Most pharmaceutical organisations don’t have a content creation problem. They have a content operations problem — and increasingly, AI is exposing exactly where that operational gap sits. In this interview, Jindrich Borovsky explains why MLR bottlenecks, low content reuse, and stalled modular content programmes are rarely technology problems at their root, what has to be true before AI can work reliably in a regulated content environment, and what changes for content teams over the next three years.
Key takeaways
- Most pharma organisations’ content problem is operational, not technological — approved content already exists in large volumes; the challenge is making it reusable, governable, and scalable.
- Low content reuse is not simply a search or discoverability problem — it’s a combination of user behaviour, content tagging, and content relevance.
- MLR bottlenecks are usually a symptom of how content is structured and governed upstream, not a shortage of review capacity downstream.
- AI amplifies operational maturity rather than replacing it — organisations need standardised taxonomy, tagging, and governance before AI can produce compliant output reliably.
- Modular content programmes fail most often when the operating model and agency incentive structures aren’t redesigned alongside the content strategy.
- As AI takes on execution work, the human role shifts from producing content to directing it — setting standards and validating output before it reaches MLR.
- Global and local teams move from a one-way handoff toward a two-way feedback loop, with local input coming earlier in the process.
Why pharma’s content problem isn’t what most organisations think
You moved from Microsoft into pharma at MSD (Merck). What surprised you about how pharma content actually operates?
Before entering pharma, I was a solution delivery lead for Global Marketing Operations at Microsoft, which was already highly digitally transformed. Coming from that environment into MSD, pharma was definitely a bit behind. What surprised me most was how decentralised everything was: legacy solutions, no global standards, completely fragmented systems and processes. While it was a bit of a shock initially, I immediately saw it as a massive opportunity to build something from the ground up and make a real impact.
What made you decide this was the space you wanted to focus your career on?
“The sheer complexity of the content lifecycle in pharma forces you to think hard and solve real puzzles. I love that challenge.”
I fell in love with the content space immediately and never wanted to leave, even though I watched many people around me rotate out into other areas. There’s a massive appetite for change and innovation in this space, which keeps things exciting. And the sheer complexity of the content lifecycle in pharma forces you to think hard and solve real puzzles. I love that challenge, and it’s exactly why I’ve dedicated my career to it.
What’s the biggest misconception organisations have about their own content problem?
Most organisations think their problem is the technology, a lack of headcount, or that they’re moving too slowly on AI. But in reality, the true bottleneck is change management, lack of governance and user adoption.
In my career, I’ve seen hundreds of thousands of dollars spent on cutting-edge solutions that ended up with only five active users. Why? Because there was no executive sponsorship or investment in driving the cultural change. We simply bought or developed a tool, threw it at the users, and expected them to magically generate the ROI on their own. Tech is the easy part. Driving the adoption is the real challenge.
Is low content reuse really just a search problem?
“Low content reuse is not purely a search problem. It’s a combination of user behaviour, how content is tagged, and how relevant the content actually is — fixing the search bar alone rarely solves it.”
No. There’s one belief I push back on constantly: that low content reuse is purely a search problem. Everyone assumes that if reuse rates are low, it’s because discoverability is poor or the search engine is broken. But if you actually look at the data — how users search, the keywords they enter, how they apply filters — you quickly realise content reuse is a multi-layered issue. It’s not just search technology. It’s user behaviour, content tagging, and the actual relevance of the content being created.
What’s the thing teams are most reluctant to hear at the start of an engagement?
An outsider coming in claiming to have all the answers and a cookie-cutter plan to “fix” everything. No one wants that. For an engagement to actually work, you have to do the opposite: listen, observe, and include everyone in the process. It’s about building trust, not creating friction.
Related: a practical guide for reducing MLR review time.
Why MLR stays slow, even as pharma invests in review technology
Why does MLR remain such a persistent bottleneck in pharma?
“MLR bottlenecks are usually a symptom of how content was structured and governed upstream, not simply a shortage of review capacity. Reducing review pressure means rethinking how content is created from the start.”
