The Drinks Business Built an AI. It's Searching the Wrong Layer.

Quick Pour

Trade media building its own AI search tool is a sign the conversation has shifted from hype to implementation. But it's also a reminder that most of what we call "AI in drinks" still sits in the information layer, which is useful, and also is not where the real compounding happens.

The Drinks Business has launched a dedicated AI-powered search tool called justaskdb, built on almost 60,000 articles from thedrinksbusiness.com across the past 20 years. It's designed to answer drinks-trade questions by searching The Drinks Business' own journalism, returning relevant articles with summaries and links. (The Drinks Business) I don't want to be dismissive about it. A digital media company sitting on two decades of specialist archive and figuring out how to make that archive conversational is a genuinely interesting product move. If you need to remember who acquired who in Burgundy last autumn, or what the last six months of on-trade headlines looked like before walking into a meeting, this is useful. It's sharper than using site search, faster than skimming through saved articles, and it keeps the answer inside the world of drinks rather than pulling from the general internet noise.

The honest framing, though, is that it helps you talk about the industry, not run your business inside it. That's a meaningful distinction when you're trying to figure out where AI actually earns its place in a small producer's week.

justaskdb and what it's actually good for

If I think about the practical moments where something like justaskdb would sit open on my laptop, they're mostly research moments: sense-checking my own assumptions before a trade meeting, scanning category trends before writing something, building context before a call with a retailer who's about to ask questions I might not have current answers to. Those are real use cases and the tool would genuinely reduce the time tax of manually hunting through a sprawling archive. For someone who reads The Drinks Business regularly, having it searchable by question rather than keyword is a straightforward improvement on what existed before.

Where it gets more complicated is when people treat a tool like this as intelligence rather than information. Context is not strategy. Knowing that a category is trending in a particular direction doesn't tell you whether you should be in it, or what it would take to compete there given your specific margins, capacity, and timing. That analysis still has to happen somewhere else, with someone who knows your actual business. The article you just surfaced gives you the raw material for a conversation; it doesn't have the conversation for you.

The information layer vs the production layer

We think about "AI in drinks" as splitting into two very different categories, and the distinction matters because the compounding works differently in each. The information layer helps you find, summarise, and narrate what's happening in the industry: trade press, market signals, competitor moves, category data. It makes you a more informed observer. The production layer is something else entirely. It's the messy, unglamorous work of actually running the business: ordering, invoicing, production planning, managing channel mix, routing decisions between DTC and wholesale and bespoke. AI that touches the production layer doesn't make you more informed, it gives you time back. And in a small team, time is the whole game.

The examples we're building at Asterley Bros and through Absolution Labs sit firmly in the second category. An automated ordering portal for trade accounts so nobody has to manage an email thread to place a repeat order. Xero-integration invoicing that creates and sends without anyone touching it. A production planner tool on the shop floor that manages the liquid production flow between all our orders (balancing bespoke, bulk, DTC, samples, and B2B) without someone holding it all in their head. Or writing it on a whiteboard. None of that is glamorous. None of it would make a headline in The Drinks Business. But it's where the hours are hiding, and it compounds every single week.

The framing I keep coming back to is this: we're not trying to save money or reduce headcount. We're trying to automate the things that don't add value to the business or to our customers, then take those saved hours and reinvest them into the work that only humans can do well. Tastings, site visits, masterclasses, new product development, proper marketing. That's the real return. Not the cost saving, but what you do with the energy you bought back.

Why ROI is still muddy for most businesses

Part of why trade-media AI feels more prevalent than production AI is that it's genuinely easier to ship. It's a contained dataset, a defined user interaction, and it doesn't have to survive contact with your invoicing system at 10pm on a Sunday. The harder work is building AI that integrates deeply enough into real operations to change how the week actually feels. That difficulty shows up in the numbers. A March 2026 "State of Digital" report covering roughly 85 limited-service restaurant chains found only 9% saying AI had a meaningful impact so far. (Restaurant Business) Different sector, same structural problem: a lot of AI activity is still happening at the edges, and edge work rarely moves core metrics.

The gap between a tool being interesting and a tool changing throughput is almost always an integration problem. The tool sits outside the workflow instead of inside it, or it requires someone to remember to use it rather than being built into the process so it happens automatically. Information-layer tools are harder to integrate by design, because information consumption is already a discretionary behaviour. Production-layer tools that sit inside mandatory processes don't have that problem. They just run.

A practical comparison: trade-media AI search vs maker-focused automation

Here's how we think about it when we're deciding what to build next.

The key question for any tool is whether it changes what you do, not just what you know. That's the test.

The bigger opportunity for trade media

There's an interesting second-order point here worth sitting with for a moment. Trade media building AI tools changes what a publication fundamentally is. If you own the archive and you own the interface to that archive's knowledge, you're no longer just distributing articles. You're distributing answers. Over time that may become the more defensible product, particularly as generic search gets noisier and people want sources they already trust. It's a smart move by The Drinks Business and I'd expect others to follow. For makers, though, the bigger win is still elsewhere: in tools that remove friction, protect quality, and give you the time back to do the human work properly. Those are different problems and they need different tools to solve them.

Frequently asked questions

What is justaskdb?
justaskdb is an AI-powered search tool from The Drinks Business that answers drinks-trade questions by searching The Drinks Business' own journalism.

How many articles does justaskdb search?
The Drinks Business says justaskdb searches almost 60,000 articles from thedrinksbusiness.com, spanning about 20 years.

Is trade-media AI useful for small drinks producers?
Yes, for context and signals. The bigger leverage usually comes from AI applied to operational friction, planning, and decision quality.

What's the "production layer" in this context?
It's the work that actually runs the business: ordering, invoicing, production planning, customer service handoffs, and the decisions that allocate scarce time and cash.

Robert Berry is co-founder of Asterley Bros, a London-based premium aperitivo company, and Absolution Labs, an AI automation consultancy for drinks businesses. He makes vermouth by day and builds AI systems in the margins.