Within consumer health, most of the conversation about artificial intelligence has so far been about innovation – how intelligent digital technologies can support consumers to self-care more responsibly and effectively, and how the data generated from this activity can be used to personalize the next generation of OTC products.
Bayer Consumer Health, for example, has created a “precision health” division focused purely on leveraging digitalization. “Precision health is making self-care even more consumer-centric,” commented Bayer CH president Julio Triana recently. “New digital tools are enabling offerings to be more tailor made.” (Also see "Bayer Offers US Consumers Precision Wellness With ‘One A Day’ App And Test Combo" - HBW Insight, 5 September, 2024.)
But what about how AI is being used to drive internal efficiency? After all, AI offers exciting possibilities for streamlining consumer health industry operations, harnessing the creativity that exists within OTC companies and, of course, improving innovation processes to support new product development.
Here Bayer CH is also leading the way. According to the firm’s head of digital transformation and IT, Cristina Nitulescu, AI is being used internally in a variety of ways, for example to save time on routine tasks and to manage supply and demand.
With regards to the latter, Bayer CH has created a “cross-functional planning and steering product” called “Foresight,” Nitulescu told HBW Insight in an exclusive interview, which uses machine learning technology to interpret demand patterns in the regions and markets it operates in.
Currently used in roughly 20 countries, Foresight responds rapidly to these signals with “sellout forecasts,” which Nitulescu said have proven to be “very accurate,” higher than 90% accuracy in some cases.
“It’s really improving our predictive business steering,” Nitulescu said. “It’s allowing us to really understand where the demands are and what our customers require, and either produce or move the inventory into that space, reducing lead time to market and getting our products in the hands of our consumers at the right time.”
In future, Bayer wants to “rapidly scale up” this model within its key markets, seeing the system as providing a “crucial advantage” for its consumer health business, she added.
Machine Learning
Supporting Foresight is a chat function for supply managers that enables them to report any supply disruptions so that the company can deal with them quickly.
“The intelligent chat functionality provides automated responses to any questions they may have related to supply incidents and interruptions, proposing mitigation and recovery measures that recover products in the fastest manner possible,” Nitulescu explained.
“It’s not yet scaled up, but shows very promising results, and now we are actually providing more interactive experiences for our supply managers.”
Another aspect of supply and demand to which Bayer is applying machine learning is in marketing, for example using trend data to assess what kinds of advertising investments are most effective and worthwhile.
“It helps reduce costs,” Nitulescu said. “But it’s also giving us better results than the traditional segmentation that we typically see in the space.”
MyGenAssist
Cost saving is the aim of Bayer’s other main application of AI. Inspired by large language models like ChatGPT, Bayer CH has created its own generative AI system, which it calls “MyGenAssist”
“MyGenAssist is Bayer’s in-house secure gen-AI platform that can now be accessed by all Bayer CH employees,” Nitulescu explains. “We have about 25% active users right now.”
Uses range from creating job descriptions to automating the innovation process for new consumer health products. “It doesn’t mean that MyGenAssist creates the entire job description, but it gives you a very good, accelerated start, so you can really focus on what matters,” she noted.
“And with regards to product development, the aim here is to have shorter and lower cost with regards to things like pre-due-diligence, and increasing the speed and quality of innovation,” she added.
Time Saver
On average, MyGenAssist users are already reporting about two to four hours of personal efficiency on a weekly basis, Nitulescu revealed.
“I think people are excited about the immediate efficiencies that they can see in their day-to-day lives, and with tasks that maybe are not the core of their business,” Nitulescu said.
“There’s still a lot of work to do in educating our users on the usage of this platform, for example with what’s now called ‘prompt engineering,’” she continued. “Learning how to ask the right questions and understand the context are probably the biggest educational efforts we see right now.”
Security Matters
As for why Bayer decided to create its own gen-AI system, Nitulescu said “data security.”
“The gen-AI models that have been released are public models, trained on public data,” Nitulescu explained. “I think one of the key tenets for using AI this way and at scale is gaining trust in the tools and the language models.”
In the end, gen-AI is still very new, Nitulescu pointed out. “I think we’re just scratching the surface at this moment. The reality is, we and the rest of the world are still learning about what we can do and how we can train these models.”
“It’s a learning journey,” she concluded. “For me, personally, it’s an exciting journey. I think the possibilities are limitless.”