SPEAKER:
Nataliya Andreychuk, Co-Founder & CEO, Viseven
KEY TAKEAWAYS:
• 80% of pharma believes it engages HCPs effectively; only 28% of HCPs agree
• 77% of content assets sit unused in DAMs, translating directly to unrealised revenue
• Content velocity pursued as raw speed produced the fragmentation crisis it was meant to solve
• Native AI trained on proprietary compliance data outperforms generic tools on editability and regulatory approval
• 41% of HCPs report social media influences clinical decisions, making DOL engagement strategically obligatory
"80% of us think that we really engage with healthcare professionals. Then otherwise, 28% of healthcare professionals think that we are not on the same page. We are spending 17 billion euros — approximately $20 billion, really on building content. So this is a fundamental gap and the crisis of content production we need to think about."
Andreychuk opened with those numbers at Pharma 2026, and the arithmetic is damaging enough to warrant repeating: an industry that believes it is performing well at HCP engagement is, by HCPs' own account, failing more than seventy percent of the time, while spending twenty billion dollars a year on the assets that are supposed to close the gap. More than half of all produced content is simply ignored. The deeper figure is worse. "77% of it is sitting in the digital asset management tool," she argued. "It's hardly ever used. That means whatever engagement or personalization we have strategized never happened. And this 5 to 10% of revenue growth through prescription might never happen to us."
Two data points. One structural conclusion: pharma has built a twenty-billion-dollar production system optimised for generating assets that no one uses. This is not a technology problem. It is an architecture problem, and that distinction determines whether the industry's current AI investments solve anything at all.
Velocity Became Its Own Trap
The industry's instinct, confronted with an engagement deficit, was to produce more and faster. That instinct compounded the original failure.
Andreychuk's diagnosis is precise. "Content velocity, we all think immediately, is about the speed, right? So it is about the speed. But I want us to go for the journey. When we started to think about fast production of the content, then scaling and we ended up into the fragmented channels and the problem of fragmentation of what we deliver." What she is describing is a decade-long strategic error that most organisations have not yet named as one. Each content sprint produced campaign assets calibrated for a specific channel, a specific market, a specific moment. When the campaign ended, those assets became orphaned files in a system that no one queries with confidence. The next sprint started from scratch. Year after year, the archive grew. The engagement numbers didn't.
The alternative definition Andreychuk proposes is not a slogan but a structural requirement: velocity must be understood as "speed multiplied by efficiency and scaling , so qualified scaling, where we can really get to the quality." The word "qualified" is carrying the argument. Scaling unstructured content production is not a strategy; it is an acceleration of the underlying dysfunction. Qualified scaling requires content treated as reusable data, modular components with defined relationships, approved claims that can be assembled rather than rewritten, localisation logic built into the asset rather than applied manually at the end of the process.
Without that foundation, digital asset management systems are not content libraries. They are content graveyards with sophisticated filing systems. Andreychuk's characterisation is direct: "The digital asset management tool where you cannot find what you will be using tomorrow is like a graveyard for the content."
Companies currently deploying AI content tools against existing production workflows are not solving the graveyard problem. They are filling it faster. AI-powered production at scale, layered onto fragmented and unstructured content infrastructure, does not change the fundamental economics. It multiplies them. The waste becomes industrial.
The Infrastructure Bet Most Companies Aren't Making
"It will not be generic AI anymore. It will be your native AI if you will be having these fundamental basis prepared."
That sentence is the session's most consequential claim, and it functions as a vendor-evaluation framework. The "fundamental basis" Andreychuk references has operational specificity: approved claims libraries, MLR decision history, localisation rules, structured content modules, all connected in a single query-able system that AI can actually learn from. Generic AI tools can produce content quickly. Native AI trained on a company's own compliance and content history can produce content that is approvable, localised, and structurally consistent with what the medical-legal-regulatory function has already accepted. That distinction determines whether AI compresses the content-to-approval cycle or extends it.
The editability problem is less discussed and more consequential than most organisations recognise. "Today we do not have autonomous AI," Andreychuk cautioned. "So that's why it is very important that whatever was generated is compliant and editable by human. Because today we might have tons of content generated by AI in the right format, it must be HTML5. But can the human afterwards edit the content?" Most generic AI tools produce outputs that are visually polished but structurally locked, formatted for display rather than for modification by content operations teams working in regulated environments. The compliance bottleneck AI was supposed to eliminate reappears at exactly the point the asset needs human review.
The pre-MLR application Andreychuk described shifts that dynamic. "Before the human sees the ready to approve content, the AI is helping us pre-MLR it. So what does it mean? We can actually catch up on the local regulations, we can catch up on the grammar, on the connectivity of references and the claims." This is AI functioning as a regulatory accelerant, catching local compliance gaps and reference-claim mismatches before they reach the review queue. The human reviewer sees a cleaner submission. The approval process moves faster.
The infrastructure logic extends downstream into field execution. Andreychuk described an agentic AI implementation in which content modules power not only production but rep training: AI-simulated HCP personas, built from approved content profiles, allow field representatives to practice conversations before live visits. The content infrastructure becomes the training infrastructure. The same approved claims that govern the asset govern the rehearsal.
