85% of UAE consumers surveyed have already used AI while shopping. Only 32% said they would currently trust an AI agent to complete checkout for them. [1]

The more I look at those two numbers, the more I think the interesting part of AI shopping is not whether people will use AI. They already are. It is how much of the buying decision we are willing to hand over.

Most conversations about AI and marketing are still focused on us, the marketers. How do we use it to write faster? Make ads? Research customers? Create images? Automate work? But customers are using it too, and that side of the shift is probably more important.

People are already asking AI to check reviews, compare prices, find gift ideas and generally make buying decisions easier. So we are happy to ask AI what to buy. We are much less comfortable letting it actually buy it.

That gap is where things get interesting.

I don't think we need to imagine some distant future where an AI assistant wakes up and decides what groceries you need. The change happening right now is much simpler.

Imagine I need a new pair of headphones. Previously I would search Google, open Reddit, watch two YouTube reviews, check Amazon and then somehow end up buying a completely different pair three hours later.

Now I can start by asking ChatGPT. Which headphones are good for long flights? Which have the best noise cancellation? What are people complaining about? Is the more expensive one actually worth it?

Visa's UAE study found that 60% of people surveyed had used AI to check reviews or product ratings, 59% to compare prices and 55% for gift ideas. [1]

That matters because these are not random things we do after deciding what to buy. They are part of how we decide. AI is starting to sit somewhere between “I need something” and “I know what I want.”

And that means brands now have to think about what happens when the person helping the customer choose is sometimes a machine.

There is a big difference between asking, “What are the best hotels in Dubai Marina under AED 800?” and saying, “Choose one, book it and charge my card.”

The first feels useful. The second requires trust.

That is probably why the 85% and 32% numbers are so far apart. People are already comfortable using AI as an adviser. We are much earlier in becoming comfortable with AI acting on our behalf.

And there are some very normal reasons for that. What if it misunderstood me? What if it buys the wrong thing? What if I change my mind? Who is responsible if something goes wrong? How much control did I actually give it?

The same Visa study found that 60% said alerts from their bank or payment app would make them feel more secure when something looks suspicious, while 33% said a familiar trusted logo at checkout would make them more comfortable. [1]

The technology can make buying easier. The uncomfortable part is that trust decides how much of the buying we are actually willing to hand over.

Discover is when AI helps me find brands or products I might not already know.

Compare is when it helps me look at price, reviews, features, availability or whatever else matters.

Choose is when it starts narrowing those options down and recommending what I should actually get.

Buy is when I give it permission to complete the transaction.

I wouldn't treat this as a perfectly linear journey. People are messy shoppers. We change our minds, go back, ask friends, open twelve tabs and sometimes buy the first thing we saw anyway.

But the distinction is useful because AI does not need to control the entire transaction to change marketing. If it is already affecting what people discover, compare and eventually choose, then it is already affecting buying decisions.

Let’s say somebody asks, “Which accounting firm in Abu Dhabi is good for a small business that needs bookkeeping and corporate tax?” There may be ten perfectly good firms. But an AI assistant still has to figure out which ones are relevant enough to recommend.

It needs information. What does the business actually do? Who does it serve? What does it charge? Where does it operate? What do customers say about it? Are there useful reviews? Does the website explain the service properly? Does information from other places support what the company says about itself?

The same applies if someone is comparing hotels, headphones, restaurants, insurance, gyms or almost anything else.

This is why I think the interesting part of AI search goes beyond trying to “rank on ChatGPT.” The more useful question is whether your business is easy enough to understand, compare and trust in the first place.

A surprising amount of marketing is still bad at this. Websites hide prices. Product descriptions say a lot without really explaining anything. Different pages contradict each other. Old information stays online forever. The brand claims to be the best at something while giving you no reason to believe it.

This is where the UAE gets particularly interesting. We are not only seeing people use AI for research. Companies here are already testing what happens when AI gets permission to actually transact.

In November 2025, Mastercard launched Agent Pay in the UAE with Majid Al Futtaim and Dataiera. As part of the pilot, AI-powered agents could help people search, discover and transact, starting with booking tickets at VOX Cinemas. It was Mastercard's first Agent Pay transaction outside the US. [2]

Then in December, Visa and Aldar completed a transaction where a customer used an AI agent through the Live Aldar app to pay real-estate service charges. The customer still went through a consent process before the agent completed the transaction. [3]

And earlier this year, Visa launched its Agentic Ready programme in the UAE. It says nearly 60% of UAE businesses surveyed are either strongly interested in or already exploring these types of commerce use cases. [4]

I wouldn't read this and assume everyone is going to let AI control their wallet next year. These are still early examples. But there is a fairly clear direction here.

First AI helped us search. Then compare. Now companies are figuring out how to let it transact safely. The question is how quickly customer trust catches up.

I don't think marketers need to suddenly create a 40-slide “Agentic Commerce Strategy.” Most businesses have much more basic work to do first.

1) If someone asks AI what we sell, will it understand us correctly?

2) If someone asks it to compare us with three competitors, is there enough public information available to make that comparison?

3) Are our prices, products, locations, policies and other important details clear and up to date?

4) Is there enough proof outside our own website for someone to believe the claims we make?

5) Where would AI struggle if it tried to help a customer buy from us today?

I like these questions because they are useful and breaks through the noise if AI shopping moves much faster than expected. A customer also wants clear information. A customer also wants proof. A customer also wants to know what something costs and what happens after they buy.

Preparing for AI might just force companies to finally fix things they should have fixed for people anyway.

I built a simple AI Shopping Readiness Check you can run on a product, service or website. It looks at five things: Discovery, Clarity, Comparison, Proof and Purchase.

The point is not to guess exactly how people will shop in 2030. It is to see how prepared the business is for a world where AI is increasingly involved in the decision.

You can download the fillable check below and run it on your own business.

Download the checklist from here

I don't think people will hand over shopping completely to AI. Shopping is too human for that.

We buy things because somebody recommended them. Because we like the brand. Because something looked good in a store. Because TikTok convinced us we needed it. Sometimes because we were bored at 1 AM and apparently that is enough reason to order something.

AI will become another part of that process. And in the UAE, that process seems to be moving pretty quickly.

People are already using AI to discover and compare what they buy. Companies are now experimenting with letting it complete the transaction too.

For marketers, I think the question is fairly simple:

If AI helps your next customer decide what to buy, will it understand enough about your brand to recommend you?

That is probably worth finding out now.

1. Visa - Stay Secure Study 2026, UAE

85% of UAE consumers surveyed used AI while shopping; 32% would currently trust an AI agent to complete checkout.  Open source

2. Mastercard - Agent Pay with Majid Al Futtaim, UAE, November 2025

Mastercard Agent Pay pilot in the UAE, including VOX Cinemas.  Open source

3. Visa & Aldar - Agentic payment implementation, December 2025

UAE real-estate service-charge payment completed through an AI agent with customer consent.  Open source

4. Visa - Agentic Ready programme, UAE, May 2026

Programme and UAE business-interest data related to agentic commerce.  Open source