Every content platform now describes itself as AI-powered. Customers running similar models on similar architectures get very different results — and the variable is almost always what was in the knowledge base before the model touched it. 

In this episode of Pharma Talks, Nataliya Andreychuk spoke with Prasant Vijaykumar, Head of Strategy at Viseven, about the parts of the content supply chain that AI doesn’t solve. 

Fifteen Years, Same Strategy 

Looking back across the therapeutic areas and products he’s worked on, Prasant’s observation is that the underlying approach hasn’t moved. 

Marketing strategy is still built around the product and its content. It starts with understanding patient needs and unmet needs, and the unmet needs of HCPs. From there it becomes a question of communicating the product’s properties in a way that addresses both. The structure of how value propositions get built through content has held. 

Why MLR Is Still the Bottleneck 

AI is being used to build content and it does save time. The blockers haven’t moved with it. 

MLR review remains the slowest stage. When Prasant raises speed with MLR teams, the answer is consistent: verification takes time because every claim has to be substantiated, and review runs longer when there are problems with quality, spelling, or references that aren’t properly attached. 

AI can help detect issues and learn from previously approved content. It isn’t a solution to the stage itself, because contextual judgment in a regulated environment doesn’t delegate. 

Nataliya’s framing of the surrounding noise: the industry currently holds unrealistic expectations that AI will resolve long-standing problems. The technology will keep improving. What today requires is a closer look at the stages in between, where the delays actually accumulate. 

Where AI Does Help 

Briefing. When a brand manager needs new content, the strategy has to reach the agency as a detailed brief — which many managers don’t have time to write. AI can produce the initial draft, leaving the manager to review and correct rather than start from nothing. 

Creation, via reuse. Core messages around a product rarely change. Efficacy, safety, ease of use — the key messages stay consistent. Modularizing those components into a structured library makes them reusable rather than rewritten. 

The Argument for Modular Content 

Prasant’s position on generation is direct. AI can create new content, and in a regulated industry that carries more risk, not less. “It’s not going to solve the MLR issue.” 

Faster generation still produces material that needs full review. Brand-new content is often riskier than reusing what’s already approved. 

Structured modular content has already cleared review. That makes it more reliable and easier to reuse — and it carries a second property that matters more than it first appears. Stored in a properly managed asset system and tagged correctly, modules are traceable. When an AI tool assembles material from them, it can reference the source of each component. 

Which is where Nataliya’s summary lands: modular content is your structured data. 

The Infrastructure Question 

Engagement demand keeps rising — more channels, more personalization, more conversation happening online including through digital opinion leaders. That raises content volume, which raises a question most organizations answer badly: is there infrastructure capable of supporting compliant production at that scale? 

Many systems remain rigid. Change suppliers or adopt new tools and teams end up redesigning templates and rebuilding workflows from scratch. 

We talk about AI being able to create the content, but in the end, those pieces of content need to be in a certain format. 

With integration options and publishing channels multiplying, the requirement is a system that lets you create once and publish anywhere. 

Nataliya’s framing for how to think about it is logistical. Define every step, know where meaningful components are stored, map how they connect — because AI requires that structure as much as any human process does. 

Human review doesn’t disappear in that picture. MLR still needs a person in the loop, and as volume grows those teams get overwhelmed without structure underneath them. 

The Missing Link: Web 

Nataliya raised the gap that shows up in most real supply chains. Website content is managed separately from everything else, almost in a parallel universe — despite web pages being among the most complex content there is, combining multiple formats, components, and layers of information. 

Prasant’s read is that the role of pharma websites has already shifted. Separate product sites drawing minimal traffic are giving way to centralized HCP portals where professionals access information across multiple products in one place. 

That’s where headless CMS platforms and content authoring meet. Modular components built for an e-detailer can serve the web portal too. Connecting authoring systems to headless CMS lets the same information publish across channels without being recreated. 

When complex elements like web pages and e-detail aids sit outside the modular structure, teams face a retrofitting problem — breaking already-created content into reusable pieces after the fact. Possible, but harder than including web in the structure from the start. 

Velocity Isn’t Only Speed 

The word usually gets read as production speed. Nataliya’s addition is the dimension that matters more for global organizations: “This is the other angle of that. How fast can I localize?” 

Picture an ecosystem covering emails, e-detailers, print, and social — with web excluded. That fragmentation lands hardest on the global teams who have to adapt everything for different markets. 

Her point isn’t about consolidating vendors. It’s about whether the chain works as a chain. 

Prasant treats localization as one of the most critical issues in the AI era. Organizations that can’t manage it well will struggle to adopt AI-powered solutions at all. And despite the pace of technological change, the underlying challenges haven’t shifted much — processes and people remain central to adoption, as they were ten years ago. 

Connections Over Tools 

Nataliya’s core message: the content supply chain isn’t defined by how many vendors or tools an organization runs. It’s defined by how they connect and function as a system. 

Poorly connected systems generate conflicts between formats, tools, and workflows — and those conflicts show up directly in how quickly a company can engage its audience. 

Improving It Incrementally 

None of this happens at once. Capability unlocks as infrastructure improves, and the improvements can be sequenced: modularization, templatization, localization, or changes to the MLR review process itself. 

The example Prasant gives is moving MLR review earlier, before the final asset is produced. “If we just remove the blockers down the road, we can be accelerated 50%.”