For companies that have already invested in generic AI content tools, this picture is uncomfortable. They are not facing a simple upgrade path. They face an infrastructure retrofit, rebuilding the modular content architecture that should have preceded the AI investment. The competitive consequence compounds over time: organisations that build proprietary content infrastructure first will improve their native AI with every content cycle, while those that bolt AI onto fragmented systems will improve their production volume without improving their engagement outcomes.
Discover more on this topic at Pharma Commercial Data & Tech Europe 2026 (4-5 November, London) Europe’s collaborative home for data and tech pioneers. Visit the website here.
The Channel Pharma Can No Longer Ignore
The structural reforms Andreychuk advocates address the production side of the engagement crisis. The distribution side has its own unresolved problem, sitting in a channel most pharma organisations have treated as either a compliance risk or a marketing experiment.
HCPs are already on social media. The behavioural data Andreychuk cited quantifies what medical affairs and commercial leaders have been observing qualitatively: "65% to 77% of healthcare professionals, they use social media, they enjoy sharing, they enjoy education online, they're listening to their peers and to their colleagues and voilà, they even say that they make some of their decisions based on the learnings they are finding in social media. So that's 41%." When nearly half of HCPs report that social media content influences clinical decision-making, the channel is no longer a peripheral communications question. It is a clinical information environment.
"We are not only finding new engagement, we are also fighting misinformation and we are fighting online for trustful information," Andreychuk argued. "So that means like whoever needs to be treated and there are serious cases like we have during pandemics, like we have with vaccination, right? People are not listening to these sources which are non-reliable. So this is our obligation to work with digital opinion leaders."
The reframe is more demanding than it appears. If 41% of HCPs are influenced by online peer content, then the absence of pharma-supported, compliant, expert-voiced content in that space is not a missed marketing opportunity. It is a ceded information battlefield. In therapeutic areas with active misinformation, vaccination, oncology, rare disease, companies that leave online medical discourse to unvetted sources take on a reputational and strategic risk that has yet to be fully priced by the industry. The regulatory and reputational dimension of that choice has not yet been fully priced by the industry.
The operating model Andreychuk prescribes for digital opinion leader engagement inverts pharma's default relationship with external voices: "It is not a commercial relationship, this is a partnership. Then the second one, it is co-creation in the strategy, in the content, in the narratives. The next is collaboration. So partnership should be collaborative. Then we must deliver value to them as well." The shift from briefing to co-creation requires a cultural adjustment as much as a channel strategy. A DOL who has shaped the content is a DOL with genuine conviction about it. That authenticity is, in the peer-influence dynamic that makes the channel work, the entire point.
The first-mover advantage in this space is not primarily about reach. It is about trust infrastructure. The company that establishes credible, compliant expert relationships in a therapeutic area's online discourse will shape the information environment that competitors must then operate within.
The Prerequisite Everyone Will Defer
Andreychuk's evidence, assembled across the session, points toward a conclusion more uncomfortable than any she states directly. Most pharma companies currently cannot demonstrate whether their content strategy produces any measurable commercial return.
The measurement problem runs through every layer of the argument. Channel activity and content volume are poor proxies for engagement. As Andreychuk put it, "the channel launch is not the activity yet, the clicks are not really guaranteeing that this is the engagement. The channels could be siloed and we cannot even understand that the volume of the content is not a guarantee of the engagement at all. And there is no link to the prescription behavior." The standard KPIs the industry relies on; impressions, clicks, open rates and content produced, measure production and distribution, not impact. They cannot detect whether a piece of content changed a prescribing decision, reinforced a clinical preference, or was read at all beyond the first scroll.
This creates a strategic paradox sitting underneath every infrastructure investment Andreychuk advocates. Native AI, modular content architecture, pre-MLR workflows, DOL programmes — each will eventually be evaluated for ROI. But without measurement frameworks that track behavioural change, time-on-content, and prescription correlation rather than activity volume, that evaluation is impossible. The 5–10% revenue growth through prescription influence that Andreychuk cites as the potential commercial outcome of effective content engagement is, without measurement reform, not just unrealised potential. It is unprovable potential.
Measurement reform is the least exciting of the reforms implied by this argument. It requires investment in data infrastructure, cross-functional alignment between commercial analytics and content operations, and a willingness to replace familiar metrics with ones that take longer to build and are harder to defend in quarterly reviews. It will be deferred in most organisations precisely because the other reforms like AI tools, modular content systems and DOL programmes, produce visible outputs that can be reported upward. Measurement produces accountability.
The companies that extract genuine value from the content architecture Andreychuk describes will not simply be the ones that build it first. They will be the ones that build the measurement capability to prove it works. In an industry that has spent $20 billion a year producing content it cannot prove anyone uses, the ability to measure is, finally, the ability to differentiate.
To get you highlights of Pharma 2026 faster, we are using generative AI technology to summarise the transcripts of the sessions. If you have any feedback about the summary, please contact lucy.fisher@thomsonreuters.com.
Discover more on this topic at Pharma Commercial Data & Tech Europe 2026 (4-5 November, London) Europe’s collaborative home for data and tech pioneers. Visit the website here.
