AI agents are starting to reshape how people discover and decide what to buy. In a joint survey by BCG and Moloco of 238 senior marketing leaders, 67% said they expect a high level of AI-driven disruption to their vertical’s consumer journey. That disruption is increasingly influencing how decisions are formed, often before a shopper ever reaches a retailer’s site. Think of a shopper asking an AI assistant for the “best running shoes for trail running,” receiving a shortlist with pros and trade-offs, and landing on a product page.
Facing this shift, many retailers think “AI” means adding a smarter chatbot to their website. But consumers rarely use these tools, and they don’t change the way people naturally shop. What’s emerging instead is ambient intelligence, AI that steps in automatically when a shopper hits a moment of friction, and steps away when it’s not needed.
Why chatbots are the wrong starting point
Retailers often feel that chat widgets are an obvious place to start because they’re familiar and are easy to point to and say, “this is our AI feature”. But this is the wrong model for how people actually shop.
Shoppers don’t abandon twenty years of search-and-browse habits just because chatbot “Bob” pops up in the corner of a page. If getting help requires a shopper to stop what they’re doing and actively seek it out, most simply won’t. And when retailers see that chatbots aren’t getting much usage, they often conclude that shoppers don’t want AI assistance. In reality, it’s usually more of a reflection on how the help is delivered, at the wrong time or in the wrong way, not whether it’s wanted at all.
As a result, retailers can fall into the trap of writing off AI’s potential based on tools that were never designed to fit naturally into the shopping journey in the first place.
Ambient intelligence – help when and where shoppers need it
A better model is something known as ambient intelligence: assistance that is embedded and context-aware, surfacing only when it’s useful but fading into the background when it’s not. The interaction is still conversational, but the emphasis shifts from asking for help to receiving it at the right moment.
Instead of waiting for a shopper to click a chat button, ambient intelligence responds to signals already present in the journey, such as pausing on product specs, repeatedly switching between similar products, or hesitating at the point of decision. It’s largely invisible by default, allowing people to browse as they always have. When behaviour suggests uncertainty, it steps in with just enough context to keep the journey moving.
What this looks like in practice
A good shop assistant doesn’t wait to be asked for help. They notice you standing in the aisle, weighing two items, eyes reading over the different specs, and step in to offer exactly the context you need to decide.
Ambient intelligence works the same way online. Imagine a shopper arriving on a product page for a monitor arm. They scroll down, scroll back up, toggle between variants, pause on the specifications, then hesitate. After a brief pause, a prompt appears: “This arm works with desks 0.5 to 2 inches thick. Your monitor up to 27 inches is supported. Most customers install in 10 minutes.” The shopper can dig deeper if they want, or add to basket and move on, without ever having to ask a question or break their flow.
The signals that trigger this kind of help are already readable within the shopping journey. Dwell time on specifications, repeated switching between variants, and jumping between similar product pages can suggest hesitation. Even scroll behaviour can suggest uncertainty. These are first-party behavioural cues only the retailer can see in real time. Offsite AI agents like ChatGPT or Gemini don’t have access to how a shopper is navigating within a site as it’s happening.
Importance of seamless intervention
The exact triggers will vary by category, but the principle stays the same. Provide assistance in line with shopper behaviour rather than interrupting the flow they’re in. Grocer Albertsons found that when help appears at the right moment, its Ask AI capability drove a 10% increase in basket size. This is notable in grocery, where shoppers are especially resistant to anything that slows a routine shop. The judgement behind when to intervene is more important than the interface, something retailers can refine through testing and learning over time.
A useful parallel can be found in Google’s AI Overviews. Rather than adding a chat button, Google embedded intelligence directly into existing search behaviour, providing context where it helped without changing how people search. Users still type queries, scan results, and click through as needed. That approach now supports billions of chats each month. Google just made an existing behaviour work better. AI in retail needs to take the same approach.
Designing AI around how people buy
The lesson from agentic commerce, and other AI-led shifts, is that AI creates the most value when it fits naturally into existing behaviour. Shoppers already know how to browse and compare when shopping online.
The real opportunity AI creates is making those moments easier and better informed. Ambient intelligence enables retailer sites to respond automatically at moments of uncertainty, such as when a shopper hesitates over specs or pauses before adding an item to their basket. Done well, this intelligence resolves doubt by drawing on first-party behavioural data and the outcomes of previous shopper choices, such as fit issues or product return patterns, that no external AI agent can see or infer on its own.
In a world where AI agents increasingly influence decisions, true agentic readiness means using the intelligence retailers already possess to make the shopping journey feel effortless.
