Cuisinart recently ran an A/B test pitting its own studio-produced commercials against Amazon's new generative video agent. The AI version outperformed the human-made spot, signaling a shift in how brands can produce digital advertising at scale.
why the ai beat the in-house cut
For the test, Cuisinart and agency Global Overview allocated equal budget to both videos and ran them concurrently for a month. The AI team selected a single video from Amazon's creative agent suggestions and refined it through prompt-based adjustments. "The results landed solidly in favor of the generative creative," said Kelsey Smithuysen, Amazon marketing director for Conair, which owns Cuisinart.
Nikhil Nanivadekar, Amazon Ads principal engineer for generative AI, explained that the system does not invent novel concepts. Instead, it analyzes a brand's historical ad performance and synthesizes variations of proven formats. The agent is grounded in the signals provided by the brand, he said. Shoppers develop fatigue when they see the same commercial repeatedly, so the tool continuously produces fresh material that matches successful patterns without exact duplication.
from rough drafts to campaign-ready assets
Previous attempts at generative static images failed Cuisinart's brand guidelines. Logos, appliance buttons, and specific color shades often appeared incorrect. Video output has improved enough to clear that hurdle. Smithuysen said the AI now delivers most of a complete advertisement, requiring only minor post-production fixes and upfront configuration. She added that within a year, her team expects to run the raw outputs directly in campaigns without manual editing.
The winning test video remains functional rather than polished. It shows a family preparing a meal while a narrator describes quick dinner options, interspersed with still images of a food processor that have light motion and sound. Text on the appliance interface sometimes misses the mark, but the overall execution passes brand review.
The brand plans to expand beyond simple product demos. Marketing teams want to align video environments with shopper life stages, such as moving into a new apartment or starting college. As generative models improve, these spots may eventually feature characters speaking to each other instead of relying on voiceover narration. The tool currently operates exclusively within Amazon Ads, which matches standard industry practice. Advertisers typically tailor creative aesthetics to match platform-specific viewing habits rather than repurposing identical assets across channels.
what's next for the tool
Amazon Ads will roll out a one-click video generation feature in June to reduce the friction of manual prompting. The current workflow requires advertisers to exchange multiple messages with the system to refine pacing, visuals, and audio. The upcoming update will analyze early prompts and generate final cuts automatically, cutting down iteration time. This shift moves the platform closer to the streamlined workflows many teams demand when managing high-volume e-commerce campaigns.
why this matters for creatives
Advertising designers and art directors will need to adjust how they approach campaign production. Generative systems are handling routine asset variation and initial scene composition, freeing teams to focus on strategy, brand consistency, and complex visual problems. Learning to direct these tools effectively will replace manual rendering tasks in many production pipelines. Professionals looking to build skills in this space can explore Generative Video Courses to understand prompt architecture and asset refinement. The shift does not remove the need for human judgment, but it does change where that judgment applies. Teams that learn to guide AI output will maintain control over brand standards while accelerating delivery timelines.
Your membership also unlocks:
