how uramaki studio works real workflow screenshot concept
ProductPROMOTIONAL7 min readApr 9, 2026

How Uramaki Studio Works: Single Prompt to Full Social Campaign (Real Walkthrough)

By Uramaki Studio Editorial Team

A step-by-step walkthrough of what one Uramaki Studio run does—from brief to planned slides, visuals, and captions—so you know what to expect before you publish.

The brief we started from

To show what one run actually produces, here is a real one — a fictional brand, a real generation, and an honest account of what needed changing.

The brief:

Artisan coffee brand for remote workers. Five-slide Instagram carousel explaining why the coffee you make at home tastes worse than the café's, even with the same beans. Audience is people working from home who have bought a decent grinder and are still disappointed. End on a soft CTA to the starter subscription.

Six things in there: format, platform, length, audience, their specific situation, the argument, and how it ends. That took about ninety seconds to write and it is the part that determines everything downstream.

What came back

Five slides with a planned arc rather than five separate statements.

**Slide 1.** "Your beans are fine. Your water isn't." — the counterintuitive hook that reframes the problem.

**Slide 2.** Why mineral content changes extraction, in two sentences. The slide that justifies the swipe.

**Slide 3.** The specific fix — filtered water, and the ratio that matters.

**Slide 4.** What it will not fix, which is grind consistency. The honest limit.

**Slide 5.** The call to action, alone.

Each slide came with a headline, body copy sized for the slide, and an image brief. Plus a caption for Instagram and a variant for Facebook.

The arc is the part worth noticing. Slide two advances the argument rather than restating slide one, and slide four concedes something — which is what makes slide five credible.

Formats and platforms

The same campaign produced the Instagram carousel at 4:5, a Facebook variant with a longer caption written for a reader who will actually read it, and a set of base images at export resolution.

What it did not produce, and what nothing does: a decision about whether this was the right campaign to run this month.

What needed editing

Four things, and this is the honest part.

**Slide 1's headline wrapped to a second line** at the brand's title size. Rewritten from "Your beans are fine. Your water isn't." to "It is not the beans. It is the water." — same claim, one line.

**Slide 3 was too general.** It said "use filtered water" without the number. Added the specific mineral range, because the whole appeal of this brand is precision.

**One image was wrong.** The generated scene for slide two showed a café counter; the campaign is about home. Regenerated with "domestic kitchen, morning light" and it was right on the second attempt.

**The caption opened with a warm-up sentence.** Deleted it. The second sentence was the real opening, which is true of most first drafts.

Total editing time: about twelve minutes.

The honest timing

From typing the brief to having five slides and two captions ready: roughly four minutes of generation, twelve minutes of editing.

That is not "60 seconds", and any tool claiming otherwise is measuring generation and ignoring review. The comparison that matters is against the alternative: the same carousel produced conventionally is three to four hours, most of it spent deciding the angle and writing the first headline.

What the tooling removed is not the work. It is the blank page and the layout labour.

What it did not do

The brief. The argument about water was mine — the model would happily have written five slides about bean freshness, which is the consensus take and therefore the generic one.

The final render also happens in the app rather than in a chat, because the composite of copy, logo and swipe label onto the image is done in the browser.

And it did not tell me whether this campaign was worth running. Nothing does.

Who this suits

Volume, mostly. Several campaigns a month, where the constraint is production rather than ideas. Multiple brands, where consistency of voice is the thing that slips. And people who arrive at the angle by talking it through and want the artefact to exist by the end.

Not for: two posts a month, single reactive images, or anything where the product must be photographed exactly.

For the copy structures used inside the slides, copywriting formulas covers the patterns, and writing a brief covers the ninety seconds that decide the output.

FAQ

How long does it really take to produce a carousel with Uramaki?

About four minutes of generation and ten to fifteen of editing. Claims of under a minute measure generation only and skip the review, which is not optional.

What does the AI decide, and what do I decide?

You decide the argument, the audience and how it ends. It decides the arc across slides, the phrasing and the images. The argument is the part that determines whether it is generic.

How much editing does the output need?

Typically a headline that runs long, one image that missed the setting, and a caption opening that warms up. Around twelve minutes for a five-slide carousel.

Can it generate photos of my actual product?

No, and it should not. Generated imagery covers scenes, backgrounds and concepts; anything a customer receives needs photography.

Does it write captions for more than one platform?

Yes — a variant per selected platform, shaped for how each is read rather than the same text repeated.

Ready to generate faster campaigns?

Generate your first campaign free on Uramaki Studio.

Related articles