a shop with a webshop

What AI doesHow it worksWhat you get
Forecast stockPredicts how much of each item you need, across all channels.Fewer missed sales, less marking down.
Read invoicesReads supplier invoices and receipts and sets up the booking.Less re-typing, fewer errors.
Personalise offersSuggests promotions based on per-customer buying behaviour.Higher basket, more repeat visits.
Help visitors pickA chatbot or guided sale that lands on the right product in two questions.More conversion online, less doubt in store.

For none of the four do you have to track anything new, provided till, webshop and stock sit in one Odoo.

Honest: not every pattern works yet

For Wijnwinkel Barneveld we built a platform where till, webshop and accounting run on one Odoo, with a loyalty programme across both channels, deposit handling and an age check at checkout. No AI on it yet, but the base is exactly the kind AI sits sensibly on top of.

Because here is the difference. A loyalty programme across shop and webshop on one database gives you a wealth of data: which customer buys what, how often, when. You can put AI on that - predict which customer is responsive to a targeted offer, or which product is about to run out in two weeks. Without that shared base you get a generic forecast that delivers nothing in your situation.

An idea we see elsewhere too

At larger member organisations we see the same pattern: a personal content feed or offer, based on what someone looked at and bought before. A visitor who keeps coming back for white wines does not see a promo for red rioja. That sounds obvious, but for it to actually work, the data has to be right and your system has to be able to show something different per customer.

Another pattern showing up more often in retail: a chatbot that guides new visitors through your assortment. For a wine shop: “What are you eating tonight, and how much do you want to spend?” Three suggestions, one click, cart filled. That is no longer far-off future.

Retail checklist: start now or wait?

Start with AI when…Wait a bit when…
…till, webshop and stock run on one system.…your stock still sits loose in three places.
…you have a loyalty programme or customer admin AI can learn from.…you have no shared customer data.
…something happens often (reordering, booking, questions).…it only happens now and then.
…you can measure the result in margin, stock or time.…you cannot say what a good result is.

A clean Odoo is the foundation. AI is the layer on top, not its replacement.

Why retail custom work now fits the budget

An age check, packaging logic that suggests a case at six bottles, deposit handling that picks up at the right moment - each a focused addition on standard Odoo. These used to be separate projects you weighed up one by one. Today we build them as part of a normal engagement, because AI speeds up the bare code work and we focus on the design.

For the wider story: custom software is affordable for mid-market now.

One stock first, then AI

Start with the basics: is everything in one Odoo? Then you can take on a first AI job. Otherwise, that is your first job. Honestly, that is also where we make the difference - we build the foundation and only then take on the smarter layers.

Read on:

Frequently asked questions

What does AI actually do in a shop with a webshop?

Four things that pay back fast: forecast stock across till and web at once, book supplier invoices and receipts automatically, personalise offers from buying behaviour, and use a chatbot or guided sale to help visitors pick the right product.

When is AI not a good idea yet?

When your stock still sits in three separate places. AI on three numbers that already do not match mostly produces three matching wrong numbers faster. First till, webshop and warehouse on one stock figure, then AI.

Can offers really become more personal?

Yes, if you have shared customer data. A loyalty programme across till and web - the way Wijnwinkel Barneveld runs it on one Odoo - gives AI something to learn from. Which customer buys what, how often, when. That is where targeted promotions sit.

How does AI fit a specialist shop with its own rules?

Well. For Wijnwinkel Barneveld we built an age check in the webshop, packaging logic that suggests a case at six bottles, and deposit handling - each a focused addition on standard Odoo. AI fits the same way: it runs on your own rules and data, not on a generic model that has never seen your trade.

What data do you need to start?

The data you already have. Till receipts, webshop orders, stock and customers already sit in Odoo, if you have it all in there. AI reads straight from that. No extra dashboards or integrations needed.

Recognize this from your own setup?

A 30-min scan turns hunches into a concrete view, what stays standard Odoo, what becomes custom, what doesn’t need code at all.

Get in touch ← Back to blog