One of the most exciting conversations at AdLab 2025 was the panel, “AI in Health Marketing: Real-World Strategies, Use Cases, and Ethics.” As AI continues to reshape how healthcare marketers target, optimize, and engage, panelists cut through the noise to explore what’s actually working and what’s most promising.
Moderated by DeepIntent Chief Technology Officer Anton Yazovskiy, the panel brought together Oleg Korenfeld, Chief Technology Officer at CMI Media Group and Compas; Parvati Vaish, SVP, Consulting at CSA; and Alfred Whitehead, EVP of Applied Sciences at Klick Health.
Here are the five key takeaways.
Panelists first established an essential point: AI in healthcare marketing is only as good as the data it’s built on. Outputs are unreliable without trustworthy inputs, especially in healthcare. As CMI’s Oleg Korenfeld said, “With the introduction of AI, validation of how good the data set is is even more important. It’s garbage in, garbage out.”
CSA’s Parvati Vaish highlighted that clean rooms are key to obtaining higher-quality data sets. “I think clean rooms—unlike in the CPG space where you’re just looking at exposure and sale—give us the ability to create connection between patient IDs and claims data and individual HCPs and their new patient starts. We can really start stitching together a far richer picture of what these engagement journeys can look like.”
Whether it’s ensuring clean inputs or fleshing out patient and provider journeys, thoughtful data strategy is what enables AI to deliver meaningful, real-world impact in healthcare marketing.
Across the panel, experimentation was championed as an imperative. Klick’s Alfred Whitehead discussed how to open the door to responsible, iterative innovation. “Perfection can be the enemy of good,” he said. “If you can get 80%, that’s great. Then, get it up to 85, 89, 90.” Such incremental improvements can still meaningfully improve outcomes.
“Test and learn is the strongest way forward,” Vaish affirmed. In a rapidly evolving AI landscape, the willingness to experiment, and to learn from imperfect results, is a strategic advantage.
Despite the tendency to speak of it as one giant machine, AI is really a collection of specialized tools. “Whoever built an AI tool probably built it for something specific in mind,” said Korenfeld. “AI is not one big monolith.” This point came into sharp focus when he described building a media-planner bot using generative AI. “It may be good at text, and it may be good at images, but it’s not that great at numbers yet.”
Moderator Anton Yazovskiy agreed, saying, “There is no one algo—it’s not how it works. You can’t just use ChatGPT for everything.” Real impact comes not from chasing hype, but from matching each problem with the model best suited to solve it.
No matter how advanced the models become, human oversight remains essential. “We want to have not just artificial intelligence—we want human intelligence as part of this,” Vaish emphasized. “At the end of the day, we are not marketing to machines. We are marketing to people.”
In a similar vein, Whitehead explained that his aim isn’t to replace people but to empower them. “One of my goals for everybody in our organization is to spend six hours with an AI system… to learn how you can do more in your job with it, but also how it can change the way you work.”
When asked what innovations they’re most excited about in the next year or two, panelists pointed to the rise of more connected, end-to-end AI systems. Whitehead said, “Something we’re just starting to work with now is the new generation of reasoning and deep research models. It’s very exciting for us in healthcare, this idea of having an AI that can read 2000 scientific papers and give you sort of the deep insights there to share…I think when that gets applied to social media, it’s going to be super interesting to see how we develop messages for our consumer audiences.”
“I’m excited to see more of the solutions stitched together more effectively,” said Vaish. “I would love to see [closed-loop systems] come to life more seamlessly…Ideally, we are rapidly creating and deploying and measuring our assets and markets so that we can continue refining more effectively.”
Korenfeld emphasized that one of AI’s most transformative uses is also one of its most established: automating time-consuming manual processes. “Taking hours and hours of manual process and automating it—to me, creating that time for our teams and elevating them in the eyes of our clients as true strategic partners in what they’re trying to achieve—this will really transform a lot.”
AI in healthcare marketing isn’t about silver bullets. It’s about solving real problems, one use case at a time. With the right data, a test-and-learn mindset, and a human-centered approach, the most exciting innovations are already underway.
Curious for more insights from AdLab 2025? Click here.