Pricing AI work, when the render is cheap and the judgement is not

When production was slow, time was a reasonable proxy for value and nobody examined it. Generative tools break that proxy visibly — the client can see that the render took minutes, and an hourly quote invites them to do arithmetic you will lose.

The answer is not to hide how the work was made. It is to price the thing that is actually scarce, which was never the rendering.

What is scarce is knowing which shot to make, whether it is on brand, whether it will hold up, and being accountable for it. That is what the number is for, and it needs saying plainly rather than being smuggled into a day rate.

• Judgement. Which direction, which frame, what to reject. This is the compressed form of years of work and it is why the output is usable.
• Accountability. If it is wrong, you fix it. A tool has no such obligation and that difference is the entire service.
• Reliability. Not one good frame — a set, on a date, meeting a specification.
• Risk absorption. Guardrails, legal safety, brand fit. You are the reason nothing embarrassing ships.

A client can generate images themselves. They cannot generate these four, which is why they are talking to you.

Quote a fixed fee for a defined set: this many assets, in these formats, meeting this specification, with this many revision rounds, by this date.

The advantages are structural. Your speed benefits you rather than reducing your fee. The client knows the cost up front. And the conversation stays on what they receive, which is what they actually care about.

The requirement is a tight definition. Fixed pricing on a vague scope is how jobs go wrong, so the specification has to be genuinely specific: counts, ratios, what constitutes a revision, what constitutes a new shot.

Start from what the deliverable would cost produced conventionally — a shoot, a crew, a location, a retouch budget. That is the number in the client's head, and it is a legitimate anchor because you are delivering the same outcome.

Then adjust down for what you genuinely are not doing, and up for what a shoot could not have given them: more variants, faster turnaround, changes after the fact.

Do not adjust down to your cost. Pricing from your own effort in a world where effort collapsed is how a whole category of work devalues itself, and the client did not ask you to.

Hiding the method is a bad position. It is discoverable, and being caught reframes the whole relationship around a deception rather than around the work.

Leading with it is also a mistake, because it makes the tool the subject. The client starts evaluating the tool, which they can also buy, rather than the outcome, which they cannot produce.

The workable position is matter-of-fact: this is part of how the studio works, here is the output, here is what I am accountable for. Answer questions directly if asked, and keep the conversation on the deliverable.

Every job has fixed overheads regardless of size: the brief, the calls, the revisions, the delivery, the admin. A small job carries almost the same overhead as a large one, which means below a certain fee the work is actively costing you.

Know that number and hold it. Declining is easier when you have calculated it, because it stops being a judgement about the client and becomes arithmetic.

The counterintuitive part is that holding a floor tends to increase enquiry quality over time. Clients who are price-shopping move on, and the ones who remain are the ones for whom the four things you are actually selling are worth paying for.