Headspace, a leading provider of digital resources for mental health and wellness, is on a mission to guide consumers to more joy, less stress and better sleep. But with no one-size-fits-all solution and countless use cases for Headspace’s offerings, it can be challenging to get the right resources and right messages to the right consumers at scale. Heading into their busiest season—the holidays—we turned to generative AI to power a new ad campaign designed to be as personalized as possible.

The process began with gaining a deeper understanding of the prospective consumers. Using Persona.Flow, we gathered consumer research insights about the seasonal stressors that impacted the target audience’s mental health during the holidays—a process completed in a matter of minutes rather than the days or weeks required by a traditional research program. This efficiency enabled us to get past the monolith of a single persona and dig into the nuances of multiple consumer profiles to springboard creative concepts that could be adapted for everyone.

We developed a modular ad concept that could pair seasonal emotional challenges with specific types of resources available from Headspace. Using Monks.Flow, our AI suite of tools, we combined ad elements—AI-generated backgrounds, product interface imagery and branded illustrations and iconography—to produce hundreds of assets at scale. For static and motion versions in multiple aspect ratios, production hours would have quickly racked up in a traditional design process. With AI-powered image generation and production, we had assets for 20 distinct use cases ready for approval quickly, and overall cut production time by two-thirds.

The ads were designed for Meta’s Advantage+ campaigns, the machine learning-powered product to deliver ads to the right users. By combining AI creative and ad delivery, we created a highly relevant experience for the users, leading to higher-quality site traffic. North American users who saw the AI-assisted holiday assets through Advantage+ converted to a signup at a 62% higher rate than Headspace’s other Advantage+ traffic.

Initially, the AI assets started out with a lower CTR than traditional ads while the learning phase was in progress. However, we improved performance by providing a larger round of AI-assisted creatives for the New Year—and the improvement was clear. With more data to work with, the algorithm could successfully narrow the CTR gap between the test and control campaigns. The AI-assisted campaigns became even more efficient, driving over 10% more conversions at a 13% better cost-per-signup compared to the control campaign.

With over 460 original ad assets designed from scratch for personalized testing, we helped Headspace work smarter, not harder, using generative AI. By combining multiple tools—from our proprietary Monks.Flow and Firefly generative technology from Adobe—with the expertise to steer the ad delivery algorithms by feeding campaigns the right asset fuel, Headspace is able to get even more people access to mental health resources, helping them lead more stress-free, joyful lives.

Few campaigns are as iconic as Dove’s Real Beauty campaign, well-known for challenging traditional beauty standards and celebrating inclusivity. Marking the campaign’s 20th anniversary, Dove sought to show that 20 years on, the need for better diversity and inclusive representation is still as relevant today—especially with concerns about AI-generated content. The Code: A Dove Film—created by Soko—was designed to bring awareness of the increasing impact of AI on beauty and was established on the key insight that by 2025, AI is predicted to generate 90% of online content. The film presents biases exacerbated by AI in contrast to the Dove-coded imagery, which shows that AI has learned from the brand’s long legacy of portraying diverse women and body types in the pursuit of real beauty.

The film served as an important anthem standing for Dove’s values, though Dove wanted to extend the campaign with a resource that substantiated its message and encouraged the industry to participate in challenging biases inherent in generative AI output. The brand believes that we are at a pivotal moment at the beginning of the AI era, where we can make a change in how AI depicts beauty. So, we had the idea to build an AI prompting playbook and other assets designed to answer the very question presented in the film: What kind of Beauty do we want AI to learn?

As an AI consulting partner, we’re well aware that generative AI isn’t going anywhere—and that it’s going to change everything. Starting out, we sought to define how we can take action now to help steer generative AI outputs in the right direction, with the goal of accurately representing diverse communities with AI. With tight timelines needed due to the pace at which AI evolves, our experience design experts worked alongside our research and development team to test prompts and outputs, resulting in a first-of-its-kind guide of best practices.

Achieving this level of nuanced representation in a guide to prompting AI hadn’t been done before, so we needed to test and explore different prompts for various use cases and types of women. For example, what’s the best way to prompt for a Filipino woman with a prosthetic leg and big, curly hair? The Real Beauty Prompt Playbook provides broad guidance on beauty and inclusion, specific information on the power of prompts, and recommendations on how to create prompts of your own.

With the overarching Dove Code campaign designed and developed by Soko, we collaborated closely with Dove’s roster of partners and sought external specialists where relevant, including female AI experts and a global community of body confidence experts to ensure we brought together best-in-class thinking across every aspect of AI. We also had guidance from the AI researcher who discovered the key insight used in the Dove Code film, ensuring continuity between the questions asked by the film and the answers provided in the playbook.

Beyond the playbook itself, we continued to support the Dove Code with an above-the-line out-of-home (OOH) campaign, including a CGI-rendered OOH ad celebrating the 20th anniversary of the Dove Real Beauty campaign on digital platforms. Overall, our efforts helped substantiate Dove’s message about ensuring diversity and inclusion in the age of AI. As one core partner among many, it took a village to build the Dove Code in celebration of the brand’s lasting commitment to real beauty—and we couldn’t be prouder.

French passenger rail company SNCF Voyageurs sought a means of producing high-volume hygiene content for web and social, without the need for 100 briefs, 100 days of planning, 80 days of post-production… you get the picture. So, we partnered with the brand to fast-track the creation of 230 visual assets, accelerated with the use of generative AI and strategically automated AI workflows. By harnessing the power of AI-generated creative, we were able to enhance productivity and refine the artistic process, all while staying true to the essence of the brand.

First, we developed a content matrix, which structured the various kinds of scenes to depict related to specific seasons or weather, time of day, holiday type, whether we wanted to depict a traveler or an SNCF employee, and more. This helped us automate and generate hundreds of combinations for landscape and people images.

Next, it was time to create the generative AI model. We began this process by reviewing brand guideline documentation, which served as a reference to generate photography that aligned with the brand and the needs of its audiences. We tested various models in Stable Diffusion to nail the results we wanted to generate at scale. We used TopazLabs Photo AI to upscale and retouch AI-generated assets at lightning speed, along with additional retouches in Photoshop, including the software’s Generative Fill feature to extend backgrounds for wide header images.

Navigating legal nuances and the ownership of intellectual property were at the forefront of our engagement with SNCF Voyageurs. We helped the brand create a framework that secured both IP protection and data privacy protection, enabling the brand to continue to use the AI-trained models without issues in ownership moving forward. As a result, SNCF Voyageurs will continue to see cost savings using the AI workflows we’ve developed with them, compared to the budgetary and time costs that come with traditional photoshoots.