Around 3,000 small distillers are now operating in the UK and Europe, most of them running on teams of six or fewer people. The ones who are figuring out AI fastest are not the ones who want to replace the work. They are the ones who want to protect the time they have to do it.
That distinction matters more than anything else in how you think about automation in craft production.
The argument against AI in craft usually misses the target
The concern is understandable: if a machine can assess botanical combinations, manage cask maturation, and generate a production schedule, what is left for the maker? The answer is quite a lot, and the more interesting question is whether the machine is even being pointed at the right problems in the first place.
Bespoken Spirits in California is the clearest named case study in how this plays out. Head of flavour Jordan Spitzer uses machine learning to compress the whiskey maturation cycle from years to weeks, iterating through flavour profiles at a pace no traditional barrel programme could match. But Spitzer does not describe himself as a passenger in that process. In a Dropbox feature from October 2025, the team described AI as "a creative partner" that sparks ideas when creativity doesn't arrive on its own. The final tasting decision, the judgment about whether this batch is the one that ships, remains a human call.
Mackmyra Intelligens, produced with Microsoft and Fourkind, went further: the world's first AI-generated whisky, with master blender Angela D'Orazio fine-tuning the result. The AI generated the recipe. The master blender made the final call. Both parts of that sentence matter.
What automation actually competes with at a small producer
At Asterley Bros, our team of six manages around 200 active trade accounts. That means trade emails arriving at odd hours, buyer follow-ups that need a specific response within 48 hours, a CRM that falls behind the moment anyone is travelling for a tasting, and a calendar full of sample requests that each need a different answer depending on the account's history.
This is not craft work. There is no flavour decision in chasing a missing invoice. There is no production judgment in logging that a bar manager at a Soho restaurant asked about ESTATE for their Negroni list and someone needs to follow up next Tuesday.
The Asterley blog post on AI in distilleries put it plainly: AI complements human expertise, offering precision and insights while leaving final decisions in the hands of skilled distillers. That is true at the production-floor level. It is just as true at the operational level, where the decisions that go to the human should be the ones that actually require human judgment.
The craft decision versus the admin decision
Here is the practical split we have arrived at, working through this in our own operation:
| Human decision (irreplaceable) | Machine task (automation candidate) |
|---|---|
| Maceration duration for a specific botanical batch | Tracking and logging maceration schedule across batches |
| Final sensory sign-off on a production run | Scheduling the QC check and routing the result |
| Deciding which trade account to prioritise this week | Surfacing which accounts have had no contact in 30 days |
| Writing a personalised response to a buyer | Flagging the email, providing account history, suggesting a response window |
| Formulating a new seasonal serve | Cross-referencing ingredient availability with existing recipes |
The left column is craft. The right column is infrastructure. The question for any small producer is not "should we use AI?" but "which column is the machine currently competing with in our operation?"
Why small producers have an advantage here
There is a counterintuitive truth in the current AI moment that I want to be careful about, because it is easy to overstate. Small producers are not disadvantaged by scale when it comes to AI deployment. A six-person operation building automation for its own specific problem (200 trade accounts, three vermouth SKUs, a weekly dispatch cycle) can get to a working solution faster than an enterprise that needs procurement approval and IT integration at every step.
The reason is that the specificity IS the advantage. You know exactly what the problem is. You feel it every Tuesday when the CRM is three weeks out of date and a buyer in Edinburgh is waiting for a call. An AI system built precisely for that problem, trained on the patterns of your actual trade relationships, is a more useful tool than a generic CRM plugin that costs twelve times as much and solves a slightly different version of the issue.
Affordable entry points exist now. The Asterley blog cited tools accessible from £15 per month for inventory tracking, batch recording, and compliance reporting. The barrier to a meaningful first deployment is operational clarity, not budget.
The part nobody writes about
What changes when you automate the data-capture layer of a small drinks business is not the work itself. It is the quality of attention you can bring to the work that remains.
When a trade email doesn't need to be manually logged because the system already did it, and the buyer's history is already surfaced by the time you sit down to respond, you are not just faster. You are more present. The meeting feels different when you have not spent the morning chasing an order that was already dispatched and nobody updated the record.
Rob Berry put it this way in a recent interview: "It's a huge release for us to be able to do the things that we enjoy so much, which is meeting people and discussing and tasting and making great drinks and just being as human as possible." That is the argument for automation in craft production. Not that machines do the craft. That they protect the space in which it happens.
Frequently asked questions
Does AI threaten craft in small drinks production?
Not when deployed for the right tasks. The risk is in automating the sensory and relational work that defines craft. The opportunity is in automating the data-capture, comms, and administrative work that crowds out craft time. Well-targeted AI leaves the human in charge of the things only humans can do.
What tasks should a small distillery automate first?
The highest-return early targets are account-intelligence capture (trade emails, buyer follow-ups, CRM updates), demand-signal aggregation (reading Shopify, email, and calendar data into a weekly digest), and order-to-fulfilment routing. These tasks repeat, require precision, and don't benefit from the human judgment that makes craft production distinctive.
Can AI improve recipe development in craft spirits?
AI can surface candidate botanical pairings faster than manual research and flag flavour interactions across large compound libraries. Mackmyra Intelligens showed that AI can generate a viable recipe, with a master blender providing the final sensory fine-tuning. The production decision remains human; the search space is compressed by the machine.
What does it actually cost a small drinks business not to automate?
The cost is in founder-hours: chasing missing deliveries, submitting HMRC compliance reports, updating a CRM nobody has time to maintain, and responding to trade enquiries that arrive late in the evening. These tasks are necessary but they don't generate revenue. Every hour spent on reactive admin is an hour not spent on product, relationships, or new business.
How does Absolution Labs approach automation for drinks producers?
We build for the specific operational problem rather than implement generic platforms. This is where Absolution Labs comes in. We bring the tooling and the build experience, but the starting point is always the problem the maker is actually trying to solve. The build-logs documenting what we've built and what we've learned are at absolutionlabs.com/blog.