Though a relative newcomer to the Davide Deiana, Head of Content Excellence at BASE Life Science, has spent the past few years developing Gen AI solutions designed to integrate with Veeva Vault Promomats. With his team, Deiana is trying to help pharma marketing and medical, legal, and regulatory (MLR) teams boost the efficiency of the content production and review cycle.
In this interview, Deiana reveals how Gen AI is reshaping the pharma content landscape, the unique challenges and opportunities that come with this change—particularly in terms of pre- and post-MLR review—and what companies need to successfully adopt Gen AI tools in their workflows.
What do you do at BASE?
DD: I'm the Head of Content Excellence at BASE, and I’m leading the advisory around content excellence, particularly with the digital asset management (DAM) life cycle in the pharma industry. My team delivers consulting projects to pharma companies and helps them with system integration, especially on the Veeva Vault Promomats platform.
AI has always been the focus of our attention. We’ve been developing our Gen AI capabilities for a while now—they’ve naturally evolved over the years as we’ve played around with Gen AI and AI in general. We’ve combined our knowledge of business processes in pharma, our knowledge of the Veeva platform, and our AI knowledge to develop Gen AI solutions that can be integrated into Veeva.
How are you implementing Gen AI in pharma marketing?
DD: We know that many other Gen AI providers are focusing on the content creation side. We don’t focus on this because we're not a creative agency. Instead, we’re experts in MLR review and approval. Our strength has always been around content compliance, so we’re tackling the challenges of getting content through the pre MLR and MLR review processes.
Reusing content across different languages is always a challenge, so we support companies with this by leveraging Gen AI to produce in-flight translations of documents. We also support the localisation process, which is very important for companies trying to implement a global to local and local to local process.
How can Gen AI help companies through the MLR process?
DD: We’re running a lot of different pilot programs and proof of concept programs with our customers. We’re trying to make sure users are supported by Gen AI to run reviews, especially in the pre MLR phase. This is to make sure that—as a brand manager and content creator—you can send high quality content to the MLR teams for review.
You want to avoid hitting the MLR team with all these pieces of content that aren’t the right quality yet—this wastes a lot of time
One of the pain points here is that the MLR process involves multiple stakeholders that run quality checks on content. These aren’t only about linguistics or grammar, but also about adherence to claims and adherence to scientific statements. With content that can be 50 pages long, all of this is extremely demanding and time consuming.
This is the perfect point where Gen AI can help brand managers and MLR teams work faster
You want to avoid hitting the MLR team with all these pieces of content that aren’t the right quality yet—this wastes a lot of time. Otherwise, you're involving an extremely expensive MLR team that’s also scarce in resources, and going back and forth because your content is not at the right quality. This directly impacts your costs and time to market.
Companies need to produce higher volumes of content to provide the right level of personalisation for HCPs, so this is the perfect point where Gen AI can help brand managers and MLR teams to work faster.
After doing lots of analysis, we also found that a remarkable percentage of the approved content out there has spelling mistakes—somewhere between 80% and 90%.
Between 80-90% of approved, distributed content has spelling mistakes
This content goes through so many hands. It goes from authoring tools and agencies to quality control teams, coordinators, reviewers, and approvers. In some countries, it’s even submitted to the health authorities too.
So imagine the frustration if a document took 50 days to be approved—went to the health authorities and came back— and you can’t change it anymore. If your content has spelling mistakes, and you can’t change it unless you get it reapproved—this would drive a lot of people nuts!
What do pharma companies struggle with when trying to adopt Gen AI?
DD: Like with all new technology, it’s important to set your expectations at the start. Gen AI is not one-click magic. You can’t just configure it once and then expect it to always work.
Gen AI is not one-click magic
This could be a point of frustration for brand managers and so on. They might think that Gen AI can perform a review in one click, for example. They need to understand what Gen AI can actually provide. Because, after all, there’s not so much awareness out there in terms of what Gen AI’s capabilities are.
I know that AI is the new buzzword, but it's more than that. In terms of technology, we’re taking an incredible step forward with Gen AI. But distributing this knowledge across all teams takes a little bit more time. Right now, Gen AI advancement is faster than what the industry can absorb.
Right now, Gen AI advancement is faster than what the industry can absorb
That is why we’ve taken a specific design principle for AI. We know the process and we know that when we talk about pre MLR and MLR, you’ll never create something deeply meaningful for pharma companies from a generic solution.
If your solutions aren’t tailored to the customer, you’ll struggle to get the refined, specific feedback from the AI models that you need. The feedback has to be precise and focused on your company's goals for it to be truly valuable.
You will never create something deeply meaningful for pharma companies from a generic solution
That's why our solution is the same at its core. But the AI models, the prompting, and the business rules behind it all need to be refined and tailored for the specific use cases of each customer. Even though the MLR process and tasks are similar across the industry, they still need to be customised for that particular customer.
Understanding the content and providing feedback on it—that's the part of Gen AI that’s specific to each company
The MLR process doesn't really change between companies. But the way you perform the MLR review does. It changes based on the audience, the product, and the brand. Understanding the content and providing feedback on it—that's the part of Gen AI that’s specific to each company.
Do you think AI would ever be able to take the role of an MLR reviewer?
DD: That’s a very interesting question, because right now, no—we don't believe that AI could replace an MLR reviewer.
This is because the MLR review is a compliance process. The regulations ensure that responsibility and accountability are clearly assigned to people, not machines.
You’d need to change how the regulatory system ensures that the promotion of medicinal products is reviewed and conducted in a specific way. If the regulations don’t change, the accountability remains on humans—on specific individuals assigned to these roles within the pharmaceutical company in each country. Naturally, if you have legal accountability and responsibility, you’ll always want to have the final say.
Gen AI can do the dishes for you so that you can focus on the creative side of things—the real subject matter
If you’re the information officer in Germany, or if you’re the responsible pharmacist in France, you shouldn't be spending time making sure there are no spelling mistakes, that the trademarks are positioned correctly next to the name of the brand, that there are links to references, and so on. These things should not be part of the core MLR team.
This is where Gen AI can provide support. Gen AI can do the dishes for you so that you can focus on the creative side of things—the real subject matter. But substituting the reviewer and the approver? That would be very difficult right now.
What advice would you give to companies trying to embrace Gen AI tools for the first time?
DD: This is something we ask every time that a customer wants to know about Gen AI tools: Are you absolutely clear on what the business challenges are in your process? Companies and customers have different priorities and challenges. Maybe it's in their content production process; maybe they don’t have enough budget; maybe their MLR process is taking a very long time. There are so many potential challenges. And then they want us to provide tools such as content authoring tools to help them easily create and reuse content.
MLR teams are mostly concerned by the fact that the quality of content is not high enough
But here, we have to ask: What is the role of Gen AI in providing aid for that challenge? Have you really assessed the exact pain points—the data points too? It’s not only about looking at how you conduct the process, but really listening to the different user groups.
This was our surprise when we started analysing all of this back in the day, in terms of where to focus with Gen AI—MLR teams are mostly concerned by the fact that the quality of content is not high enough when it reaches them.
It wasn’t that the content subjects or review processes were too complex—they have the knowledge and are perfectly fine and efficient in handling these. The burning issue was that quality is too low. So how do you improve the quality of material before it reaches the MLR approval process?
Really assess the pain points in your process—you may be surprised
That’s where we focused, developing Gen AI to improve tone of voice and language, fast translation, grammar, in doing trademarks checks and code compliance checks. If you have a set of rules that you can pre assess your content towards, then it is there that you need to focus and that's where Gen AI should focus. Really assess the pain points in your process—you may be surprised.
.png)
.png)
