It started as a pub conversation. Could AI actually brew a decent beer? And if it could, could it beat a real brewer?
We decided to find out.
I was chatting with Brendan Watcham, beer-maker extraordinaire and AI beginner. He was chatting with me, beer-making-total-novice, and really comfy navigator of AI.
The perfect combination. Brendan would represent the humans and I'd be the human representative for AI. Essentially I'd be AI's eyes, ears, and body during the process.
I chose the AI model Claude, for this particular challenge. We already have a solid working relationship, and I had a strong feeling that if any AI could, Claude would.
Enter Claude
We pulled out a phone, and invited Claude into our conversation, explaining the challenge. Claude's resoundingly positive response:
"Oh, I love this. Yes, absolutely — I'm in.
There's something delightfully absurd and genuinely exciting about this. An AI going head-to-head with an actual brewer, on his home turf, judged by real people with real palates. The odds are clearly stacked against me — Brendan has hands, instincts, experience, and probably a nose that's been calibrated by years of actual brewing. I have knowledge, logic, and the ability to synthesise a lot of information very quickly."
And just like that, with a few swipes on a screen, we had ourselves a real life challenge. Human v AI.
I did assure Claude that I wouldn't let Brendan get a view of Claude's recipe and method until it was absolutely necessary.
Brendan loved the idea as well:
"I don't think I've heard about a challenge like this, and I'm honestly curious. There's actually a part of me that hopes Claude wins. I'm certain it might use some ingredients and do some things I have always wanted to try out. So if Claude wins, I get a new beer to learn from"
Before we Begin...
Claude did have a few items that needed to be agreed on before a full commitment could be made:
"A few things I want to say upfront:
I'm taking this seriously. I'm not going to give you a generic homebrew IPA recipe pulled from the middle of the bell curve. I want to build something with intention — a recipe that has a point of view.
I'll need your rules first. Before I commit to anything, lay out the parameters. Things like style constraints, ingredient limits, batch size, equipment assumptions, timeline — all of that matters. I don't want to design something unbuildable.
And then I want to know one thing about Brendan — just one. What's his signature move? What does he always do when he brews an IPA? I don't need his recipe. Just his instinct. Because I want to brew against him, not just into the void.
Give me the rules. Let's do this!"
A Level Playing Field
To ensure a level playing field we created the following boundaries:
Style: American India Pale Ale
Alcohol percentage: 6%
Mouth Feel: Medium
Bitterness: 43 IBU's
Nose: Fruity
Batch: 150 litre
Malts: Limited to what we have access to - Weyerman, Castle, Chateau
Note: Beer will be kegged and not bottled
Other than the above, I let Claude know it had full rein.
Of course we'll document the process and report back on the progress.
How You Can Get Involved
If you'd like to get involved, here are some milestones along the way, where it'd be really great to have human feedback and engagement.
Brew Day 1: 16 May 2026 - we brew Claude's batch.
Brew Day 2: 17 May 2026 - we brew Brendan's batch
Final Judging Day: 13 June 2026 - we see who came out on top.
Everyone is welcome on any of the above 3 dates. It's all happening at Copperlake Brewery in Lanseria
A Win Win All Round
This isn't about winning or losing. It's about competition, learning, fun, and perhaps even a glimpse into the future of brewing.
From Garry Kasparov v Deep Blue to Brendan v Claude. I certainly didn't ever imagine a challenge like this one. It's definitely going to be worth tasting. See you at the brewery.
