“We’ll use AI instead.”
The sentence can end a proposal before anyone has agreed what the tool is being asked to replace. “Instead” can mean anything from “help me get a first draft” to “take over a decision whose consequences I still own”. Those are different jobs, even if the same prompt can start both.
It is tempting for a specialist to reply that expertise is about judgement. That is true, but it is not enough. The client may be right that a particular production task is simple enough to handle directly. An expert should be able to say so without pretending every piece of work needs an expert attached to it.
The useful question is where the uncertainty sits.
An output is not the whole job
Most professional work contains several decisions: what problem to solve, what information is reliable, what to make, how to judge it, and how it will fit into the wider business. A tool may help with one or more of those stages. It does not make the stages identical.
A generated page of copy can be fluent and still describe the wrong offer. A visual can look polished while creating an expectation the business cannot meet. A code fragment can run and still be difficult to maintain inside the existing system.
The question is not simply whether the tool can produce an output. It is what the person using it has to know in order to choose, check, adapt and support that output.
Sometimes using AI directly is sensible
If the brief is clear, the result is easy to inspect, the decision is reversible and the cost of a mistake is small, a client may not need to pay someone else to produce the first version. The same can be true when the client already has the context and skill needed to judge the result.
In those cases, defending the old production step does not make it valuable. A specialist can help set a useful standard, identify the source material or review a sample without insisting on owning the whole process.
That boundary matters. Expertise should not become a toll booth around work a capable team can do safely for itself.
The harder work is deciding what good means
The balance changes when the brief is ambiguous, several constraints collide or the output will be difficult to assess without experience. Then the client may need help with the decision around the output: what question to ask, what material to trust, what a failure would look like and what should happen if the first answer is wrong.
For a brand project, that could mean deciding which audience problem matters before generating campaign options. For a website, it could mean testing whether a page helps the right person understand the offer, rather than judging it only by how polished the copy sounds.
Those are not mysterious expert instincts. They should be made visible through criteria, examples and proportionate checks. The specialist’s contribution is easier to assess when the client can see what was uncertain, what was compared and why one route fits the stated problem better.
Keep responsibility attached to decisions
Using AI does not settle who will check the result, who can approve it or who will respond if it causes a problem. Those decisions belong in the working arrangement, especially when material is private, the output will be public or a mistake would be costly to correct.
The professional is responsible for the decisions and work they take on. The client still owns the business choices it makes. A tool should not become the convenient place where responsibility disappears.
Ask what the tool is replacing
When a client says “instead”, ask:
Which part of the work do you want the tool to do?
What has already been decided, and what remains uncertain?
How will you recognise an unsuitable result?
Who has the context to check it and the authority to act on the result?
What would be costly, risky or difficult to correct later?
If those answers describe a bounded production task, using AI directly may be the right call. If they expose an unresolved decision, make that decision the work. Do not defend a fee by saying a tool cannot replace expertise. Show exactly where expertise changes the choice, and be willing to leave the rest to the client.
No published comments yet.