Written by Erin Smith
At the Drinks Business Fine Wine 2026 conference last week, a panel of leaders from across the sector agreed that AI will reshape the trade. The sharper question was what it should be pointed at. The answer that kept surfacing: the industry's cost base and its capital, not its expertise.
The fourth panel of the day at DB’s annual conference, held at the British Library's Knowledge Centre on Wednesday 23 September, "Fine Wine Meets AI: Opportunity, Insight and Innovation", set itself a simple brief. Where does AI genuinely add value, and where does human expertise remain essential?
Moderated by Patrick Schmitt MW, the panel brought together Fero CEO and co-founder Mitchel Fowler, Jeremy Howard, CEO of Cru World Wine, Nick Martin, CEO of Wine Owners, Jane Renwick, director of fine wine of London City Bond, freelance consultant Siobhan Turner MW, and Callum Woodcock, CEO of WineFi. It overran, as the moderator predicted it would. By the end, a clear consensus had formed: AI will not replace the people and communities around fine wine that make it special, but it may finally fix the economic drag that is holding it back from investing in building those communities.
The competition is fierce for fine wine
Fine wine's competition isn't the château next door, Fowler said. It's losing consumer relevance to completely different categories like cocktails in a can. And to compete, the trade needs to stop letting capital dictate how it operates. That argument, that AI's biggest prize in fine wine is fixing the economics rather than replacing the expertise, ran through the whole panel.
He argued that this is an enabler that will allow the sector to compete more effectively, against the rising tide of competition from other drinks categories. Wine more generally it was argued has less margins than spirits, which allows for that sector to invest three to five times as much in marketing and advertising. This is translating in an explosion in growth in Fero’s innovative spirits and RTD clients, have gone from zero to in some cases millions of units each month in the UK. That is the competition.
In order to compete there needs to be adoption in certain parts of fine wine businesses, such as supply chain operations, just to stay on the pitch. Fowler offered the clearest first-hand evidence of what that could look like. In nine months, Fero's engineering and product team has shrunk from 30 people to 13 while shipping three times as fast, twice as often, and with far fewer errors. This has not only allowed it to become more productive as a business, but Fero now incorporates the frontier models into its core workflow management tools to put its clients at the cutting edge. Roadmap items the business expected to take five to ten years have already shipped and things that were not possible before are augmenting our clients’ capabilities in completely new ways.
AI is a power tool, not a sommelier
The most useful AI in fine wine today is unglamorous. Martin described it as swapping a hand drill for a power tool: the same job, done faster and at greater scale. His examples were operational. They included standardising supplier invoices for import, integrating third-party apps and improving stock velocity and sales forecasting.
Renwick said London City Bond uses AI daily across its 16 sites, particularly to streamline deliveries. She drew a firm line, though. As a service provider, LCB must not reach past its customers into their relationships with private clients.
Woodcock, whose wine investment platform draws many clients from outside the traditional trade, uses AI agents to improve engagement. If a client skims a market report, that is one signal. If they read ten slides in, that is a far better reason for a salesperson to pick up the phone. AI decides who to call; a human still needs to make the call.
Howard added a practical warning. AI is not a silver bullet. Businesses must first write down, precisely, the logic of the process they want to automate, then iterate with the model. Only then does automation free up headcount or slow the need to hire.
Why the human expert isn't going anywhere
On the limits of AI, Turner MW was the panel's sharpest voice. Provenance remains a struggle, she said, partly because the trade likes to be opaque about where it bought a wine. Technology that could preserve confidentiality while proving provenance would be transformative. It does not yet exist.
She was equally sceptical of AI authentication tools trained on a narrow set of vintages from a handful of properties. They help little with a 1950s Burgundy, where a bottle can be authentic yet not be what its label claims. Her ultra-high-net-worth clients, many of whom made their fortunes in technology, could automate almost everything. They still want a Master of Wine on the phone, or in their cellar.
