Goodwipes Spoofed OpenAI's Astra Ad in 4 Days With AI

On September 16, 2026, Adweek reported that Goodwipes — the flushable wet wipe brand founded in 2013 by Sam Nebel and Charlie Siciak — had published a 58-second spoof of OpenAI's Astra ad. In the original, polished people instruct AI companions from armchairs. In Goodwipes' version, titled "Asstra," a fart opens the spot and the camera pulls back to reveal that everyone is on a toilet. The verdict line: "The future of butt wiping has arrived."

Two details make this a case study rather than a stunt.

First, the original Astra ad has over 135 million views on X. Second, the spoof — described by Adweek as just as glossy as the original — was made in four days, using AI, by OK Future, the AI-native creative agency founded by former MullenLowe U.S. CEO Frank Cartagena.

The timeline is the story

Cartagena's explanation is the sentence to keep:

"A few years ago, we'd have spent months selling the idea, finding the money, then shooting it. And we would have missed the cultural reference point."

That is the whole argument in one line. Reactive creative — the ad that answers something happening this week — used to be structurally impossible for most brands. Not because the idea was hard, but because the process (pitch, budget, pre-production, shoot, edit, legal) ran at a speed that guaranteed the reference would be dead on arrival. By the time you were allowed to shoot the joke, nobody was talking about the thing you were joking about.

The four days were still four days of real work. Adweek's account lists character studies, storyboards, and set design — done "from your living room" rather than on a stage. The process also hit two failures that will sound familiar to anyone who has put a product into an image generator:

• The AI kept seating people on toilets with the seats down, and getting the seats up in every shot "took the better part of a day."
• The platforms sometimes added a lid to Goodwipes packs or garbled the copy on the front — which matters more than it sounds, because Goodwipes' distinctive assets are the product on the pack. As marketing SVP Meredith Diehn put it: "Our distinctive assets are so important to us, so some of the biggest challenges were the back-and-forth and questions around whether the product looked right."

Her closing note on the finished spot is the most useful part of the whole story: "It's still not quite perfect, but that's the discomfort you have to get comfortable with as a marketing leader. It's 90% right."

What an ecommerce team should take from it

Three separable lessons, in order of how much they change your week.

1. The scarce resource is permission, not production. Diehn said the campaign "wasn't a tough sell internally because there was already permission to experiment with the tech while ensuring everything remains 'human-led.'" The four-day clock only works if the decision to allow AI-assisted creative — with a named human owner and a review step — was made before the idea arrived. Teams that still negotiate that question per campaign cannot move at this speed no matter what tooling they buy. Value: turn a product URL into your first ad in about five minutes, on a plan whose commercial rights cover paid media, so the test doesn't need a legal conversation of its own.

2. The idea is not generated, and neither are your brand assets. Nothing in the tooling supplied the joke, and nothing in the tooling reliably supplied a correct pack. The model invented a lid and garbled on-pack copy; a reviewer caught it; that loop cost time. Practically, this means the parts of the ad that touch a claim — the pack, the label, the price, the rating — should come from an asset you control and can verify, not from a generator's interpretation of your catalog photo. Our own pipeline is built that way: it starts from your product URL and writes script, hook and offer around the real product, with a human review gate before anything runs.

3. "90% right" is a choice that has to be made per category. For a humor brand whose whole position is lightheartedness, shipping at 90% and moving on is a reasoned trade with a deadline attached (the reference point's shelf life). For supplements, finance, or anything with a regulated claim, that same trade is a compliance incident waiting to be filed. Decide which one you are before the campaign, not during it.

What reactive creative costs now

The reason this story is interesting to us is arithmetic, not aesthetics. If a four-day reactive campaign is now available to a mid-size brand, then reactive creative stops being a special project and becomes a budget line: "how many variants can we afford to answer the moment with?"

We can only speak to our own numbers. Prizmad turns a product URL into a finished UGC-style ad — script, hooks and offer copy written, one of 50+ avatars lip-synced to an AI voiceover, captions and music composited — in about five minutes. Token pricing is flat and published: ≈1,000 tokens per finished ad, on plans of $39/mo for 3,000 tokens (≈3 ads), $99/mo for 8,000 (≈8), and $249/mo for 35,000 (≈35), full commercial rights included.

So a four-variant reactive set — what Goodwipes effectively shipped, in one idea rather than four — is roughly a fifth of a month on Starter, at about $12 per ad. The agency relationship is still where the idea came from; the production ceiling that used to cap how many ideas you could answer with is what changed.

The caveats, stated plainly

• Parody has legal edges a generator will not flag. "Asstra" plays on a live, well-known campaign from another company. That worked here, with a named agency and an established brand voice behind it. It is not a template to copy by default — review names, marks, and comparative claims before producing, not after.
• AI disclosure rules apply to output, not intent. A platform label or an AI-disclosure obligation attaches to how the ad was made, regardless of how good it is. We keep a walkthrough of the current rules in AI ad disclosure and labels — worth reading before a reactive campaign, because a fast turnaround is exactly when nobody re-checks the policy.
• Speed does not fix a weak offer. The Goodwipes spot worked because the brand's own core message (tech advances, wiping doesn't) already existed and the audience already knew the brand for humor. Reactive creative amplifies a position you have; it does not create one.
• Human-led was a stated constraint here, not a slogan. Diehn described an existing internal rule that AI assists while a human owns the output. That is the same review gate we recommend regardless of tool: generate from real product data, approve before spend.

The pattern is consistent across the AI ad stories of the last few months: the production layer keeps getting cheaper and faster, and the parts that decide whether the money comes back — the idea, the offer, the product truth, the review — stay exactly where they were. The interesting news this week is that the slow part is no longer the render.

Start on the cheap end of the curve: turn a product URL into your first ad, or read what AI video ads actually cost before you budget a reactive set.

Related reading: how a Hollywood AI video startup actually affects ad teams, and why the production cost wall moved.