What happens when AI starts doing the weekly shop?

The weekly grocery shop used to start with a pen and a piece of paper. You’d scribble down what had run out during the week, head to the supermarket and work your way through the aisles. If you forgot something - bad luck - it was back onto the list for the next store visit.

Today, the weekly shop looks very different. We have apps that remember what we buy and help us plan our list, AI that can recommend what we should make and, increasingly, agents that can start doing parts of the shopping for us.

And I suspect grocery could be one of the first categories where consumers delegate parts - or even all - of the weekly shop to an agent to complete the shop on their behalf.

McKinsey’s research revealed that 51% of consumers would be open to AI-assisted product search, and 50% to personalised product recommendations, signalling a genuine level of confidence in AI to support discovery and deliver relevant suggestions that improve the shopping experience. Meanwhile, Logica research suggests 35% of Australians would use agentic shopping tools for grocery shopping.

But what is it about grocery that makes it such a suitable candidate for delegation? When you think about the weekly shop, it isn't particularly enjoyable. Milk, cereal, detergent, toilet paper, lunchbox staples - many of these purchases are repetitive, utility-driven and relatively low regret. We buy the same products on a relatively predictable cadence, over and over again.

That makes grocery particularly ripe for delegation.

And many grocers thenselves see it heading that way. Nearly one in two US grocers surveyed by McKinsey expect AI agents to assist with a third or more of all transactions within five years. When both shoppers and grocers are signalling a willingness to embrace agentic shopping, it starts to feel less like a distant vision of commerce and more like a very real shift in how the weekly shop could get done.

Imagine opening your supermarket app and saying: "Build my usual weekly shop, but we have friends coming for dinner on Saturday, I need school lunches for five days and keep the total under $180."

Rather than searching, browsing and adding 30 or 40 products individually, an AI agent could interpret the mission, look at what you normally buy, account for your budget, evaluate prices and promotions and build the basket. You review it, make a few changes and check out.

The weekly shop suddenly becomes less of a chore and more of an approval process.

But I don't think we'll wake up one morning and decide AI can take over the entire household grocery budget. McKinsey's research shows willingness drops as AI gains more autonomy. While around half of consumers are comfortable with AI helping them search or recommending products, that falls to 32% for an AI-built basket they review, 23% for automatic ordering within predefined rules and 20% for fully automatic ordering without review.

That suggests the journey towards grocery automation will happen gradually.

First: help me find it.

Then: build it for me.

Then: buy it for me, within my rules.

Eventually: manage it for me.

Perhaps my agent knows we go through two bottles of milk a week, the dog food arrives every four weeks and we're almost certainly going to need toilet paper before Sunday.

At that point, I'm not really asking AI to shop for me. I'm asking it to manage some of the mental load of keeping a household running. And that may ultimately be the more compelling proposition.

For retailers, removing friction from the weekly shop is an obvious opportunity. For brands, the implications are more complicated.

Today, if I search for Greek yoghurt on a retailer website, I might see 30+ products. Brands compete for my attention through search position, sponsored listings, promotions, ratings, imagery, packaging and familiarity.

But imagine I don’t see the shelf at all. Instead I ask my agent to "add a high-protein Greek yoghurt the kids will eat, under $8."

Now the agent is doing the evaluation. It might consider nutritional information, price, availability, ratings, previous purchases and promotions before deciding what goes into my basket. I may never see the other 29 products it considered.

The battle starts shifting from winning the shelf to making the shortlist.

The Appetise Grocery Gap research gives us an interesting glimpse into what can influence those decisions. Its analysis of grocery behaviour across Australia and New Zealand found that when shoppers override AI-generated product suggestions, three recurring factors emerge: price, format and fit. A shopper might swap a recommended brand for a cheaper option, replace a processed product with something fresher, or change the pack size or variant to better suit their household.

These are decisions we make instinctively as shoppers. An agent needs to learn them.

This is perhaps where it gets most interesting for brands.