Most pharma organisations focus their MLR efforts downstream, buying new review platforms or throwing AI at the final approval stage to clear the traffic jam. That falls short because it’s downstream policing, treating the symptom instead of the cause. The real bottleneck is an upstream quality gap: creators don’t have embedded compliance guardrails, so bad content enters the pipeline and just gets rejected faster.
What upstream issue rarely gets enough attention?
What happens at the creation stage itself. Brand and local regulatory rules should be embedded directly into the tools people create with, cutting out avoidable errors and endless back-and-forth. But even when standards exist, creators often aren’t aware of them, so you have to back the technology with continuous coaching on local SOPs, for internal teams and agencies alike. And it needs to be effortless to find and adapt pre-approved global or regional assets — that takes a huge amount of pressure off the final review stage.
Related: how pre-check tools catch compliance issues before content is submitted.
Have you seen an organisation genuinely fix MLR? What did they do differently?
Yes, and the pattern is always a combination, never a single fix: review-ready checklists so creators catch mistakes before submitting, tracking the actual reasons content gets rejected so the systemic issues get fixed rather than repeated, and AI-driven tools that automate basic checks and give creators instant feedback. It’s the combination of process, data and automation that moves the needle.
See how one pharma team applied this in practice: how Idorsia cut content production time from weeks to days.
Is pharma actually ready for AI, or does it just think it is?
Are pharma organisations actually ready for AI, or do they just think they are?
The biggest gap isn’t a lack of interest or budget, it’s the difference between pilot enthusiasm and scalable execution. Most commercial pharma organisations think they’re advanced because they have active AI pilots running. In reality, they’re just layering modern AI onto legacy infrastructure. Closing that gap means rethinking the entire content supply chain, not just plugging an AI tool into an existing workflow.
What needs to be true about a content ecosystem before AI can deliver real value in a regulated environment?
“AI does not fix a broken content process, it scales it faster. Before AI can deliver compliant, reliable output in a regulated environment, organisations need a standardised taxonomy, consistent tagging, and governed source content.”
You have to fix the foundation first — you can’t automate chaos. Standardisation is non-negotiable: clean inputs, a unified global taxonomy, consistent tagging, consolidated channel requirements without market or brand deviations. If the inputs aren’t structured, AI can’t generate meaningful, compliant outputs.
Related: how a structured MLR knowledge layer prepares content for AI.
What’s the risk of moving too fast on AI without the right foundations in place?
AI doesn’t fix a broken content process, it just scales it faster. Without strong metadata models, clear taxonomies and standardised workflows, moving fast on AI creates an overwhelming MLR bottleneck, breaks content traceability, and results in expensive tools that can’t scale across markets.
Why modular content programmes stall — and what “good” looks like in three years
What’s the biggest mistake pharma teams make with modular content?
Rolling out a modular content strategy without reimagining the underlying operating model and agency workflows. On paper, breaking assets into pre-approved components promises speed, compliance and efficiency. In practice, if you don’t redesign how teams actually author and approve content, modularity just adds overhead — and it falls apart even faster without end-to-end traceability of where modules are used and how they perform. Truly mature modular implementations are still rare in pharma for exactly this reason.
Related: why modular content programmes stall between pilot and scale.
Is there a common way of thinking about content transformation you’d challenge?
The belief that technology replacement is the primary engine of content transformation.
Real transformation in pharma is about 80% operating model, MLR risk tolerance, and how agencies are incentivised. If creative agencies are still billing time-and-materials, they have zero financial incentive to adopt modular, automated authoring.
What wastes the most time or budget in pharma content operations?
Treating local agencies as primary creators rather than adaptors, constantly reinventing the wheel.
It happens for three reasons: approved global content is hard to find and adapt; there’s poor governance over agency deliverables and tech stacks; and MLR gets used as a late-stage review instead of an early filter.
Fix content discoverability, incentivise reuse, and enforce strict asset ownership with agencies, and you save a substantial amount of budget almost immediately.
What’s the most visible change for teams doing the day-to-day work, if an organisation gets content operations right?
“Humans move from being content builders to content directors.”
