GarthBox

Last month I turned a robot loose on the bars of Wisconsin to ask a single question: how much are you charging for Spotted Cow?

There was no business case for this. I run a brew bar. I visit bars. I already know what Spotted Cow costs. I built a conversational AI that could dial a bar, sit through the ringing, chat with whoever picked up, and log the number, purely because the idea made me laugh at my desk. (I saw someone do it for Guinness in Ireland and got inspired.)

That is the whole method of differentiating advantages in work, life, love, etc.,. But let me back up.

When I started with AI, I made one mental move that changed everything after it. I decided to treat it the way I treat Crimson Desert. I run multiple characters. I take the snackable missions. I grind the repetitive stuff without resenting it, because grinding is part of the game. I told myself I was going to enjoy every element of the journey, including the boring ones, and I stopped watching the meter. Nobody calculates the cost-per-hour of a game they love. (130 hours into Crimson Desert because my PS5 tells me, but I’ve got no clue how many I’ve put into Claude…far more, that’s for sure.)

Meanwhile, a lot of the smartest people I know were treating AI like a forklift. Efficiency. Career advantage. Prompts as productivity. They got exactly what they optimized for: slightly faster versions of the errands they already ran.

The side quests are where you find the sword.

While I was building a voice line to answer FAQ calls for the bar, the good idea arrived without an appointment: what if the calls went the other direction? A robot calls you, offers one cool beer fact to brighten your day, and wants nothing. No booking link. No survey. No strings.

Every call a business makes to you wants something back. A call that only gives might be the strangest call you get all year. It’s not quite the same as receiving a check in the mail, but it’s close! You would tell somebody about it. I cannot buy that with an ad budget, and no efficiency roadmap on earth would have produced it. Roadmaps improve the errands you already run.

I don’t believe anyone can treat this technology like a video game for six months and come out empty-handed. Play generates its own proof.

And this game does something no other game I own does. It patches itself while you sleep. Every few months (weeks? days? hours?!) the engine upgrades, and you can walk back through your old save files and watch them level up. Work I finished last winter runs better today than the day I shipped it. I’m not sure much else in my life works that way. The bar floor does not re-mop itself better each quarter.

The strange accounting of all this is that I went looking for fun and came back with an advantage. Every time I went looking for the advantage directly, I came back with a subscription and a headache…frustration..rage quitting.

So, one cool beer fact, no strings attached: New Glarus refuses to sell Spotted Cow outside Wisconsin, and bars across the border have been caught smuggling kegs of it. The beer’s best marketing is the state line.

Stay Positive & The Next Round Is On Me

We don’t serve Spotted Cow at Garth’s Brew Bar. Never have. It’s a fine beer and half of Wisconsin grew up on it, but our whole thing is the beer you haven’t met yet, so New Glarus’s flagship has never touched our lines.

Which is exactly why, across a few Saturday nights, I had a robot call 96 bars across the state to find out what everyone else charges for it.

I’d read about a guy in Ireland who used an AI voice to call three thousand pubs and track the price of Guinness. It went viral, and pubs actually started lowering their prices to look good on his board. I wanted the Wisconsin version. So over a weekend and about eight dollars, I wired up an AI agent that could dial a bar, introduce itself honestly (“quick heads up, I’m an AI, and I’m not selling anything”), ask what a pint of Spotted Cow runs, and then ask the question I actually cared about: if you had to send a Spotted Cow drinker to one other beer, what would you pour them?

Then I pointed it at New Glarus, Madison, Milwaukee, the Dells, Minocqua, and Green Bay, and let it go for the pilot.

Here’s what I expected: a tidy map, cheapest pint to priciest, across the whole state.

Here’s what I got: 96 calls, and 5 prices.

It turns out people do not want to tell a robot the price of a beer. Most of the calls hit voicemail, a supper-club phone tree (“dial one for the front desk”), or a bartender who picked up, heard a synthetic voice, and hung up before the second sentence. On a Saturday night, with the place slammed, I honestly don’t blame them. The clean statewide price index I set out to build does not exist, because the premise was wrong. You cannot cold-call your way to that number. People will give it to a person, not to a machine.

But the handful of bars that did talk handed me two things I wasn’t looking for.

