Pfizer has developed its own generative AI platform, named after the pharmaceutical giant's founder.
Since last year, Pfizer has been developing a new AI platform to support its content supply chain and overhaul its entire internal marketing workbench. Charlie, named after Pfizer co-founder Charles Pfizer, is currently being rolled out across the organization. Executives say it's still early, but the platform is currently being used by hundreds of people on Pfizer's central marketing team and thousands of people across the company's various brands. It is also used by agency partners such as Publicis Groupe and his IPG.
Bill Worple, Pfizer's vice president of customer engagement platforms and technology, said Charlie's main focus will be improving the company's content supply chain. Charlie not only helps with content creation and editing, but also fact-checks and legal reviews. This is especially important in highly regulated industries such as pharmaceuticals. Using his system of “red, yellow, green” risk when labeling content, Charlie can identify assets that medical review teams may want to spend more time investigating. For example, a heading that is used many times may not require as much attention. Other creatives whose assets may have used previously approved languages but are now displayed in the new settings, while other content making new claims will take the most time. You may.
“The whole idea is how do we triple the [or] We increased our content creation five times to really create messages that resonate with both providers and patients,” said Worple.
Another focus is turning Charlie into a workbench for the entire marketing organization. This includes media analysis of the company's brand, insights about different competitors, and the integration of data from different websites. Charlie also integrates with Adobe platforms such as Workfront and Experience Manager, allowing users to take action based on insights across various dashboards. Other features include integrated platforms like Slack that allow employees to communicate with each other and share information.
When it comes to the type of content Charlie helps create, Pfizer is starting with digital media, emails, and digital presentations for its sales team to use with physicians. Another area the company is considering is assisting with research and drafting of medical papers. Charlie is also able to gather insights across treatment areas to better understand his clients and treatments. For example, Worple mentioned how migraines can affect parents differently than those without family members.
“You actually start generating different insights into who your customers are,” Worple said. “after that [knowing] What are the real issues for them? It is not classified as medical research. The insight “X percent of these people are this type of individual” is very helpful in understanding who your customers are. Now you know how to talk to them better. ”
marcel spinoff
According to Worple, Charlie was trained based on data from a variety of sources. To generate marketing content, Pfizer uses training data from approved content registered for each therapeutic category (oncology, endocrinology, etc.) and specific products. We also use a segmentation model to teach Charlie about the types of messages associated with each segment and which are most important.
Although a custom version of ChatGPT is utilized to create the content, Charlie uses a recommendation algorithm. Pfizer is also developing ways to use natural language processing (NLP) for queries related to internal research, case studies, and performance marketing data. Answers will also be verified with source material based on previously published and verified Pfizer content to prevent Charlie from giving inaccurate information or other “hallucinations”. There is also a process for reviewing and validating the output.
Pfizer's effort is a collaboration with Publicis Groupe, which helped build Charlie on the existing Marcel AI platform. Arpit Jain, president of Marcel, said Marcel is like a “base house” and Charlie is a customized version tailored to Pfizer's needs. He added that Publicis Groupe also has a dedicated team that helps maintain and evolve Charlie on a daily basis. Last month, Publicis announced major updates to its proprietary AI, including a new platform called CoreAI.
Because medical data is highly sensitive, Pfizer has ensured that the collection and use of Charlie's data meets various internal and external privacy requirements. Based on who is using Charlie, the platform can tailor its features to the employee's role, how they use it, and the type of data they engage with. This makes data governance particularly important, both from an accuracy and privacy perspective.
“We have to be very careful,” Jayne said. “The last thing you want is to be left with weird data that you didn't clean up properly. Especially in the medical field, that can be a matter of life or death. If you recommend the wrong iPhone, that's okay. It's the world's best. It's not the end. But God forbid that should happen if you recommend the wrong medicine.”
Along with content creation, using generative AI for medical, legal, and regulatory reviews is a key capability that other pharmaceutical companies are also considering, according to analysts covering the space. It is also considered to be low cost, low risk, but high reward.
According to Gartner analyst Chris Beland, marketers are excited about leveraging generative AI to help them globally deploy AI-translated and localized content for each market.
“The potential for cost optimization is huge, but it's not without risk,” Beland said. “There are clear implications for consumers and users. [health care providers] Not only will you see flawed or inauthentic content, but if not planned properly, it can also have unintended impacts on strategy, measurement, and approval workflows. ”
generation of advertisements, etc.
Beyond marketing, generative AI could create between $60 billion and $110 billion of annual economic value for the pharmaceutical and medical products industry, according to a January report from the McKinsey Global Institute. The report also states that using generative AI in marketing could cut content creation costs in half and increase production pipelines by 20%. This technology also has the potential to double or triple the speed of content approval.
According to WARC, pharmaceutical and healthcare companies represent the third fastest-growing product segment in global advertising spending after financial services and technology/electronics. In the U.S. alone, the company projects pharmaceutical and healthcare ad spending to reach $34.6 billion in 2024, an increase of 16.7% from 2023.
Some companies across industries are exploring ways to generate synthetic data based on their own data, said Paul Stringer, WARC's managing editor of research and insights. By doing so, businesses can create an audience that looks like their customers without exposing personal information. However, he said it's important that companies make sure they don't reproduce real-world biases in the process.
“If you don't feed these generative AI models the right training data, you run a huge risk of amplifying existing biases and gaps in your customer data,” Stringer said.
Other agencies are also building their own generative AI tools for pharmaceutical clients. One is CMI Media Group, a media agency owned by WPP. (A few weeks ago, WPP announced he would invest £250 million in 2024 to build out AI initiatives.) According to Justin Freid, CMI's chief media and innovation officer: The tool is scheduled to debut later this month and is already being tested on social and social media. display advertising. I plan to include a video later.
In addition to assisting with legal reviews, CMI uses this tool to help generate content that is personalized to each audience or optimized based on platform performance data. For example, this might include changing the background of your ad based on the season. All variations remain within set creative guidelines, but will open up “countless testing opportunities” to see what works for each audience, Freed said.
“They still require approval, but these are creative variations and not like completely new campaigns,” Fried said. “You don't have to dissect that new ad… there's nothing crazy about a new tagline with a new claim or something like that. We're keeping it simple, but more relevant to the end user. We found that it was more effective.”
