For one full year, someone handed everything over to AI — morning schedules, dinner plans, even how to spend the weekend. Joanna Stern, senior technology analyst at NBC News, turned that year into a book. Her verdict after the experiment was brief. 

"Amazing and uncomfortable." 

Somewhere between those two words is where Korea's solo entrepreneurs stand today.

AI tools are seeping into our work at remarkable speed. But the more they seep in, the blurrier the question of how much to hand over has become. As the tools improve, the important question is no longer whether to use them, but where and how. Stern's experiment brings that blurry question back into focus.

365 Days: Everything She Delegated

The scope of what Stern handed to AI didn't stop at work. It included drafting emails, coordinating schedules, planning travel, and booking restaurants — but also helping with her kids' homework and curating her hobbies. Using AI as a productivity tool and using it as the operator of your life are two different things. That difference is captured, intact, in her year of records.

The gains were real. Time spent on repetitive tasks dropped, and decisions came faster. The hours she once poured into email drafts shrank dramatically, and travel logistics went the same way. It was the early efficiency bump anyone who adopts a new tool experiences. With the time saved, she could think more deeply or take on more projects.

But the discomfort accumulated alongside the gains. The schedules AI built were efficient, Stern recalls, yet they felt stripped of her own rhythm. The moment AI picked the restaurant on a family trip, the small anticipation she used to get from planning it herself disappeared. Only after the tool took over that role did she realize that the weight of the process — coordinating, comparing, choosing — had been as much a part of the trip as the outcome. That was the substance of what she came to call the "trade-offs."

What makes the experiment interesting is that Stern is no stranger to technology. She has spent roughly two decades writing one of the most widely read tech columns in America. Being first to try every new device and service is literally her job — and after a year of use, her conclusion was "amazing and uncomfortable." This is not someone who doesn't understand what the tools can do. It's the complexity felt by someone who came to know them deeply.

The More You Delegate, the More You Review

Hand more to AI, and you end up making more decisions, not fewer. It's counterintuitive. Delegation is supposed to reduce decisions. So why do they multiply?

It's structural. AI gives you options. Applying context to those options and weighting them differently is still your job. At first, "just handle it" seems to work. But as AI's output piles up, so does the number of times you have to ask whether it's right. The total volume of decisions doesn't shrink — the place where decisions happen moves. From execution to review.

Take an example. Hand a client email to AI and you get a draft. Can you send it as is? Most people don't. They edit. And while editing, they're judging: Is this phrasing my voice? Does this content fit this client? Is this proposal right for this moment? The decision didn't disappear; its nature changed. A decision that starts from a blank screen and one that starts from a draft are different things. Both are still decisions.

This shift isn't a bad direction in itself. Editing on top of a draft often beats starting from a blank screen. But build your operation on the assumption that "AI will take care of it," and the decisions that pass through unreviewed will, at some point, clump together into a problem. The pattern recurs throughout Stern's records: efficiently processed work piles up, and somewhere inside it, a direction she never wanted quietly takes hold.

The Case Against Overreliance on AI

Stern's experiment has its critics. The argument is that the deeper AI tools are woven into daily life, the weaker your judgment becomes. The instincts you build by making choices yourself, the logic goes, dull a little with every delegation. Among management researchers, there is steady concern that AI assistance can boost short-term performance while eroding long-term problem-solving capacity.

There are more concrete worries, too. The options AI suggests come from its training data, and that data skews toward the average. Average destinations, average restaurants, average sentences. High on efficiency, low on distinctiveness. The more deeply a solo entrepreneur or founder leans on AI, the more their output may start to look like their competitors'. Social media content flattening into a uniform AI prose style, proposals settling into identical structures, brand language fading — these patterns are already visible across many fields.

Stern herself recorded moments when she "couldn't tell whether a choice was mine or the AI's." This is an especially sensitive problem for solo entrepreneurs, because what they're selling is, in most cases, their own judgment and perspective. The tension between efficiency and distinctiveness goes beyond personal taste — it cuts straight to what makes the business different.

Why the Delegation Criteria Have to Come First

Stern's story comes out of American media, but its structure maps directly onto solo entrepreneurs in Korea. AI writes the client emails, builds the content calendar, generates the social media copy. Each delegation is reasonable on its own. But what kind of person all of it adds up to — that's the question to ask before delegating, not after.

The standards you set when you first adopt a new tool or system end up shaping how you operate from then on. The idea that your first 90 days create the patterns for years to come doesn't apply only to office jobs. AI tools work the same way. Unless you deliberately decide, at the outset, what scope and what standards you'll work with, the tool becomes a habit and the habit becomes your operating model. Once a delegation boundary settles into place, pulling it back in is hard.

In practice, the first question to ask is repetition. Is this a task where the same format recurs? Work with a fixed structure — email reply templates, quote drafts, meeting summaries — is what AI handles well. Tasks where the context changes every time and judgment is required are a different category. Drawing that line first is what keeps the delegation boundary from quietly expanding.

The next criterion is reversibility. If you shipped AI's output without review, could you take it back? Work that's hard to undo — contracts, a first proposal to a client, publicly published content — must go through a review step, no exceptions. A moment will come when you're tempted to skip it: the deadline is tight, or the output looks good enough. Setting the reversibility rule in advance is exactly for that moment.

Last is the brand surface. Will customers judge you by this output? The more a piece of work serves as their basis for judging you, the more the final call has to be human, even if AI drafts it. The content a first-time customer sees, the opening paragraph of a proposal, every touchpoint that carries your brand language — none of it can be filled with average options.

Joanna Stern didn't stop using AI when her year-long experiment ended. She redrew the boundaries. Choosing, each time, where to delegate and where to stay — that, she says, is how you hold the balance between efficiency and distinctiveness.

For a solo entrepreneur, AI is useful. But usefulness doesn't come from the tool's performance alone. It comes when you can decide for yourself where to delegate and where to make the call. Will enough decisions still be yours to claim a year from now? That standard is something you can set today — before the tool sets it for you.