The first was the Dells. Four of my five prices came from Wisconsin Dells supper clubs, and they were the priciest in the state. The Del-Bar quoted eight dollars a pint. Fur, Fin & Feather, five. So the tourist corridor is both the only place that reliably answers the phone and the place that charges the most for the same beer. That’s not an index. That’s a tourist tax, and now I’ve got the receipts.

The second one is the one I’m keeping. What they said to drink instead. Kleeman’s in New Glarus said Totally Naked. A Madison tavern said Moon Man. And The Del-Bar, a white-tablecloth supper club charging eight bucks a pint, told my robot to pour a Spotted Cow drinker a Miller High Life. I’ve been doing this a long time, and that might be the most honest beer recommendation I have ever heard.

A few things I’m taking from the whole exercise, in case you’re thinking about turning an AI loose on an idea of your own.

The cheapest experiment is the one reality can kill in a weekend. I was out eight dollars and a few Saturdays, not a month and a real budget. If the hypothesis is wrong, you want to find that out for the price of a sandwich.

Also, your hypothesis is almost never the story. I set out to rank prices. The leftovers, the tourist tax and the “try this instead” map, turned out to be far more interesting than the thing I was chasing. Build the experiment loose enough that its byproducts can surprise you.

And people talk to people. AI made it absurdly cheap to ask a dumb question 96 times in one night. The answer it kept getting back was that trust doesn’t scale the way phone calls do. Worth sitting with before you point a robot at your own customers.

Alas. I’m scrapping the price index and building the thing the data actually wanted to be: a map of what Wisconsin bartenders tell you to drink when you’ve had your fill of Spotted Cow. Which, it turns out, is the beer education I’ve been trying to do for years. I just needed a robot and a bad idea to find it.

Stay Positive & Grab A High Life, I Guess

There’s a track where I sometimes run my bike to empty on purpose. I can afford to, because there’s a fuel tank sitting right there and the worst case is a short walk and a little embarrassment. The first time I did it, the low fuel light came on and I tensed up, waiting for the engine to starve. I still had a few laps left in my session. It kept running. Lap after lap, well past the point where the gauge swore I was finished.

That’s the first thing empty teaches you. E is not where you think it is. Every single time I’ve run something down to what I believed was the bottom, there was more. The gauge in your car was calibrated by people managing liability, not measuring fuel. The light comes on early because nobody wants to be the engineer who stranded you on a highway. It’s a warning built for the manufacturer’s lawyers.

I’ve tested this outside the bike, too. We ran a neighborhood appreciation night at the bar, free beer until the kegs blew, and I kept waiting for the moment we’d have nothing left to pour. It arrived hours after any of us predicted.

I once spent sixteen hours in a single day on a touchpoint plan I was pitching to a client, certain at hour ten that I was done thinking. The hours after that held the best ideas in the deck.

The pattern never breaks. The reserve is always bigger than the reading.

Which is what makes this a business problem and not a car problem.

Companies pay serious money to never feel any of this. That’s what most efficiency metrics are. Utilization rates, capacity dashboards, burn reports.

They are warning lights, calibrated conservatively, installed by people whose job is making sure nothing ever sputters. Useful, sure.

But a company that always obeys the light never learns where its actual bottom is. Its understanding of its own capacity is a legal document. The teams that win are the ones that occasionally ignore the light and chase impact anyway, because they’re the only ones operating on real data about what they can actually do.

Now the other half, the part I got wrong. Early in my career I lived at E. Not visits. Residence. It was horrendous for my health and worse for my relationships, and the damage didn’t come from running dry. It came from how I refilled. A splash here, a quarter tank there, just enough to keep moving. I tear up when I think about what I missed out on because of it.

When you never fill all the way back up, you lose track of two numbers at once: where your empty is, and how big your tank was to begin with. I spent years driving a body whose gauge I could no longer read at all.

So the discipline has two halves, and the second is the one I failed…

Lesson is…Run to E rarely, on purpose, someplace with a tank waiting nearby. And when you refill, go to full. All the way to the click. The people who burn out are usually not the ones who ran to empty. They’re the ones who never came all the way back.

Stay Positive & Next Fill Station 44 Miles…Can You Make It?

“Hey, I’ll give you ten minutes back.”