Fowler framed why that matters. Fine wine, he argued, is in a sense a social or neurological product, what happens in our neurons around those products is where we attribute value. Recognition, community and the stories told about a producer are what turn a small grower into a fine wine icon. That places the trade on the right side of AI disruption. The information and service economies are being upended first while experience-led categories built on community stand to gain.
The elephant in the room: capital
Howard put numbers to the trade's structural problem. By his figures, the UK's top 50 wine merchants together turn over around £750m a year, carry around £200m in admin costs and lose roughly £20m a year between them. They also hold around £220m of inventory and £250m of debtors on their balance sheets.
The result, he said, is an industry that is unprofitable, expensive to run and capital-heavy, with its capital tied up in the wrong places. Almost every company in the room was undercapitalised to some degree, he suggested, even though pools of capital exist that would gladly lend against fine wine as collateral. He called it the elephant in the room, and noted that finance had barely come up all day.
He sees AI working on all three fronts. It can cut the cost base. It can make inventory visible, so a buyer searching for a specific case can find a merchant who holds it. And it can support better capital allocation across a market he described as a series of walled gardens.
Fero's view: fix the deterministic problems first
Fowler's argument was that much of what holds wine back does not need a large language model at all. The trade is largely untouched by software, and many of its problems are deterministic. Supply chain optimisation and demand and supply management are solved problems in other consumer verticals. Wine has simply never applied the solutions, leaving the industry, in his words, "not match fit".
The cost of that shows up in reinvestment. Fowler contrasted mature spirits portfolios, which spend roughly 15–25% of revenue building and creating brands, with wine businesses that typically spend under 5%. He also noted that Diageo's marketing budget alone is equivalent to around two-thirds of the listed wine industry's revenue base. Without unlocking efficiency, wine will not have the earnings to invest in the communities it depends on.
Fero has narrowed its own focus to three capabilities: risk management, which is how it reliably raises capital for the industry; supply chain management; and trading management. The aim is to stop capital being the constraint the industry organises itself around, from payment terms to capital-intensive inventory, so the trade can concentrate on what it does best.
Fero is facing a dual challenge, fix the basics in these areas by bringing a new operating model to an industry that has historically been slow to adopt new technology, while simultaneously keeping its tools at the absolute cutting edge of what’s possible so that its client become (and remain) the iconic drinks businesses of tomorrow.
What changes in five years
Asked for one thing AI will fundamentally change, the panel's answers pointed in a consistent direction.
- Agents that buy. Howard expects collectors to have personal AI agents that find, negotiate and arrange delivery of wine to a brief. Merchants and warehouses will need to make inventory visible to machines, not just people.
- Machine-readable wine. Martin pointed to Google's Universal Commerce Protocol, which lets AI agents check out on a shopper's behalf. Merchants need to publish richer data, such as food pairings, in a form agents can read.
- Price discovery. Woodcock expects arbitrage between exchanges to narrow as pricing becomes more transparent.
- Provenance. Turner hopes, optimistically, for far more transparency on where bottles have been.
- The supply chain and its capital. Fowler predicted both will be completely transformed.
On what will not change, the room was unanimous: the human experience of discovering and sharing wine.
Analysis: be ruthless about the thing only you can do
The panel's most useful message was not really about AI at all. It was about focus.
Fine wine's value lives in expertise, trust and community: the Master of Wine who knows the client’s cellar inside out and can make recommendations she knows the client will love, the merchant who knows a collector's taste and the stories that make a producer matter. None of that is at risk. What is at risk is a trade that spends its margin on administration, ties up its capital in the wrong places and cannot afford to invest in the things customers actually come back for.
Fowler's advice to producers, distributors and retailers alike was to know with confidence what makes them unique and be completely open-minded about everything else. Customer expectations and switching behaviour will normalise around the pace of change AI brings. The businesses that come through it will be the ones that let technology and better-structured capital carry the load underneath, so their people can do the work only people can do.
