The industry is consumed by appetite suppressants. But the far more radical threat to how people buy, cook and eat is already being built and it doesn’t require a prescription, access to disposable income or a trip to a private clinic.
Spend ten minutes in any food industry conference room right now and the conversation will turn to GLP-1s. Which category gets hit hardest? How to reformulate. What happens to confectionery volumes? Whether the company’s new high-protein SKU can capture some of that appetite-suppressed-but-still-eating occasion. Consultants are billing. Decks are being built. Everyone is wearing their best concerned face.
The concern isn’t entirely misplaced. According to UCL research published in January 2026, an estimated 1.6 million adults in England, Wales and Scotland used drugs such as Wegovy and Mounjaro for weight loss in the past year alone, with a further 3.3 million saying they would consider doing so in the next twelve months. That is a potential market of nearly five million people actively changing how they eat.
As a GLP1 user myself, a GP that has no idea I am on it and a network of people who have also quietly admitted it once I outed myself a few weeks ago….I’m telling you it’s way more than that and its growing faster as the results speak for themselves.
But here is what strikes me as odd: while the industry fixates on a drug that suppresses appetite in a significant but still-constrained minority, a technology is quietly arriving that has the potential to suppress the very mechanism that most food marketing depends on and its freely available to everybody.
That technology is agentic AI. And compared to GLP1s, it barely gets a whisper…
What “agentic” actually means
For the uninitiated: agentic AI isn’t ChatGPT. It’s not a chatbot you have a conversation with. It’s an autonomous system that acts on your behalf – browsing, comparing, deciding, purchasing – without you needing to be involved at every step. The terminator but (hopefully!) less world-ending and more admin angel.
The analogy that lands best is this: if the large language model is the brain, agentic AI gives it hands and eyes. It doesn’t wait to be asked. It executes based on your intent.
The capabilities are already here, and the UK grocery market is directly in scope. IGD (note it’s not a fringe technology commentator, but the Institute of Grocery Distribution) published a formal report in January 2026 warning that agentic AI can already automate basket building and replenishment, compare prices and availability across multiple retailers, optimise purchases for budget, health and sustainability goals, and complete checkout without any shopper intervention. The report’s conclusion was blunt: businesses that fail to prepare “may find the future has already been decided without them.”
£23.4bn – UK online grocery market value in 2024 — the market agentic AI is targeting first
£500m – weekly revenue at risk if UK retailers fail to adapt to AI-driven shopping, per Algolia
47% – surge in UK consumers using AI to shop in 2025, per Adyen’s 31-country retail index
70% – of 25–34-year-olds would switch supermarket for a third-party AI shopping app, per Algolia
And the direction of travel is clear. Currently, only 3% of UK shoppers use AI tools specifically for grocery shopping, a figure that sounds reassuring until you remember that online grocery itself sat at similarly marginal levels before convenience became the accelerant. IGD is explicit that “the uncertainty surrounds when agentic AI will change behaviours, not if.”
The System 1 problem nobody is talking about
For decades, the food and drink industry has been built on a single psychological insight: most purchasing decisions are made fast, emotionally, and without conscious deliberation. This is what Daniel Kahneman called System 1, the brain’s autopilot. The reach-for-what-you-know. The pack-colour recognition. The end-of-aisle impulsive pick-up. The Friday evening takeaway order driven by craving rather than calculation.
Enormous sums flow from this insight. Neuroscience research suggests System 1 thinking accounts for up to 95% of our daily cognitive activity. Brands have spent decades trying to make consumer loyalty instinctive because that is where most food purchasing lives: in habit, familiarity and emotional resonance, not rational analysis. Shelf placement, packaging design, colour cues, promotional mechanics, advertising all engineered to hit the gut before the head gets involved….half of my behavioural science post-grad was focused on this concept!
So, here is where it gets difficult if you have marketing budget and sales targets…
If an AI agent builds your basket, it doesn’t feel anything when it sees a Walkers display at the end of the aisle. It doesn’t get seduced by a meal deal. It just optimises.