Imagine I ask an agent to do my normal weekly shop but keep it under $180. It builds my usual basket and comes back at $202. It then discovers it can save $24 by swapping six of my regular branded products for cheaper alternatives.

What should it do?

I might be completely indifferent about canned tomatoes but fiercely loyal to my coffee. I'll happily swap detergent for a meaningful saving, but don't touch the yoghurt my kids actually eat. My loyalty isn't binary - some preferences are much more negotiable than others.

An agent therefore needs to understand not only what I prefer, but the strength of that preference and when it becomes negotiable.

And that potentially creates a very different competitive environment for brands. Being the product someone usually buys may not be enough if a machine is constantly evaluating whether another product offers better value, availability or fit.

Brands have spent decades learning how to influence people through packaging, shelf presence, brand building, promotions, retail media and, more recently, digital shelf optimisation. None of that disappears. But increasingly, there may be another audience they need to convince: the machine acting on the shopper's behalf.

Can an agent find your product and understand what makes it different? Does it know who it is suited to? Can it recognise when a shopper has demonstrated a strong preference for it? And can it justify your premium when it has been asked to optimise the overall basket?

Those questions make product data, ingredients, nutritional information, ratings, reviews, availability, pricing and promotions increasingly important, not just because they help shoppers make decisions, but because they give machines the signals they need to make decisions too.

How much attention we give to the grocery shop is at the heart of the shift we are seeing. There will always be products we care about choosing, brands we love and occasions where we want to browse and discover something new.

But there is also a huge part of the weekly shop that most of us would happily think about less. And the appeal of delegation could be greatest for busy, time-poor families with large baskets, where the weekly shop involves more products, more planning and more time. For these households, handing some of that work to an agent could be a relatively easy trade-off - less time spent shopping in return for greater ease and convenience.

We’ve already moved from pen and paper to apps and increasingly AI-assisted shopping. Agents take that evolution one step further - allowing us to hand over more of the work, and potentially more of the decisions, involved in the weekly shop.

For brands, that leaves an important question: when your customer starts delegating more of those decisions to AI, will the agent still choose you?

This week we welcomed 130+ brand leaders to the second annual Digital Shelf Future Forum, hosted by Arktic Fox in partnership with Six Degrees Executive.

The ambition behind the Forum is simple: create a space for brand leaders to step away from the day-to-day and openly explore how digital commerce is changing and what they need to be thinking about next.

Across the morning we tackled everything from AI and agentic commerce to marketplaces, quick commerce, product content, digital shelf capability and retail media. While there was plenty to take away, a few themes stood out.

🔎 Discovery is changing again. Roger Dunn challenged brands to not only think about SEO, AEO and GEO but to also consider ACO (Agentic Commerce Optimisation) as AI increasingly influences which products are discovered, recommended and ultimately chosen.

🛠️ Digital shelf needs to become a business capability, not a project. Vera Skocic from RYOBI shared the importance of diagnosing where the real gaps and opportunities lie on the digital shelf before jumping to solutions, and making digital shelf a cross-functional priority as key ingredients for success. From sales and marketing to product and supply chain, all teams play a role in managing the levers of digital shelf performance - from discoverability and online shelf presence to credibility and availability - making shared ownership critical. Her advice: measure the right leading indicators, build internal understanding and focus on creating an enduring capability.

⚡ Emerging channels are becoming mainstream. With Australian marketplaces now capturing $1 in every $4 spent online, and 66% of Australians buying through quick commerce platforms like Uber and DoorDash, more brands than ever are building strategies to win in these channels. Success requires brands to think differently about their portfolio, propositions and shopper missions by channel to win across increasingly diverse commerce environments.

🤖 AI makes understanding the human even more important. As technology increasingly sits between shoppers and brands, knowing who your customer is, what they value and why they choose you becomes even more critical.

The channels will keep evolving and the technology will keep changing. The brands best positioned to navigate it will be those that build an adaptive and curious culture, while building the capability and skills to respond as commerce evolves.

