How do you see AI ultimately benefit consumers? Do you envisage AI enabling better products, enhanced safety, personalization, affordability, etc?

AI has the potential to fundamentally improve how consumer products, particularly in health, beauty, and personal care, are developed, evaluated, and recommended. Its greatest value lies not in replacing human expertise, but in enabling better decisions grounded in data, biology, and measurable outcomes.

AI can enable better products by improving how complex biological data is analyzed and translated into formulation and recommendation decisions. Human biology is multifactorial and highly variable; AI excels at identifying patterns across large datasets that would otherwise remain invisible.

One important application is the use of clustering methodologies. Rather than relying on broad demographic segmentation or treating each individual as entirely unique, AI can identify biologically meaningful profiles based on shared characteristics within microbiome, physiological, or molecular data. These clusters capture real-world biological variability while remaining scalable, enabling products and recommendations to be designed for defined biological profiles rather than one-size-fits-all assumptions.

In parallel, scoring models help translate complex, multidimensional datasets into interpretable indicators, such as balance, resilience, sensitivity, or functional performance. When scientifically validated, these scores make biological data accessible to both professionals and consumers, while still preserving the underlying complexity. This allows personalization to evolve over time and supports decision-making based on measurable change rather than static snapshots.

AI also enables consumers to measure the real impact of products on their bodies, rather than relying solely on claims or short-term perception. Depending on the technology used, personalization can operate at different levels of depth and cost. Image-based AI analysis, such as standardized photography or selfie analysis, can track visible changes over time, while biological evaluations, including microbiome testing via swabs or biomarker assessment through blood draws, provide deeper insight into physiological response. Together, these approaches shift personalization from prediction to feedback, allowing consumers to assess what is actually working for them.

What are your top three recommendations for companies in your sector considering AI and what should they absolutely avoid?

First, companies should anchor any AI initiative in high-quality, purpose-built data and, critically, in clear ownership or long-term rights to that data. AI performance is directly tied to the relevance, depth, and representativeness of the datasets used for training. For companies intending to bring an AI model to market, proprietary datasets are often the primary source of defensibility. Without control over the data used to train and refine models, even technically strong AI systems risk becoming easily replicable.

Second, AI systems must be developed and deployed with direct involvement from domain experts, both during model training and throughout validation. Scientific, clinical, and regulatory experts are essential to define relevant inputs, interpret outputs, and continuously assess accuracy and limitations.

Third, companies must clarify early the intended commercial role of AI. Is the model meant to remain an internal tool supporting R&D and formulation decisions, or is it intended to be deployed as a consumer-facing product? For models brought to market, companies should critically assess defensibility: what differentiates the model, how easily it can be replicated, and whether the underlying data, validation framework, or continuous learning loop creates a sustainable advantage. This assessment should inform whether the level of investment is justified given the competitive landscape and the likelihood of rapid imitation.