The Karpathy Loop, Applied to Marketing
Google will optimize your ads for you. Inside their platform, using their algorithm, measured by their definitions. You don't see how it works and you don't get to change what "better" means. You get a dashboard that says trust us.
Meta does the same thing with a different name. Advantage+ picks the creative and the audience. Both call it automation.
I used to call it that too, until I noticed what it actually resembles: a casino automating your betting strategy. The house built the algorithm. The house defines what winning means. The house optimizes for the house.
You're not the customer in that arrangement. You're the supply.
In March 2026, Andrej Karpathy released a repo called autoresearch.
630 lines of code, one GPU, one file the agent can edit, one file only the human writes, one metric that decides what lives and what dies.
The agent forms a hypothesis, edits the training code, runs a five-minute experiment, checks the score, keeps the change or rolls it back, then loops again. About a hundred experiments overnight, no human in the loop for any single one of them.
Tens of thousands of people starred it in its first week. Most of them saw a machine learning tool.
I saw the opposite of a black box, which was an uncomfortable thing to notice, because I'd been living inside one for years without questioning it. Just reading the dashboards it handed me.
Autoresearch works the other way. The human defines the constraints. The human defines success.
The human sees every experiment, every result, every decision the system made and why. The agent does the work. The human owns the logic.
Nothing in it is hidden, and nothing is quietly optimized for someone else's revenue. The whole system fits inside your own head, which is not how your ad platform, your attribution tool, or any vendor in your stack works.
Every tool you rent is someone else's opinion about your business, repackaged as a fact.
The dashboard says you're winning. The P&L says something else. You can't explain the gap, because the logic that produced it lives in someone else's box.
So I built a tool called LoopKit and mapped it onto Karpathy's setup almost one to one. Autoresearch has a file the agent can't touch, which sets the data and the rules. In LoopKit, that's the brand brief: product, audience, voice, the limits the agent works inside, and they're mine, not Google's.
Autoresearch has one file the agent rewrites every round. In LoopKit, that's the ad copy, with angles, hooks and structure all fair game.
Autoresearch has one number that decides everything. In LoopKit, that's a score I define myself from relevance, clarity, urgency and brand fit. You can see how it's built, instead of trusting a number that means whatever the vendor needs it to mean that quarter.
Same setup, different domain. The difference is who owns the loop.
"AI writes ad copy" is a commodity now. Every tool does that.
What almost nobody does is run fifty experiments on that copy, let an agent guess and fail and learn against a fixed standard across dozens of rounds. One is a question. The other is a study.
The pattern carries over to anything with a small set of options and a score you can trust: email subject lines, landing page headlines, bid strategies.
And you stop renting someone else's judgment about your business and start building your own, one checked experiment at a time.
You can look at the same market two ways. In one, you log into a platform, read a dashboard someone else built, make decisions filtered through their priorities, and scramble every time the algorithm changes underneath you.
I know that morning routine. Open the laptop, check the dashboard, scan the green numbers, close the laptop, learn nothing.
In the other, you run your own loops, define your own metric, own every experiment, and already have the data to check an algorithm change the day it happens.
The first makes you a customer. The second makes you a competitor.
I don't think the interesting question left is what AI can do. I think it's what loop you should be running overnight, on your own terms, with your own data, measured by a definition you actually wrote yourself. Not inside their black box. Inside yours.