What does AI actually replace in influencer marketing?

Picture a normal campaign. A brand wants four creators for a product launch in six weeks.

That's a lot of work: a plan, a shortlist, prices, contracts, payments, shipping, approvals and a mountain of comments at the end. So how much of it could a machine do?

More than you might think. And that's very good news, because the half a machine can do is the half almost nobody enjoys.

The simple rule

Three economists, David Autor, Frank Levy and Richard Murnane, wrote the rule down in the Quarterly Journal of Economics in November 2003. Computers take over tasks that follow clear rules. And they create more work in the tasks that don't.

Here's the surprising part: it isn't about hard versus easy. Chess is hard, and a machine plays it brilliantly. Knowing which of two photos will make someone smile is easy for you, and no machine can do it.

So the real question for any step is this: could someone write down the right answer before the work starts? If yes, a machine can take it. If no, it needs a person.

A brand deal, step by step

Let's walk through that campaign and sort every step.

• Writing the first plan: AI does it. A good plan has a known shape: the product, the buyer, what to show, roughly what to say. A machine can draft that in seconds.
• Deciding if the plan is any good: you do it. No rule can tell you an idea is boring. You can tell in a second.
• Finding creators who match: AI does it. Checking audience data against a brief is rule-following, and it used to eat a junior person's fortnight.
• Saying yes to the shortlist: you do it. The machine ranks. The final yes is taste, and taste is what clients pay for.
• Chasing approvals, tracking every parcel and watching every post: AI does all of it. Each step has a right answer you could write down on day one.
• Making the video: the creator does it. A machine can generate a video. It can't make hundreds of thousands of people choose to watch one particular person every week.
• Reading every comment afterwards: AI does it. Nobody reads eight thousand comments by hand. Now someone finally can.
• Deciding what to do about those comments: you do it. If hundreds of people asked where to buy, deciding that the next video opens with the link is a person's call.

Look at the pattern. The machine takes the paperwork. You keep the ideas, the taste and the relationships.

Faster, and better too

This isn't just about speed. Shakked Noy and Whitney Zhang tested it with 453 professionals doing real tasks from their own jobs, and published the results in Science in 2023. With AI help, people finished about 40% faster, and the quality of their work was rated about 18% higher.

The best part is how the work changed. In their words, the tool "restructures tasks towards idea-generation and editing and away from rough-drafting".

In plain words: you stop staring at a blank page and start improving a draft. Less typing, more thinking.

People and machines are better together

Here's one of my favourite stories about this.

For more than ten years, scientists could not work out the shape of one enzyme from a monkey virus related to HIV. Every automated method failed. So in 2011, a team at the University of Washington around David Baker turned the problem into a puzzle in Foldit, a game where ordinary people fold proteins on screen for fun.

The players solved it in three weeks.

The paper in Nature Structural and Molecular Biology lists the scientists as authors, and also two teams of gamers. The computers did the maths. The people did the seeing.

The story has a lovely ending. On 9 October 2024, the Nobel Prize in Chemistry went half to Demis Hassabis and John Jumper for an AI that predicts the shapes of proteins, and half to David Baker for designing brand new proteins. According to the Nobel announcement, more than two million people in 190 countries have used the prediction tool, on problems like antibiotic resistance and enzymes that break down plastic.

The machine took an enormous amount of drudgery off people's hands. And people didn't do less. They reached further than ever.

Try this on Monday

1. Write out every step of one of your deals, from the first message to the final payment.
2. Mark each step "rules" or "judgement". Be strict. If someone could have written the right answer on a form before the work began, it's "rules", however long it takes you today.
3. Hand the "rules" list to your tools and watch what happens to your week.

What's left is the part you're really paid for, and probably the part you wanted to do in the first place.

Where Lumienzo fits

Lumi runs every rule-following step on its own. It learns your brand and product and researches what's trending. Then it writes the plan, finds the creators, tracks every sample and watches every comment. It all just happens, and you review it. The ideas and the final yes stay with the people whose names are on the work.

See how that works for creators, brands and agencies, or read why AI is making creators more valuable.