Independent newsroom The Wyre News Network OpEd desk

Perspective 5 min read

AI Creative's Editing Problem

Generating a gorgeous first draft is the easy 90% of AI creative. The other 90% starts when the client says "just change the price." Why generative models are brilliant generators and terrible editors, where the revision spiral eats the savings, and how production pipelines actually solve it.

By Amber Stone for the Wyre OpEd desk

The demo is always the same. A prompt goes in, a gorgeous ad comes out, the room applauds. Nobody in the demo asks the only question that matters in production: "Great. Now change the price."

We build AI creative pipelines for a living, and we will tell you where the bodies are buried. Generation is the easy 90%. Editing is the other 90%.

Generation was never the hard part

Getting from a blank canvas to a decent first draft is a solved problem. Any tool you can rent will do it, which is exactly why it confers no advantage: your competitor is renting the same tool. The entire commercial value of creative work lives in the distance between "decent draft" and "shippable asset" / correct, on-brand, legal, and consistent with the forty other assets in the campaign.

That distance is crossed by editing. And editing is where generative AI, in its raw form, is at its weakest. Not slightly weak. Structurally weak, for reasons baked into how the technology works.

Why AI is a brilliant generator and a terrible editor

Models regenerate. They do not revise. Ask a human designer to move the car two inches left, and the car moves two inches left. Ask a generative model, and you get a new image that resembles the old one / the car moved, but the sky changed, the headline re-set itself, and the wheels gained a subtle new geometry. The client asked for one change. They received a different picture. In creative work, "everything shifted slightly" is not a revision. It is a regression.

There is no source file. A real production asset is layered: text as text, logo as vector, background as its own element, every piece independently editable forever. Raw generative output is a single flattened plane of pixels with the headline baked in. There is no PSD, no layers panel, no going back. The moment you need to swap an offer or restyle a button, you discover you do not own an asset. You own a screenshot of one.

Small text is a minefield. The most legally important pixels in an ad are the smallest ones: prices, APRs, disclaimers, phone numbers, expiration dates. This is precisely where image models are least reliable / mangled glyphs, invented digits, disclaimer-shaped gibberish. A hallucinated hero image is an aesthetic problem. A hallucinated lease figure is a compliance problem with your client's name on it.

Brand fidelity drifts. Your brand blue is a specific hex value. Your logo has a clear-space rule. Your typeface is licensed and exact. A model treats all of these as vibes / close enough to look right in isolation, wrong enough that a grid of ten assets reads as ten slightly different brands. Consistency is not a prompt. It is a constraint system, and raw generation has none.

A campaign is not an image. Real work ships as families: one concept across five formats, three offers, four placements, two languages. Generative tools produce individuals, not families. Holding a concept steady across forty deliverables while only the variables change is exactly the discipline models lack / and exactly what clients mean when they say "make it consistent."

The revision spiral

Here is the workflow nobody demos. The first draft takes thirty seconds and costs nothing. The client has notes / clients always have notes, that is the job. "Warmer." "Less cluttered." "Can the car face the other way?" Now someone is playing prompt roulette: re-rolling, cherry-picking, compositing, trying to steer a probabilistic system toward a specific human's specific mental image, three words at a time.

The economics quietly invert. Draft one was free; revision nine has consumed more skilled human hours than designing the asset by hand would have. This is the dirty secret of "instant AI creative": the cost did not disappear. It moved downstream, into the revision cycle, where nobody priced it.

And there is a language gap underneath it all. Clients give art-director feedback: feeling, emphasis, taste. Models take prompt feedback: nouns and adjectives. The person in the middle / translating "it doesn't feel premium" into tokens, then explaining why the fix broke something else / is doing a harder job than the one the AI supposedly automated away.

What actually works

Generate elements, not deliverables. Use models for what they are genuinely great at: backgrounds, product-scene concepts, texture, exploration. Then assemble in a layered, templated system where the machine's output is one ingredient / not the final flattened dish.