A shift toward an AI-augmented operating model across the whole content lifecycle. AI isn’t replacing human expertise, it changes what daily work looks like. Humans move from being content builders to content directors: setting up the standards, taxonomies and design systems that make content LLM-readable, and then validating AI output for accuracy, compliance and brand alignment before anything goes to MLR.
Related: how Shaman works with teams on content, technology, and people.
How does the relationship between global and local teams change?
It creates a much tighter, two-way feedback loop, replacing the traditional top-down approach. Global becomes a stronger content engine, producing master templates and modular components that local teams can adapt in a fraction of the time. And local teams get a seat at the table much earlier, brought into planning rather than just receiving finished assets that don’t fit their market.
What becomes possible in three years that isn’t possible today?
We move from static, manual content cycles to a continuous, predictive content supply chain, where local adaptation happens instantly, performance data suggests content tweaks directly, and MLR shifts from a bottleneck to an automated pre-clearance process.
Why Jindrich joined Shaman
What made you want to join Shaman specifically?
I’d been following Shaman for a while, and I’ve known Maurice and Erik for several years. Every conversation proved they’re true experts in this field, with a vision that aligns with my own. What sets Shaman apart, and what you don’t often see in this space, is that we don’t force a one-size-fits-all solution or oversell. We listen, respect the client’s existing ecosystem, and tailor the strategy accordingly. I’m also genuinely proud of our focus on driving adoption and proactive user support — high adoption is notoriously hard in enterprise pharma software, and that’s exactly where I plan to focus my energy.
What are you genuinely excited to work on, and what impact do you want to have?
The breadth: seeing the diversity of tech stacks and operating models across pharma organisations of every size, from emerging biotechs to global enterprises, and helping solve their operational challenges directly. I want us to be recognised as genuine industry leaders in content operations, sought after by clients and respected by peers, while using our hands-on expertise to keep pushing solutions that solve the fundamental challenges pharma is facing today.
From the CEO
“When we decided to expand how we work with pharma teams beyond technology, we knew we needed someone who’d actually lived inside the problem, not just sold solutions to it. Jindrich has spent his career inside content operations, most recently at Microsoft’s Global Marketing Operations before moving into life sciences.”
Maurice van Leeuwen, CEO & Chief Product Officer, Shaman
Frequently asked questions
Is pharma’s content problem really about technology?
No. Most life sciences organisations have already invested heavily in authoring platforms, DAM systems, and AI tools. The persistent bottleneck is usually the operating model around that technology: governance, adoption, and how workflows connect teams — not the tools themselves.
Why is content reuse low even when organisations have large content libraries?
Low reuse is rarely a pure search or discoverability problem. It’s typically a combination of how content is tagged, how users actually search, and whether the content available is relevant enough to reuse in the first place.
Why does MLR review stay slow even with modern review technology?
Because most efficiency efforts focus downstream, on the review stage itself, when the root cause is usually upstream: content entering the review pipeline without embedded compliance guardrails, inconsistent reference management, or unclear ownership of previously approved assets.
Can AI work reliably in a regulated pharma content environment?
Only once the content foundation is in place: a consistent taxonomy, standardised tagging, and governed, structured source content. Without that foundation, AI tends to scale existing inconsistencies rather than resolve them.
Why do modular content strategies fail to scale in pharma?
Most commonly because the operating model and agency incentive structures aren’t redesigned alongside the content strategy itself — for example, agencies still billing on a time-and-materials basis have little incentive to adopt modular, reusable authoring.
Want to go deeper?
What’s Preventing Pharma Teams from Scaling Content Excellence?
These are exactly the challenges we’ll explore in our upcoming webinar, together with Jindrich and the Shaman team.
Register for the webinar →
Jindrich has nearly 20 years of experience across enterprise platforms and content operations, including 10+ years at MSD (Merck), where he led a global team of 40+ IT professionals and drove global rollouts of Veeva Vault PromoMats — and before that, Microsoft’s Global Marketing Operations. He leads Shaman’s new Content Excellence Practice. Connect on LinkedIn.
