We Asked AI to Run a Business Here’s What We Would Never Let It Decide

AI can write your emails.

Analyse your customers.

Build workflows.

Answer enquiries at 3:17 in the morning without demanding overtime.

It can help write code, generate concepts, analyse thousands of rows of data and summarise a meeting you mentally left twenty minutes before it ended.

So naturally, we had a question.

How much of a business should we actually let AI run?

Quite a lot, as it turns out.

But definitely not everything.

Because while artificial intelligence may be extraordinarily capable, there are still certain decisions we would prefer not to hand over to something that has never had an angry customer standing three feet away from it.

So, if AI arrived at Nayéh tomorrow morning, put its imaginary briefcase on the table and announced:

“I’ll take it from here.”

Here is what we would happily hand over.

And what we absolutely would not.

Give AI the Boring Stuff. Immediately.

There are tasks human beings continue doing manually for reasons nobody can adequately explain.

Copying information between systems.

Renaming files.

Sending the same confirmation email 600 times.

Creating reports from information already sitting perfectly happily inside a database.

Moving customer details from a form into a CRM.

Reminding somebody for the fourth time that an invoice is overdue.

This is exactly where AI and automation become interesting.

Not because robots are coming for everybody’s jobs.

Because Sandra from Accounts has better things to do than copy email addresses into spreadsheets.

The best use of AI isn’t always spectacular.

Sometimes it simply removes hundreds of tiny, repetitive tasks that quietly consume thousands of hours every year.

That is not futuristic.

That is just sensible.

Would We Let AI Answer Customers?

Yes.

And absolutely not.

Imagine you run a hotel.

A guest messages:

“What time is breakfast?”

AI: perfect.

Breakfast is served from 7:00 to 10:30.

Nobody needed to interrupt a human being for that exchange.

Now imagine:

“This was supposed to be our anniversary. We’ve arrived and our reservation doesn’t exist.”

AI should probably stop feeling ambitious.

Because this isn’t an information problem anymore.

It is a human problem.

Someone is disappointed.

Someone needs authority to make a decision.

Someone needs to understand tone, context and the fact that complimentary champagne may currently be a better business strategy than explaining the booking policy.

This is where businesses get AI spectacularly wrong.

They automate conversations instead of automating friction.

There is a difference.

Good AI should make customer service faster.

It should not make your customers feel like they are arguing with a microwave.

Would We Let AI Set Our Prices?

We would certainly let it have an opinion.

AI can analyse competitors, demand, historical sales, margins, customer behaviour, seasonality and enormous quantities of information faster than any human team reasonably could.

Useful.

But imagine AI notices something.

Your 40€ product sells extremely well at 55€.

Excellent.

Then 65€.

Still selling.

Then 79€.

Computer says fabulous.

Margins say fabulous.

Customers?

They may quietly begin hating you.

Numbers can tell you what is happening.

They don’t always understand what you are doing to your brand while it happens.

Pricing isn’t simply mathematics.

A 2,000€ handbag is not priced according to the amount of leather required to prevent someone’s lipstick falling on the pavement.

Perception matters.

Positioning matters.

Trust matters.

Human irrationality matters enormously.

We would let AI analyse the chessboard.

We would still want a human deciding which piece to move.

Would We Let AI Hire Someone?

Imagine this.

Two candidates.

Candidate A perfectly matches every requirement.

Correct experience.

Correct keywords.

Beautiful career progression.

Excellent assessment results.

Candidate B has taken a strange career path, changed industries twice and doesn’t perfectly match the job description.

AI ranks Candidate A first.

Except Candidate B walks into the room and within twenty minutes you know:

That’s the person.

Businesses are built by unusual people surprisingly often.

People with weird combinations of experience.

People who don’t fit templates.

People whose greatest strength doesn’t appear neatly under “Skills” on a CV.

AI can help screen 4,000 applications.

Please let it.

Nobody wants to read 4,000 CVs.

But deciding who belongs inside the culture of a company?

We still want humans in that room.

Would We Let AI Create a Brand?

This one is particularly interesting.

Ask AI for:

“A luxury skincare brand for environmentally conscious women aged 25–45.”

Within seconds:

Beige.

Serif font.

Something involving leaves.

Possibly the word “Élan.”

Congratulations.

You now own approximately 14,000 existing skincare brands.

AI is exceptionally good at understanding patterns.

Which creates an interesting problem.

Great brands often become great by breaking them.

Nobody remembers the company that looked exactly correct for its category.

We remember the one that made us look twice.

AI can accelerate research.

Explore creative territories.

Analyse competitors.

Generate variations.

Test ideas.

Help enormously throughout a creative process.

But creativity isn’t simply generating something attractive.

It is deciding what deserves to exist.

That still requires taste.

And taste is annoyingly difficult to automate.

Would We Let AI Make a Million-Euro Decision?

Imagine AI analyses everything.

The market.

Competitors.

Financial projections.

Customer demand.

Economic conditions.

It concludes:

Do not expand.

The numbers are clear.

Except you know something it doesn’t.

You’ve spent twelve years in this industry.

You’ve watched customer behaviour change before the data reflected it.

Three competitors are quietly struggling.

A location has unexpectedly become available.

You can feel an opportunity forming.

Do you listen to the machine?

Maybe.

Maybe not.

This is where AI becomes dangerous if businesses misunderstand what intelligence actually means.

Data is incredibly powerful.

Experience is also data.

It is simply stored inside people.

A brilliant executive has spent years accumulating patterns from decisions, failures, negotiations, customers, employees and markets.

Sometimes they know something is wrong before the spreadsheet knows.

The future probably isn’t:

Human vs AI.

It is:

Human instinct + machine intelligence.

That combination is considerably more interesting.

The Question Isn’t “What Can AI Do?”

That question is already becoming outdated.

AI can do an extraordinary amount.

The better question is:

What should it do?

We would happily let AI:

Analyse enormous datasets.

Automate repetitive administration.

Identify patterns humans might miss.

Personalise customer journeys.

Improve forecasting.

Assist customer support.

Optimise workflows.

Help teams work faster.

Research.

Draft.

Organise.

Monitor.

Recommend.

And perform approximately seventeen thousand boring tasks nobody will mourn.

But we would be considerably more cautious about handing it:

Judgement.

Culture.

Taste.

Empathy.

Leadership.

Moral responsibility.

Major strategic decisions.

Or the keys to the office.

Not yet, anyway.

The Best AI Strategy Might Be Surprisingly Human

Businesses rushing into AI often start in the wrong place.

They ask:

“Where can we put AI?”

We prefer another question:

“Where are humans wasting their time?”

That is where things become interesting.

If your best salesperson spends three hours every day doing administration, automate the administration.

If your customer service team answers the same twelve questions 400 times a week, automate those questions.

If managers spend Friday afternoons manually producing reports, automate the reports.

Then give the humans something more valuable.

Time.

Time to sell.

Time to create.

Time to solve unusual problems.

Time to speak to customers who actually need another person.

Time to think.

Perhaps the most exciting thing about artificial intelligence isn’t that it might replace people.

It is that, used intelligently, it might finally stop talented people spending half their lives doing things a computer should have been doing anyway.

At Nayéh, that is the kind of AI future we’re interested in.

Not businesses run entirely by machines.

Businesses where machines handle more of the machinery, so humans can get back to building something remarkable.