Now consider what happens when purchasing decisions are increasingly delegated to an algorithm. The agent doesn’t have a favourite brand from childhood. It has no limbic response to a crinkle-cut crisp. It won’t grab a Twix because it’s been a long afternoon. As IGD’s Toby Pickard put it: “Impulse, emotion, and human discovery are redundant when it is machine visibility, not shelf visibility, that shapes choice.”
That is a fairly devastating sentence for anyone who has spent a career building brand salience through above-the-line investment.
The inversion of the decision ratio
Here is the scenario worth stress-testing: what happens if the current ratio of intuitive-to-rational food decisions shifts materially? Not entirely, not overnight but structurally, in exactly the categories most exposed to automation.
UK grocery is the perfect first domain for this. Unlike fashion or beauty, grocery doesn’t run on aspiration. It runs on rhythm, habit and replenishment (and ok a bit of aspiration if you are a Waitrose shopper!) which makes it the ideal candidate for AI delegation. Algolia’s 2025 survey of 2,000 British consumers found that nearly half already use AI to help decide what household items to buy. Half of those surveyed had also abandoned an online shop due to frustrations finding products or managing substitutions (Amen to that!). This is precisely the friction that agentic tools resolve instantly. Sign. Me. Up.
What agentic AI can already do in UK grocery
IGD’s January 2026 analysis confirms agentic AI can currently: automate basket building and replenishment based on past behaviour; compare prices and availability across Tesco, Sainsbury’s, Ocado and others simultaneously; optimise the weekly shop against budget, health, and sustainability parameters; and complete checkout without the shopper ever visiting a retailer’s website. The only meaningful barrier remaining is consumer trust and among 25–34-year-olds, that barrier is already low.
The food categories most dependent on System 1; confectionery, crisps, soft drinks, impulse snacking, are precisely the ones most vulnerable to an AI optimising on nutritional density, price per unit, and whatever health goals the shopper set up on a Tuesday morning. The agent isn’t reaching for the Haribo. It’s cross-referencing macro targets against whatever Ocado has in stock.
How brands will need to compete with the algorithm
A world where the consumer’s AI builds the basket means your route to purchase is no longer through a standout fixture on aisle five or an emotionally resonant TV spot. It’s through structured product data, nutritional metadata, and being the option the algorithm surfaces when optimising against someone’s stated parameters.
This is a genuinely new kind of brand challenge. As Bain & Company noted in a recent analysis of agentic commerce, full agent-to-agent purchasing — where the buyer’s and retailer’s AI systems transact directly — could eventually bypass brand websites and category fixtures altogether. The warning for suppliers is stark: brands that fail to make their product data legible to AI agents risk becoming invisible not because consumers rejected them, but because the algorithm never surfaced them in the first place.
That is a very different capability from making an emotionally resonant advertisement. And it demands a very different kind of marketing investment… are industry marketers even ready for this shift?
The GLP-1 parallel
There is actually something these two disruptions share: both attack the automaticity of food choice, just by different routes. GLP-1 drugs suppress the neurological signals that make highly palatable food compelling. Agentic AI bypasses the decision-making process entirely. One quiets the appetite. The other removes the human from the equation.
The difference is scale and reach. GLP-1 adoption in the UK, however fast it grows, is constrained by NHS eligibility, private prescription costs averaging hundreds of pounds a month, side effects, and the fact that most users cycle off within a year. Agentic AI has none of those barriers. The models are getting better, faster and cheaper. The friction of delegating a weekly Tesco shop to an AI is already lower than most people think — and it is falling every month.
The food industry is right to take GLP-1 seriously. But if it is spending its strategic bandwidth almost entirely on reformulating for a medicated minority, while failing to prepare for a world in which algorithms shop on behalf of the majority, it may be facing the wrong disruption in the wrong direction.
The bot doing your shopping doesn’t care which brand your grandmother bought. It doesn’t feel anything when it scrolls past a beautifully shot pack shot. It cares about price, availability, nutritional parameters and what you told it you were trying to achieve this week.
That is a more profound challenge than appetite suppression. It’s just harder to put in a board presentation….perhaps it needs more research…because where there is great change….there is usually great opportunity for the strategic advantage. Just sayin…