Context

AI is a very good writing tool. It improves rhythm, cleanliness and structure. The issue is that it is often used too early: in place of decisions, not after them.

And when that happens, a very convincing illusion appears: “Now it sounds better, so it must be right.” But “sounds better” only means it is more readable. It does not mean the brand has direction.

The simple rule

AI is an amplifier, not a compass. If you give it a clear direction, it makes it more effective. If you give it a confused message, it gives it back to you confused — just better written.

The illusion of clarity

Clarity is not only a matter of form. It is above all a matter of choice: what to say, to whom, with which priority and with which proof.

That is why AI can improve the text without really improving the brand. The copy becomes more fluid, but the promise stays broad, generic and interchangeable.

A text can flow well and still remain weak. Not because it is badly written, but because not enough has been chosen beforehand.

The symptoms

If you are using AI first — that is, to replace decisions — this usually happens:

Typical signals

  • the bio sounds good, but it could belong to anyone
  • the services are a list, not a choice
  • every week the angle changes: “maybe we are also…”
  • you have more content, but less clarity
  • the tone improves, but the focus remains negotiable

These are fairly clear signals: form grows, direction does not.

When AI lets you down (even if it writes well)

AI becomes risky not when it is wrong, but when it reassures you too early. It gives you texts that feel credible on the surface and makes you believe the work of direction has already been done.

It usually happens when:

  • it gives you phrases that could belong to anyone
  • it helps you say “everything better”, instead of helping you say less, more clearly
  • it optimises tone while the message priority is still undecided
  • it gives you the illusion that direction is there, simply because the text feels smooth

A practical clue

If you keep ending up with phrases like “tailored”, “360° approach”, “growth”, “value” but still cannot say for whom, in which situation and with which proof, you do not need a better prompt. You need to choose.

The 10-second test

Remove the name and the logo from the text. Then ask yourself: could it sit on a competitor’s website?

If the answer is yes, writing is not the bottleneck. Choice is.

When a text works “for anyone”, it is usually not guiding anyone.

The fix before any prompt

Before asking AI to “write this better”, close three decisions yourself. These are the ones that turn AI from filler into a genuinely useful tool.

1) For whom

You do not need to be hyper-niche. You need to be selective. If you speak to everyone, AI only helps you speak to everyone… better.

Useful question: who is the right person who, reading this, thinks: “Okay, this is about me”?

2) What problem do you really solve

“Growth”, “value”, “success”, “strong brand” are umbrella words. Direction begins when the problem becomes recognisable.

Useful question: what situation is the client in before they get to you?

3) What proof makes it credible

This is where many brands collapse: they define themselves well, but they do not demonstrate. And by proof I do not only mean testimonials: I mean method, standards, constraints, processes, criteria, examples.

Useful question: what can I see, read or verify that makes the promise credible?

For whom

A specific person, not “anyone who needs it”.

Problem

A recognisable tension, not an umbrella word.

Proof

A verifiable element, not only a well-written promise.

Right order

First the decisions, then the prompt.

How to use AI properly (after)

Once you already have those three decisions, AI becomes genuinely useful. Not to decide what to say, but to say it better without losing logic.

Where it really helps

  • turning a promise into a simpler sentence
  • creating coherent headline variations
  • adapting the same message across different channels without losing identity
  • reducing redundancy and improving rhythm

A simple example

Before

Prompt: “Write me a professional, high-impact bio.”

Result: often smooth… and interchangeable.

After

Prompt: “This is my direction: for whom…, problem…, proof… Now rewrite it in 3 versions while keeping these constraints.”

Result: cleaner text without losing identity.

Useful base prompt

Act as a brand strategist and editor. Do not invent anything. Use only the information below.

For whom: …

Problem: …

Proof: …

Write 3 variants of the bio (max 35 words) + 1 website headline (max 10 words).

Constraints: no vague adjectives, no “solutions/innovative”, one concrete proof in each version.

Takeaway

AI can make texts clearer. But real clarity starts earlier: in the decisions.

If your texts sound good today but the brand still does not come through, it is not a writing problem. It is a direction problem.