10 AI marketing campaigns: real examples and creative lessons
Quick answer
An AI marketing campaign uses artificial intelligence for a defined part of planning, creation, personalization, delivery, or measurement. In the creative examples below, Coca-Cola invited fan-made AI artwork, Heinz used generated ketchup imagery as the idea, Virgin Voyages personalized video invitations, and Nike created an imagined match between versions of Serena Williams. The useful question is what AI contributed and what a smaller team can adapt.
Below are 10 documented examples, with sources explaining what AI actually contributed and a practical idea to adapt. This is a selection of creative approaches, not a ranking by sales lift. Brand and agency case studies establish how a campaign was presented; they do not by themselves prove incremental revenue.
1. Coca-Cola: Create Real Magic
AI's role: fan-created brand artwork. In 2023, Coca-Cola invited creators to use generative AI with recognizable brand assets. Its account of the initiative describes the creative platform and a subsequent summit for participating artists.
The idea gave people something more specific to do than “try AI”: create within a familiar visual world. That connection between the brand and the creative task is what makes the example useful.
Adapt the idea: offer a small set of approved assets and a clear theme for submissions. Define usage permissions and moderation before inviting contributions. Judge participation and the quality of usable work separately from sales.
2. Heinz: AI Ketchup
AI's role: generated imagery as the campaign idea. Rethink's AI Ketchup campaign entry uses AI-generated ketchup imagery to make a point about Heinz's recognizable branding.
The execution turns a simple prompt into a visual brand-recognition story. It should not be read as a scientific test showing that every model independently prefers Heinz. Prompt selection, training data and the images chosen for publication all affect what an audience sees.
Adapt the idea: identify a distinctive asset your audience already associates with you, then build the creative around that recognition. Do not describe selected generated examples as an objective market survey.
3. Virgin Voyages: Jen AI
AI's role: personalized video invitations. Virgin Voyages used an AI-powered Jennifer Lopez experience to help people invite friends and family on a cruise. VML's case study describes the combination of personalization, video and a social invitation.
The useful mechanism is the invitation: the user creates something to send to another person. The celebrity is part of this particular execution, not a prerequisite for the underlying idea.
Adapt the idea: build a personalized invitation around a real event, milestone or shared activity. Use a presenter or likeness you are authorized to use. Track invitations completed and the subsequent booking path separately; sharing an invitation is not a booking.
4. Nike: Never Done Evolving
AI's role: a simulated sports story. For Nike's 50th anniversary in 2022, AKQA modeled different eras of Serena Williams to stage a match between her younger and later selves. The agency's production account explains how gameplay data informed the simulation.
This is a useful example of starting with a narrative question—how had Serena's game evolved?—then choosing technology to explore it. It is not simply a text prompt producing a finished commercial.
Adapt the idea: turn a documented change into a visual story, such as a product's design evolution. Label simulations and reconstructions clearly. Do not present an imagined comparison as recorded performance.
5. BMW: The Ultimate AI Masterpiece
AI's role: generated artwork projected onto a car. BMW's 2021 project combined AI-generated art with a physical BMW 8 Series Gran Coupé. Its official announcement describes the artists, training material and projection concept.
The car remains a physical product, while the generated artwork changes its presentation. That is an important distinction for brands worried about AI altering the item being sold.
Adapt the idea: preserve the real product and experiment with the setting or visual treatment around it. Compare the result with approved product photography so the creative does not change colors, packaging or included features.
6. Mango Teen: Sunset Dream
AI's role: fashion campaign imagery. Mango announced its Sunset Dream campaign for Mango Teen in July 2024. The company describes collaboration across design, styling, art direction, AI model work and its photography studio.
The production account matters because “AI campaign” can hide the human work around the output. A fashion image still needs to represent the garment, even when the surrounding presentation is generated.
Adapt the idea: begin with approved product references and have someone responsible for the product check the final image. Review fabric details, fit, construction and color rather than approving an image solely because it looks polished.
7. H&M: digital-twin fashion imagery
AI's role: imagery using digital twins. H&M published its first digital-twin campaign imagery in July 2025, describing it as an exploration of creative possibilities with AI.
This example is useful for the production question it raises: what can a team vary while keeping a recognizable subject and a coherent art direction? It is not evidence that digital models remove the need for creative teams or product review.