I’ve said it. You’ve heard it. It gets delivered in the tone of a man handing over a wrapped present, and for a second the room actually warms up, because ending early is the closest thing corporate life has to snow day.

But look at the grammar. Give it back. To give something back, it had to be yours. The hour was never mine. I borrowed it, from you, from everyone on the invite, and now I’m returning a fraction of the loan and waiting for applause. No bank teller has ever handed you part of your own deposit and called it a gift.

And even calling it a return is generous, because of what the minutes come back as. Ten minutes in the middle of a stacked afternoon is change in pennies. Technically money. Try buying anything with it.

You know this because you’ve watched what happens next. The meeting ends twelve minutes early. Nobody stands up. The little squares hang there for a beat, then go black one by one, and every single person slides the same six inches to the left, into Slack. The inbox eats the gift whole. Nobody stretches. Nobody steps outside. Nobody walks the dog. The dog, who heard the blip sound of a meeting ending and had opinions about it, goes back to sleep.

I spent years behind a bar, and here’s what the bar taught me about gifts: a comped beer costs the house something. That’s the whole reason it means something. Generosity has a receipt. When I end your meeting early, I spend nothing. My calendar didn’t get lighter, yours didn’t get freer, and the phrase cost me four words. Any gift that costs the giver nothing is not a gift. It’s positioning.

Companies run this same play on customers all day. “We’re giving back to you” usually means points, and points usually mean handcuffs with a bow on them. Customers can smell generosity that doesn’t cost anything, the same way your team can. The smell is identical.

There is a version of this I’d respect. It’s heavier. End the meeting early and then ask, out loud, “What will you do with your twelve minutes?” Make everyone say it. Watch how fast the room gets uncomfortable, because the honest answer is Slack, and now everyone has to hear themselves say Slack.

I haven’t done it yet. I’m not sure I want to be asked, either. Customers probably don’t either. Probably time to rewrite our value statements.

Stay Positive & Let’s Walk The Dog Next Time, Shall We?

I built a robot whose only job is to reject my writing.

Not edit it. Reject it. It reads my drafts and answers one question: could anyone have written this? It knows my tics. It keeps a list of phrases I’m no longer allowed to use. When a draft fails, it doesn’t cushion the news. It says no, and it says why.

The test at the center of it is simple. Strip the byline off any piece of content and ask: if you swapped in another name, would anybody blink? If the answer is no, it isn’t writing. It’s inventory.

You can run that test on anything now. One prompt. Every founder letter, every brand manifesto, every post in your feed: remove the name, ask if any other name would fit. Most of it fails instantly. That should bother more people than it does.

What surprised me was what the robot caught in mine. Not weak sentences. Structure. I had a skeleton I’d been walking for years without seeing it: odd image up top, dramatic one-liner, three quick business examples, punchline in italics. The rhythm I thought was most mine turned out to be the thing that made posts smell generated. The machine found the machine in me.

Which is the actual point?

Everyone is building AI skills right now, and most of them are automation. Onboard the new keg, chase the invoice, close the month. Good. Those buy back hours, and hours matter. But the skills I love are the other kind. The ones that don’t do your thinking so much as raise the price of it.

A skill that asks what you believe that your smartest peers would argue with. A skill that fails your draft if it contains zero evidence from your actual life. A skill that forces you to think out-of-the-box.

I spent years behind a bar, and you cannot hand a customer a palate. You can pour a flight that reveals the one they already have. Taste has never once arrived on command. It arrives when the scenario corners it.

You can’t manufacture taste. I think we all get this. It’s what makes us feel like we’re still needed.

But…

But you can manufacture the scenario that drags your taste out of you, blinking, and puts it to work.

That’s what a good editor always was. That’s what a deadline was. That’s what the friend who reads your draft and says “you’re better than this” was.

The scenario was always the machine. Now the machine can be the scenario.

The same tool that writes ten thousand interchangeable posts for one person becomes the strictest editor another person has ever had. Same tool. Same price. The difference is whether you wanted to be pushed in the first place.

Stay Positive & What You Build Is Up To You

I was on a call with my new team this week and someone joked that the reason our work gets taken seriously is that we don’t take ourselves seriously. Everyone laughed. Nobody disagreed. It was the truest thing said on the call, and it was delivered as a throwaway.

