"Couldn't I Just Ask AI?"

The Question That Kills Most Solo Businesses in 2026

Key Takeaways:

1. A Washington University and NYU study found the best freelancers were hit harder by AI than the mediocre ones. That is the opposite of what almost everyone assumes, and it changes what being excellent is actually worth.

2. If your price rests on being better than the average person at producing the work, you are standing exactly where AI now stands. That position is no longer defensible, whatever your track record says.

3. Clients never bought the deliverable you think they bought. They bought something else, and once you see what it was, the fee conversation stops being about your rate.

4. There is a one-minute test that shows whether your offer still holds up. All you need is your own service description and any AI assistant. Most people do not like the result.

Prefer watching a video? https://youtu.be/9CPCFI6iCgM

---

It usually arrives in a friendly tone, which somehow makes it worse.

"Thanks for this. Can you just clarify your fee for me, because we can probably ask ChatGPT to do this too."

If you have been independent for a while, some version of that sentence has already landed in your inbox. The instinct is to defend the work: the years behind it, the nuance the machine missed, the three things it got subtly wrong. All of that may be true. None of it wins the argument, because you are now arguing about quality of output, and output is exactly the thing that just got cheap.

There’s something that should worry you even more.

Researchers at Washington University and NYU tracked a large online labour market after ChatGPT launched. Writing freelancers lost around 2% of their monthly jobs and 5.2% of their earnings. For image work, it was worse: 3.7% fewer jobs and 9.4% less income. Modest numbers on their own. The finding underneath them is the one that matters.

The most experienced freelancers were hit hardest. Higher hourly rates, top-rated badges, long track records: those people lost more work, not less. Weaker performers got a productivity boost from the tool. Stronger performers simply lost their edge.

Read that again, because it breaks the comfortable story most of us tell ourselves. "AI will replace the average, and the excellent will be fine." That is not what the data showed. If your premium was built on being better at producing the thing, AI did not protect you. It took aim at you specifically, because your entire price was the gap between your output and an ordinary person's output, and that gap is where the model lives now.

So the question is not how to compete on output. It is what you were really selling all along.

Clients never wanted the document

I have interviewed a lot of solopreneurs for soloSei, and the ones growing right now have all worked out the same thing, usually the hard way.

No client ever wanted a strategy document. They wanted to stop arguing internally about the strategy. Nobody wanted a training needs analysis. They wanted to stop wasting budget on courses that changed nothing.

We sold documents because documents are easy to scope, easy to invoice and easy to hand over. The document was never the value. It was the receipt.

That was a workable shortcut for twenty years. It is not workable now, because your client can generate that receipt in ninety seconds. What they cannot generate is the confidence that the receipt is right.

Three moves that actually reposition you

Sell the decision, not the deliverable. AI produces options with total confidence and no accountability, which is a dangerous combination for anyone who cannot tell a good option from a plausible one. Your value is not "here are five logo design options", because that is now free. It is "this one, for these reasons, and here is how I would implement it." Judgement is the product, so price it as the product. In practice, your proposal stops listing artefacts and starts naming the decision the client will be able to make with confidence once you are done.

Put your name on the result, not the effort. The strongest defence against commoditisation is to be measured. I recently experienced this first hand, as a customer. I marketing agency gave me a proposal to help me with lead generation on LinkedIn. All competitors told me how many leads they will get for my budget. This agency committed to converted sales, and their contract stated that they will “stay involved until we get there". They were 60% more expensive than the nearest competitor, but they removed the risk. Very few people will take that bet, which is precisely why it prices well. Be careful and be honest here: only promise a number you can influence, and be explicit about what the client has to do for it to land. An outcome you cannot control is a lawsuit waiting to happen, not a positioning strategy.

Be in the room. There is a category of work AI cannot touch, not because it lacks intelligence but because it lacks presence and consequence. It can be: Sitting with a nervous founder about to make an expensive mistake; Reading a leadership meeting and knowing which objection is real and which is performance; Being the person who gets phoned on a Sunday, with a reputation on the line alongside the client's. That is not the domain of artificial intelligence. It is the domain of you, and it has real value.

The test to run on your own offer

Take your current service description, the one on your website or a deliverable in your last proposal. Paste the whole thing into any AI assistant and ask it to deliver what you promised.

If it produces something 80% as good in under a minute, you are selling output, and you have a pricing problem that no amount of better copywriting will fix.

If it produces something that looks right but you can immediately see where it would fail in a real client's context, that gap is your business. Write that gap down in plain language. That sentence, not your CV, is your positioning.

None of this is about being AI-proof, which is the wrong ambition anyway. The solopreneurs pulling ahead are using these tools harder than anyone, producing more in a week than a small team used to manage in a month. They just stopped selling the part the machine does.

Sources

• Xiang Hui and Oren Reshef (Washington University in St. Louis Olin Business School) and Luofeng Zhou (NYU), "The Short-Term Effects of Generative Artificial Intelligence on Employment: Evidence from an Online Labor Market", summarised at olin.washu.edu and INFORMS

• Framing inspiration: "How to Build a Profitable 1-Person Business With AI" (the outcomes-not-outputs argument and the experience, judgement, credibility, audience and involvement list)

• soloSei solopreneur interview series, youtube.com/@soloSei