Google's Asset Studio Upgrade: End-to-End Ad Creative With Gemini, Veo 3.1, and Nano Banana
What Asset Studio actually does now
At Google Marketing Live on May 20, 2026, Asset Studio got rebuilt into a multimodal creative engine. The old loop had you writing copy, shooting images, and cutting video yourself. The new loop flips that: you hand it a product URL, a marketing brief, and your goals, and it starts producing assets — copy, images, and video, all in one pass.
Four models split the work behind the scenes. Gemini reads your brief and pulls product details from the URL, then drafts the copy. Veo 3.1 handles video. Nano Banana, Google’s image model, generates the stills. A fourth model, Gemini Omni, also contributes to video generation. These aren’t running in isolation. They’re chained into a pipeline that carries a project from storyboard all the way to a finished YouTube ad in the same place.
Here is the division of labor, so the rest of this makes sense:
| Model | What it generates | Typical use |
|---|---|---|
| Gemini | Copy, headlines, value props | Reads the URL and brief, turns product info into usable lines |
| Veo 3.1 | Video clips, finished cuts | Converts a storyboard into a shot-by-shot short video |
| Nano Banana | Stills, product image variants | Produces a consistent set of images for PMax and Demand Gen |
| Gemini Omni | Video generation (alongside Veo) | Fills the gap from frames to finished cut |
For a small DTC brand with no creative team, the upgrade doesn’t change one feature. It changes the cost structure. The old question was “can we afford to shoot this.” The new question is “how many variants can we test this week.” Those are very different problems to be solving.
The global English rollout is planned for summer 2026, so if you open it today and the full feature set isn’t there yet, that’s expected. Wait it out.
Brief to storyboard to video, step by step
Start with the inputs. You need two things: a product landing page URL and a brief that actually says something. The URL lets Gemini scrape the product details — price, value props, imagery — so you skip filling in a dozen fields by hand. The brief should name the audience you’re after, the angle you want to push, and the tone. The more specific it is, the less the output drifts.
Have it generate copy and a storyboard first. Don’t rush to a finished cut. Gemini will return a few copy directions, and Veo will hand back a storyboard structure. This is your best moment to steer. Changing direction at the storyboard stage costs almost nothing; killing a fully rendered video to redo it burns both time and compute. If the way the product shows up is wrong, or the pacing feels off, fix it right there.
Once the storyboard holds up, let it render. Veo 3.1 and Gemini Omni turn the boards into a finished cut while Nano Banana produces a matching set of stills. What you walk away with isn’t a lone video. It’s a full kit: video for YouTube and Demand Gen, images for PMax, copy to spread across placements.
Then drop it into Google Ads. The whole run happens in one place, brief to ready-to-ship YouTube ad, no exporting and re-importing between tools. For a small team, that means you don’t need someone who knows an editing suite, and you don’t outsource a round and wait three days.
One caution worth repeating: a vague brief produces vague output. Plenty of people treat AI as “I’ll just describe it loosely and it’ll get me,” then wonder why everything comes back generic. Ten minutes spent making the brief concrete beats ten rounds of fixes later.
The Merchant Center to Demand Gen video pipeline
The same May 20 batch shipped a quieter update that’s easy to miss and genuinely useful: Merchant Center product videos can now flow straight into Demand Gen.
Here’s how it works. If you already have product videos in Merchant Center — many sellers do — those videos can now feed Demand Gen for personalized video ads at scale. No re-uploading, no reconfiguring. The assets sitting in your catalog get reused as-is. For sellers with a lot of SKUs, that’s a real chunk of work removed.
The back half goes further. A Demand Gen viewer who sees an ad and decides to buy can be routed directly to the merchant checkout page. One fewer hop sits between “saw the ad” and “paid.” Video does the seeding, the link does the closing, and the two ends connect.
Read this pipeline next to Asset Studio and the shape gets clear. Asset Studio mass-produces video, Merchant Center reuses what you already have, and Demand Gen distributes it and connects to conversion. One makes, one stores, one sells. The connective tissue small teams always lacked staff for is now stitched together by the platform.
Worth noting on the targeting side: Performance Max also picked up a new “search themes” beta that lets you nudge PMax onto placements it hasn’t been reaching. If your PMax has felt like a black box, that’s a small handle to grab.
Using one-click A/B testing without burning spend
The release added 1-Click Creative Testing to surface your top performers. The feature is effortless to start, and effortless is exactly how budgets leak. A few rules I’d hold to.
Don’t dump a pile of variants in and brute-force it. Cheaper generation doesn’t mean cheaper testing — every variant still needs impressions to produce data. Run ten versions at once and none of them gets enough volume to read. Narrow to three or four versions with a clear hypothesis behind each, like “swap the opening hook” or “reorder the value props,” then test.
Keep each variant one variable apart. If you want to know whether the problem is the first three seconds or the value prop, change one thing at a time. A video that swaps the open, the copy, and the color grade all at once will tell you nothing useful even if it wins, because you can’t reproduce why.
Wait for enough data before calling it. One-click testing will hand you a “leader” fast, but an early lead is usually noise. Let it run to a sample size that stands up before you act, instead of piling spend onto whatever’s ahead on day two.
Most important: test generated creative against your own-shot creative rather than assuming the generated assets win. Put your best-performing real footage in as the control. Generated video is often cheaper and faster, but in some categories human-shot footage simply converts better. Let the data settle it, not the gut feeling that AI-made must be the smarter pick.
Where human review still earns its keep
However smooth the tooling gets, AI-generated creative still needs a human pass, especially anywhere brand accuracy or product claims are involved.
Brand accuracy is the first gate. Whether Nano Banana’s stills and Veo’s video render the product correctly — right colors, right logo placement — is something AI gets wrong now and then, and a wrong asset shipped live is an incident. A quick look at each set before it goes out costs almost nothing.
Claims are the second gate and the riskier one. Gemini might surface an unsupported efficacy line or an inflated promise in the copy. Ship that and you’re looking at rejected ads at best, compliance trouble at worst. Any sentence carrying a number, a “best,” a “first,” or a “guaranteed” gets checked against something real before it runs.
Storyboard control is your main lever against drift, and it bears restating: lock the direction at the storyboard stage, because fixing it after the cut is rendered costs far more. AI can’t fully absorb how you think about your brand. The brief and the storyboard are the two openings where you push that brand voice into the pipeline.
The longer-term hazard is brand drift. Generate dozens of assets a week, each a little off from the last, and a quarter later your ads might not read like the same brand. Pull a recent batch every so often and ask whether it still looks like you. Fast generation is a gift, but consistency needs a person watching it. The platform won’t do that part for you.
Related Articles
Google AI Brief in Practice: Three Layers of Natural Language Control for AI Max
AI Brief lets advertisers steer Gemini inside AI Max using plain English across three layers: Messaging Guidelines, Matching Guidelines, and Audience Guidelines. Here is how to use each layer and what cross-border sellers should write in each field.
Google Veo 3 in Google Ads: Generate Video Ads From Product Images at No Cost
Since May 6, 2026, Veo 3 image-to-video is built into every active Google Ads account for free inside Asset Studio. Upload product photos, get a 10-second HD video with ambient audio. Here is the full workflow and what it actually produces.
2026 AI UGC Video Tools Compared: Arcads vs Creatify vs HeyGen vs MakeUGC
AI-generated UGC video tools exploded in 2026. A side-by-side comparison of Arcads, Creatify, HeyGen, MakeUGC, and Tagshop on pricing, avatar libraries, product feed integration, and bulk output capacity.