Menu
Close
.svg)
About four years ago, I found myself participating in one of the most interesting, unusual, and, at the time, remarkably forward-leaning marketing research studies of my career.
The business question itself was straightforward. We were working in a rare disease category and trying to evaluate what might prompt specialist physicians to escalate therapy as a patient's symptoms progressed. Traditionally, we might have presented respondents with a written patient profile or a series of case vignettes and asked them to react. Instead, we decided to experiment with something considerably more immersive.
We introduced them to a patient. Or more precisely, a simulation of one.
Appearing on screen was a digital avatar controlled in real time by a member of the research team or one of our clients, working from a semi-structured script. Physicians could query the avatar, explore symptoms, probe inconsistencies, and gather information much as they would during an actual exam-room clinical encounter. Following the interaction, the moderator would transition into a broader discussion about treatment decisions, escalation triggers, prescribing rationale, and the overall experience itself.
By today's standards, the technology was admittedly primitive. The avatar was closer to a digital puppet than the sophisticated AI-driven systems available to us mid-2026. Yet even then, I observed that something fascinating emerged.
The physician respondents in our study behaved differently.
They leaned into the interaction. They asked more questions. They displayed greater curiosity and, perhaps most interestingly, more empathy. The experience felt less like evaluating a research stimulus and more like navigating an authentic clinical conversation. As a result, we uncovered new dimensions and nuances of physician decision-making, patient communication, and treatment behavior that would have been difficult -- if not impossible -- to surface through a static case presentation alone.
I remember leaving those sessions with the distinct feeling that we had stumbled onto something much larger than a novel research technique. At the time, I viewed it as an interesting methodological innovation. Looking back, what we were actually seeing was the beginning of a new way to create more realistic human interactions inside research...and to learn from them.
Fast forward to June, 2026.
Over the past year, our AI team have been applying itself and investing heavily in what we call ThinkMATES: digital avatars that are dramatically more sophisticated, more lifelike, and more capable than anything we were experimenting with only a few years ago.
What strikes me is not simply how much more realistic they have become. It is how quickly clients have begun identifying practical applications for them. Use cases for AI avatars is among the most frequently discussed topics when we engage with clients these days.
A few months ago, a client posed an interesting challenge. They were developing a new therapy and wanted to understand whether strong clinical data alone would be sufficient to drive adoption in the real world. As we talked through the problem, it became clear that what they really wanted to understand was not the product itself. They wanted to understand the friction surrounding the product. What I mean is:
These are not hypothetical situations. They are daily realities of workflow challenges that get in the way of a doctor's using a new product, despite their finding its profile attractive.
Historically, we might have asked physicians how they think they would react in those circumstances. Increasingly, however, we can simulate those interactions and observe the response directly.
That is a fundamentally different breed of learning.
Earlier this year, I wrote about the use of AI avatars in visual aid testing (just on the heels of being blown away by what I saw at PharmaUSA). For decades, our industry relied on either live sales representatives or various forms of "rep-in-a-box" recordings to present promotional materials during research. AI avatars represented a natural evolution of that model, offering consistency, scalability, compliance, and multilingual flexibility.
But the opportunity is much broader than visual aid testing.
Today, we are using avatars to create more immersive future-state simulations, helping physicians react to products that do not yet exist in the marketplace. We are bringing patient personas to life in ways that create a stronger emotional connection between stakeholders and the people they are trying to serve. We are experimenting with avatar-based outputs that allow insights to be delivered in different languages and customized for different stakeholder groups.
In many cases, the goal is not efficiency. It is realism.
The more closely we can approximate the real-world environment in which decisions occur, the more confidence we can have in the insights that emerge.
The really exciting developments are still ahead of us.
Most avatar applications today remain largely scripted. They present information, guide interactions, and support simulations. Increasingly, however, they are becoming conversational.
Our teams are already experimenting with avatars capable of engaging in dynamic dialogue, adapting responses, and interacting in ways that feel remarkably human. One can easily imagine a future in which avatars serve as simulated patients, physicians, caregivers, moderators, trainers, or even stakeholders inside workshops.
None of these ideas feel particularly far away anymore.
The most important thing to remember, however, is that avatars are not the story. Understanding people is the story, and that remains our North Star.
The technology is simply giving us new ways to access that understanding in a more realistic, immersive and engaging way.
After spending decades talking to physicians, patients, nurses, caregivers, and healthcare leaders, I've learned that the most valuable insights often emerge when we can place people into situations that feel authentic enough for real behavior to surface. Digital avatars are becoming one more tool that helps us do exactly that.
Three years ago, we were experimenting with digital puppets. Today, we are building digital colleagues, digital patients, and digital stakeholders.
And if the questions we are receiving from clients are any indication, we are still very early in the journey.