That joke has been rattling around in my head ever since, because it explains something companies keep saying without knowing what they mean. ClickUp says it. Half the job descriptions I read say it. We hire for how people think. It sounds like a nice sentiment about potential over pedigree. It’s actually a confession: the playbook is not the asset. It never was.

Consider Dick Fosbury. In the 1960s, every high jumper on earth cleared the bar the same way, rolling over it face-down in a technique called the straddle. It was the playbook. Coaches taught it, judges expected it, and the record books were full of it. Fosbury, a mediocre straddler, started throwing himself over the bar backwards. His own coaches tried to talk him out of it. Sportswriters compared him to a fish flopping in a boat, and they did not mean it kindly. Then he won gold in Mexico City in 1968, and within a decade the laughingstock technique was the only technique. Today nobody jumps any other way. The heresy became the playbook, which means the playbook you’re studying right now is just the last heresy that happened to win.

Beer did the same thing. When hazy IPAs first came out of Vermont, competition judges marked them down as flawed. Murky, unfiltered, wrong. The style guidelines literally had no box for them. Brewers who followed the guidelines made clear, correct, forgettable beer. Brewers who trusted their own palate over the rubric made the style that ended up carrying the entire category. I’ve poured a lot of both from behind a bar, and I can tell you which one people drive across town for.

So when I say how you think matters more than your resume, your age, or your passport stamps, I don’t mean it as encouragement. I mean it as a warning, specifically to the person who believes a proven framework will carry them. It won’t, and the reason is structural. By the time something becomes a playbook, its edge is gone. Everyone has it. A playbook is a record of how one person thought about one problem at one moment, and the moment is over. Running it faithfully doesn’t make you strategic. It makes you the fourth-best straddler in a world that’s about to jump backwards.

Which brings me back to the joke on that call. Not taking yourself seriously is not a personality quirk. It’s an operating condition. The team that can laugh at itself is the team where someone can say the fish-flopping idea out loud without calculating the reputational cost first. Seriousness is expensive. It makes every meeting a performance, and nobody proposes a backwards jump in the middle of a performance.

The uncomfortable part is that this never ends… Fosbury’s flop is now the orthodoxy, taught by the same kind of coaches who once mocked it. Hazy IPAs have style guidelines now. Whatever you figure out will calcify too, and someday a kid who doesn’t respect your playbook will clear a bar you didn’t think was clearable, backwards, while everyone winces.

Stay Positive & Say The Fish-Flopping Idea Out Loud

In Thornton Wilder’s The Matchmaker, a rich grump named Horace Vandergelder hires Dolly Levi to find him a wife. Dolly, who wants the job of wife for herself, does not compete for it. Competing is for amateurs. Instead she invents Ernestina Simple: a heiress of great beauty, impeccable manners, and zero existence.

Ernestina’s job is not to be chosen. Ernestina’s job is to stand next to Dolly and lose.

Economists eventually caught up to this trick and named it the decoy effect. It’s why the menu carries a lobster nobody orders. The lobster isn’t food. The lobster is a spotlight operator, and it points its beam at the steak.

Here’s the uncomfortable part. Nothing in your life gets chosen on its own merits. Not products, not job candidates, not you. Everything gets chosen against a lineup, and whoever builds the lineup wins before the judging starts. Dolly understood that the question is never “am I good?” The question is “what am I standing next to?” Or, as we more regularly say in our day-to-day… “Compared to what?”

Marketers who get this stop polishing the product and start composing the shelf. Good, better, best. The oversized option that makes the middle one feel wise. The premium tier that exists so the standard tier can look like a steal wearing a tuxedo.

And people who get this stop letting strangers cast their comparisons. Scroll long enough and you’ll find yourself standing in a lineup you never agreed to, somewhere between a guy with a boat and a woman whose kitchen has never known crumbs. That lineup was composed too. Just not by you, and not in your favor.

You can’t opt out of comparison. Nobody can; it’s how choosing works. But you can notice who arranged the shelf. You can ask what the lobster is selling. You can decline to be somebody else’s Ernestina.

…

Dolly gets her man, by the way. The fictional heiress retires undefeated, having lost exactly as planned.

Stay Positive & Shelves Are Ready For (Your?) Stocking