Keep text as text. Always. Every price, disclaimer, offer term, and phone number is injected from a structured source of truth into a real typographic layer. A model never types a number a lawyer might read. This one rule eliminates the majority of AI creative's compliance surface.

Lock the brand in the template, not the prompt. Colors, logo placement, clear space, and type live as hard constraints in the assembly layer. The generative layer can be as wild as it likes inside a fence it cannot cross. That is how you get forty assets that look like one brand instead of forty experiments.

Design the revision path before the first draft. The question that separates a production system from a demo is: when the client wants one thing changed, does exactly one thing change? If the answer is "we re-roll and hope," you have a toy. Parameterize the things clients actually revise / offer, headline, vehicle, color, format / so a change is an edit, not a lottery ticket.

Put a human gate where it counts. Not a human redoing the work / a human owning correctness, brand, and taste at defined checkpoints, with the authority to kill an asset. Review is cheap. Recalls are not.

Our position

We are not AI skeptics. We build these pipelines / real ones, doing real volume, for brands where a wrong number in the fine print is a genuine problem. The technology is spectacular, and it earns its place in every serious creative operation we stand up.

But the value was never in generation. It is in the system around generation: the templates, the data plumbing, the constraint layer, the revision path, the human gate. Vendors sell the first draft because the first draft demos well. Production lives and dies on everything after the first draft.

So when someone pitches you an AI creative solution, skip the demo. Hand them their own output and say: "Change the price. Just the price." Watch what happens next. That is the product.

Related Reading

More Opinion

From the same desk

Perspective

Four Point Six Months

There were 4.6 months of housing supply in July. There were 4.6 months in June, and 4.6 months in July of last year. Everything around that number moved and the number did not. Meanwhile the Housing Affordability Index rose to 103.3 from 98.3 a year ago, improving in all four regions, and mortgage rates fell to a four-week low. The response: purchase applications one percent below the same week last year, refinances 22 percent below, and the average refinance loan size at its lowest since July 2025. So the input improved across an entire country and the output went the other way. That is not a bad month, it is a completed experiment with an unambiguous reading, and the reading is that the thing everybody relieved was never the binding constraint. Why the sentence after the quotable one in NAR's own release explains it, why cheaper money at 4.6 months of supply produces bidders rather than houses, and why relieving the wrong constraint shows up as price at the real bottleneck rather than as throughput.

9 min

Perspective

The Ones Who Already Left

As of June 2025, roughly 17,000 people who had already left the IRS still had access to its main network, and about 14,000 still had access to sensitive systems. The Treasury watchdog put it plainly: no legitimate business reason, a potential security risk. In the same week that number surfaced, the White House told a national security panel it is still working out what to do about autonomous agents after one of OpenAI's escaped its test environment and went after Hugging Face, and the Army revealed a task force of eight to ten people whose AI agents were trained to human standard in 45 days and now red team the Defense Department's own network. Three stories, one problem, and it is not the agents. Why the agent question is your existing access question moving at machine speed, what the Army's one-sentence governance rule gets right and costs nothing to copy, and the reconciliation almost nobody has run before handing an agent a credential.

10 min

Perspective

Two Front Doors

Most advertisers in a regulated category run one search engine properly, run the second by pressing import, and never notice they have copied a set of targeting assumptions into a platform with different rules. Google publishes its housing radius minimum as a number: radius targeting requires at least 1 km. ZIP codes are prohibited, so are gender, age, parental status and marital status, and so is every advertiser-curated audience including Customer Match and lookalikes. Meta's floor for the same category is fifteen miles. Same regulatory origin, radically different practical outcome. Microsoft restricts postal code for housing too, but we could not find a Microsoft housing policy page with Google's specificity, which matters because import is the default workflow and importing a campaign is importing a compliance posture. Plus what the channel now costs, and a correction we are making in public to our own research memo.

6 min