I’m really proud to create a space for the industry to have important conversations about digital shelf and the future of commerce. A huge thank you to everyone who joined us, our speakers and panellists for sharing their perspectives so openly, our partners Six Degrees Executive for teaming up with us to deliver the event and to our event sponsors Bazaarvoice, Bynder, Salsify and Uber Advertising for their support in making the forum possible.

Meta has launched Muse, a new personal AI agent currently rolling out in the US, designed to move beyond answering questions to actually completing tasks. Muse operates through its own secure browser and can navigate websites, fill out forms, book appointments and make purchases, with users approving critical actions before they happen. It can also connect to email, calendars, Instagram and other apps, and is accessible through its own app or WhatsApp. What’s particularly interesting is the move from AI helping us decide to AI acting on our behalf. Another sign that brands need to start thinking about how they show up when an agent, rather than the customer, is navigating the web and completing the transaction.

Australia’s proposed Digital Duty of Care bill, which dropped yesterday, could see Big Tech platforms face fines of up to $109 million for breaches. One of the most interesting proposals for marketers is “My Feed, My Way”, which would require social platforms to give users a clear choice between an algorithmically recommended feed or a following-only feed, showing content from people, brands and creators they’ve actively chosen to follow. The concept isn’t entirely new - Instagram already offers Following and Favourites feeds - but social media algorithms have become a powerful engine for organic discovery, enabling content to reach people who don’t already follow an account. If more users opt out, that discovery engine becomes weaker, potentially making organic reach harder and followers and genuinely engaged communities more valuable again. It could also increase brands’ reliance on paid distribution to reach new audiences. The irony is that the draft doesn’t restrict algorithmic optimisation of paid advertising, potentially making paid media even more important for brands looking to reach new audiences.

SKIMS has partnered with DoorDash to offer on-demand delivery from its entire US retail footprint, with more than 23,000 styles across shapewear, underwear, loungewear and clothing available for delivery, on average, in under an hour. What stands out isn't simply SKIMS joining DoorDash, it's the continued expansion of what constitutes a quick-commerce mission. Forgot your pyjamas while travelling? Need shapewear before a night out? Last-minute plans and nothing to wear? These are urgency-led missions traditionally associated with groceries, takeaway or convenience, now being applied to fashion. It's another signal that quick commerce is evolving from “I need dinner now” towards “I need almost anything now.”

What started as a way to order dinner has evolved into one of the fastest growing channels in Australian retail.

Consumers are now buying everything from groceries and beauty products to hardware and home essentials, expecting delivery in minutes rather than days. As shopper expectations shift, retailers are expanding into new categories, new partnerships are emerging, and brands are rethinking how they engage consumers in a convenience-first world.

In this episode of Unpacking the Digital Shelf – APAC Edition, I’m joined by Matt McGinley, Head of CPG and Retail Ads at Uber Advertising ANZ, to unpack what's driving the rapid growth of quick commerce and what it means for brands.

We explore how shopper behaviour has changed over the past 12 to 18 months, why retailers are investing in convenience and rapid delivery, what the best-performing brands are doing differently, how to approach quick commerce as part of a broader commerce strategy, and the metrics that matter, from incrementality to new-to-brand customers.

I’m an advisor, trainer and thought leader with 25+ years’ experience across digital, marketing, loyalty and commerce, working with consumer, FMCG and retail brands. I’m also the host of Unpacking the Digital Shelf – APAC Edition and have been recognised as a RETHINK Retail Top Retail Expert (2026), and Top 20 CMO (2018).

Before you go…

🧭 Need strategic clarity or expert advice? Let’s chat.

🛠️ Looking to build capability in your team? Explore training & workshops.

🎤 Planning an event or offsite? Book a keynote or thought leadership session.

Get in touch at [email protected] or find out more here, and don’t forget to follow along on LinkedIn and Substack.