Adapt the idea: agree on likeness permissions and the intended uses before creating a digital representation of a person. Keep approval of the person, garment and setting explicit. A digital twin is not permission to generate any scene involving that person.
8. Burger King: Million Dollar Whopper
AI's role: turning a customer's product idea into shareable creative. Burger King's 2024 contest let participants imagine a Whopper. The official AI terms describe generated assets associated with the experience, including imagery and a personalized jingle.
The creative task fits the brand's customization theme. It also gives a participant an artifact to share rather than only a form to submit. The existence of the campaign does not substantiate an unqualified claim that it sold out or caused a sales increase.
Adapt the idea: ask customers to assemble an idea within clear limits, then generate a shareable preview. Make it clear which options are concepts and which products are actually available to buy.
9. Toys“R”Us: a Sora-assisted origin film
AI's role: generated scenes for a brand-origin story. Toys“R”Us and Native Foreign created a 2024 film imagining the young Charles Lazarus and the origins of Geoffrey the Giraffe. The brand's studio page identifies Sora as part of the production.
Treat the film as a historical example of a production approach, not a promise about the availability or limits of the tool today. It is also a dramatization, not archival footage of the founder's childhood.
Adapt the idea: storyboard a short brand story and distinguish invented scenes from documented history. Use text-to-video for concept exploration, then review continuity and the truthfulness of the story before publishing.
10. Lexus: Driven by Intuition
AI's role: developing an advertising script. Lexus's 2018 Driven by Intuition announcement describes an AI scriptwriting process using IBM Watson, with the film directed by Kevin Macdonald.
This example separates the idea stage from the production stage: an AI-assisted script did not mean that the finished footage, performances and direction were generated by AI.
Adapt the idea: use a model to propose story structures, then have a person check whether the story actually communicates the product's benefit. Keep the approved brief and factual claims fixed while exploring alternatives.
What do these campaigns have in common?
They give AI a defined creative job. The brand remains recognizable, the audience has something to watch or do, and the execution serves a story beyond the novelty of the tool. Those are creative principles to adapt, not guarantees of commercial success.
A useful campaign source identifies the brand, the execution, the technology's role and the result being reported. A general recommendation engine, an operations platform or a virtual try-on feature is not automatically a specific AI advertising campaign. That is why those broader software examples are not part of this list.
How to adapt an idea without a global-brand budget
- Choose one customer question and one approved claim. For a product launch, that might be what the item looks like in a real setting.
- Prepare a real product photo or recorded demonstration. Keep the product fixed while changing the scene or message around it.
- Plan a short sequence: hook, product evidence, next step. Use these product video script templates to draft it.
- Make a small set of variants and review the complete exports. Reject altered labels, invented features and unsupported testimonials.
- Link the creative to the matching product page and measure qualified actions. Keep the offer and audience consistent when comparing versions.
If you have a product photo, start with Image to Video. If you need a platform for an assembled ecommerce ad or presenter-led explainer, compare the ecommerce video tools by workflow before committing to a plan.
Try text-to-video
Curious how your brand could create impactful AI-driven videos? Explore Magic Hour’s text-to-video platform to bring your ideas to life.
Try Text to VideoHow should you measure an AI campaign?
Choose the metric before production. For participation, count completed submissions or invitations. For acquisition, examine qualified visits, purchases and acquisition cost. Include production and editing costs when judging efficiency.
An agency's reported views or impressions can describe reach, but they do not isolate the effect of AI or prove profit. For a practical setup, see our guide to AI in advertising.
Frequently asked questions
A clear idea, recognizable brand assets, accurate claims and an execution suited to the audience. AI should help perform a specific creative job. Judge business performance using the campaign's objective rather than the presence of AI.
They use AI in different ways. Lexus used it in script development; BMW combined generated art with a physical vehicle; other examples use generated images or personalized video. The contribution should be stated explicitly rather than treating every production as fully automated.
Yes. Start with a short, accurate product clip and one clear next step. The reusable lesson from a large campaign may be its invitation, story structure or treatment of a brand asset—not its celebrity, budget or technology stack.
Runbo Li is the Co-founder and CEO of Magic Hour, where he builds AI video and image tools for content creation. He is a Y Combinator W24 founder and former Data Scientist at Meta, where he worked on 0-1 consumer social products in New Product Experimentation. He writes about AI video generation, AI image creation, creative workflows, and creator tools.
