The Velocity Trap: Why Pharma’s AI Content Boom Isn’t Paying Off

This time, there’s no guest. Nataliya Andreychuk hosts Pharma Talks alone for an unfiltered look at what pharma has actually gotten for its AI spend — and it isn’t quite the story the investment numbers suggest.

A $250 Billion Question

Nataliya opens with a figure worth sitting with: companies across the world spent $250 billion on AI in 2024 alone, and by her estimate that number has grown exponentially since. From where Viseven sits — deep in content and communication — she can trace much of that spend directly to how pharma produces content. Most of it, she says, has gone toward one thing: volume. New channels, new formats, faster localization across markets. The industry has, in her words, democratized information for healthcare professionals and patients to a degree it never has before.

But she poses the question the rest of the episode is built around: is more actually better?

The Velocity Trap, in Numbers

Nataliya lays out the tension with two statistics. Content production alone grew 29% year over year — more than at any point in pharma’s history. Speed is up, localization is easier, and at some level, cost per asset has come down (she’s careful to note that the true cost of AI production itself — compute, tokens, environmental impact — is a bill still to be settled, just not the topic for this episode).

Set against that acceleration is a harder number: 77% of that content — three out of every four approved, MLR-cleared assets — was never used. Produced, reviewed, published into the world, and then forgotten.

That gap between output and impact is what she calls the velocity trap: optimizing so hard for faster, cheaper, and more that the industry stopped checking whether any of it was landing.

Becoming an “AI-Native” Organization

The clearest evidence, she argues, is on the receiving end. Nataliya noted that roughly 40% of healthcare professionals say they’re less likely to engage at all when content isn’t tailored to them — when it doesn’t reflect prior communication history or any real human understanding of who they are.

Her read on this isn’t that AI failed. It’s closer to a paradox: pouring investment into AI has made it obvious exactly where the real effort needs to go. Broken workflows didn’t get created by AI — they became visible because of it.

Nataliya’s own thinking here started at home. She describes conversations inside Viseven about what it would take to become a genuinely AI-native organization — including hiring more AI-native talent and comparing their output against standard productivity to see what a real step-change looks like.

She connects this to what’s playing out across the wider industry: layoffs, job changes, restructuring. In her framing, this is pharma going through the discomfort of becoming AI-native, and every organization has to work out what that actually means for itself.

Even with efficient, self-service AI content generation, she argues three things remain non-negotiable: digital asset management tied to medical and regulatory history, AI trained on the organization’s own compliance framework, and a real role for human-in-the-loop (HITL) review. She points to major vendors already moving in this direction — acquisitions and agentic AI tools built specifically to review promotional and medical materials, aiming to cut human review effort by 70% while deliberately preserving that remaining 30% for people.

The Humics Framework

To describe what that 30% is actually for, Nataliya introduces a framework she first encountered through the work of researcher Pascal Bornet, the author of Irreplaceable: The Art of Standing Out in the Age of Artificial Intelligence. Humics — a way of defining what makes humans valuable inside an AI-native organization. She names three components:

• Genuine creativity — not recombination or templating, but ideas that originate from lived human experience and judgment, which AI doesn’t have.
• Critical thinking — the willingness to question AI output, sit with uncertainty, and reason your way to a decision rather than accepting the first plausible answer.
• Social authenticity — real human-to-human connection built on trust, the kind AI can imitate on the surface but not originate.

She illustrates the stakes with a comparison: an MSL’s real interactions with a healthcare professional versus an AI-generated summary of the same relationship. The AI output, she says, will often be statistically plausible and even scientifically accurate — but reasoned through fact by fact, it’s shallow. Without the real human interaction behind it, it doesn’t carry the same value, and the ethical territory around treating it as a substitute is genuinely murky. AI, in her view, is a powerful vehicle for executing across the content supply chain. The decision-making still belongs to people.

Three Takeaways Before the Next Episode

Nataliya closes with three points she wants listeners to sit with:

Speed isn’t the same as impact

Every time content moves faster and cheaper through the content supply chain, ask whether it’s actually closing the gap between what’s produced and what healthcare professionals and patients expect — or just widening the velocity trap.

Build AI on your own foundation

Pharma-native AI has to be grounded in your own digital asset management, your own compliance history, your own record of healthcare professional interactions, and your own scientific knowledge. She’s blunt that this is no longer optional — competitors are already building this way.

Technology only pays off through people

The return on AI investment shows up only when people can do what AI can’t: create, think critically, and connect. Organizations that build those capabilities will see AI amplify the gaps they already have. Organizations that don’t, won’t.

Final Words

Nataliya ends with a direct challenge for listeners to take into their next AI roadmap review: is it a roadmap to scale your content, or a roadmap to grow your people? Her argument is that these are not the same question, and pharma has mostly been answering the first one. The organizations that shift toward the second, she says, are the ones that will avoid ending up with more content and the same velocity trap — just faster.

Enjoyed this conversation? Want to talk through what an AI-native content strategy could look like for your organization? Talk to us, and let’s explore it together.

---