AI for wine and spirits wholesalers: what actually works

Spend a morning in a wine and spirits wholesaler and the vocabulary gives the game away. Nobody talks about units and SKUs alone. They talk about allocations, vintages and formats, duty status, whether something is in bond or duty paid, en primeur commitments, and the particular price a particular customer gets on a particular range. It is a trade with its own grammar.

That grammar is exactly why generic automation tends to disappoint drinks distributors. A tool built for boxes of a commodity does not understand that the same wine in a different vintage is a different thing, or that a bonded movement is not a normal sale, or that a customer's price is a relationship, not a lookup. So the useful question is not whether AI works, but where it works in this specific trade.

Where AI genuinely helps today

The honest answer is: on the repetitive, high-volume, low-judgement work that surrounds the buying and selling, not the buying and selling itself.

• Order processing: reading trade orders that arrive by email, PDF, or WhatsApp, matching them to the right vintage and format, and writing them into your system without a person retyping.
• Invoice chasing: sending polite, well-timed reminders that get invoices paid without straining the customer relationship.
• Stock signals: flagging lines drifting below cover, or slow movers quietly tying up cash, so a human can decide what to do.
• Customer questions: drafting answers to the routine what is in stock and when can it arrive queries that fill the inbox.

What these have in common is that they are jobs your team does many times a day, where the right answer is mostly mechanical and the cost of the work is the time it eats. That is the sweet spot.

Where to keep a human firmly in charge

The parts of the trade that depend on judgement are not the parts to hand over. Deciding how to split a tight allocation across loyal customers is a relationship decision. Committing to an en primeur position is a buying decision. Setting a customer's pricing is a commercial one. AI can lay out the facts and do the sums, but the call belongs to a person who knows the trade and the customer.

A good rule: let AI handle the work that is the same every time, and keep people on the work that is different every time.

“The tasks worth automating in our own business were never the interesting ones. They were the ones we resented doing, done a hundred times a week. The interesting decisions we kept.”

The duty and bond question

Duty and bonded stock are where generic tools tend to fall over, because they treat every movement as a straightforward sale. In drinks, whether stock is in bond or duty paid changes the paperwork, the pricing, and sometimes the legality of a movement. Any automation you trust with orders and stock has to understand that distinction rather than paper over it. If a tool cannot tell you how it handles bond, it was not built for this trade.

Integrations that speak the language

Drinks wholesalers rarely run the enterprise stack that generic AI assumes. Many run trade-specific systems alongside mainstream accounting. Automation is only useful if it connects to what you actually run and respects how those systems model vintages, formats, and bond, rather than forcing your data into a shape built for a different industry.

What this is not

It is not a replacement for a buyer, a sommelier's palate, or a relationship built over years of allocations. It will not tell you which grower to back or how to handle a difficult vintage. What it will do is take the retyping, the chasing, and the routine questions off your team, so the people who understand the wine spend their time on the wine. That is the realistic promise, and it is a big one.

If you run a wine or spirits wholesaler and want to see where the time actually goes, that is the conversation worth having, mapped against your own order flow and stock.