How Sketchy Brings Medical Learning to Life with Voice AI

Medical school is challenging, and that’s putting it lightly.

Just a few years ago, students would spend their days memorizing endless flashcards and information from dense textbooks. Still, there was a lingering sense that no matter how many facts you cram in, nothing fully prepares aspiring doctors for real patient interactions. 

Sketchy’s founders knew that feeling all too well. They were med students themselves in 2013, sketching characters onto napkins to help make learning microbiology topics stick. 

Soon those napkin doodles turned into a full-blown visual learning platform now used by hundreds of thousands of medical students worldwide.

But there’s one part of the clinical journey that even the best illustrations couldn’t capture: real-world patient conversations.

When Sketchy began experimenting with patient simulations, the goal wasn’t just to recreate scenarios. It was to evoke an emotional response in students. The hesitation in a patient's voice. A subtle confusion. The subtle cues that turn clinical knowledge into clinical judgment.

That’s where ElevenLabs came in.

Rethinking How Medical Students Practice ClinicalSkills

Sketchy had already made a name for itself, turning complex medical content into visual memories. But as the team dug deeper into clinical education, they noticed a gap. 

Students were graduating with strong foundational knowledge, but often lacked practice in clinical reasoning, communication, empathy, and the messy uncertainty that characterizes real-life patient care.

OSCE prep, clerkships, and oral boards demand more than recall. They require flexibility and an active presence. And no flashcard deck teaches you how to respond when a patient asks their first question, or how to navigate cultural nuance when delivering difficult news.

Sketchy started exploring ways to simulate these moments. The answer wasn’t more static content or branching-choice videos. It needed to feel like a real conversation, with as much of the real-ness and unpredictability that comes with it.

ElevenLabs Integration: Building Voices That Don’t Sound Like Bots

“We were already using AI to build interactive cases,” Ben Muller, MD, Sketchy’s Chief Content Officer shared. “But the experience didn’t feel fully human until we added voice.”

Early experiments with off-the-shelf voice models were underwhelming. Flat tone. Robotic pacing. Emotional disconnect. In clinical learning, those missing pieces matter.

What made ElevenLabs different was how real it sounded, down to the inflections. The platform gave Sketchy access to emotionally expressive voices in dozens of languages, with fine control over pacing, tone, and even phonetic quirks.

One of the early challenges? Temperature. Literally. 

“Voice models were reading things like ‘98 °F’ as ‘ninety-eight degrees… F,” Dr. Muller recalled. The fix involved creating phoneme dictionaries to guide pronunciation. A small technical detail, but a telling one. Because in clinical dialogue, precision counts.

How Did Voice AI Benefit the Medical Learning Platform?

Once ElevenLabs was integrated, changes in the student experience were immediate. 

Patient simulations stopped sounding like actors reading scripts and started feeling like conversations you might actually have on the ward. Sketchy used ElevenLabs to design voices for characters with distinct personalities — hesitant, warm, assertive, and confused. Each one was tailored to match specific clinical scenarios.

A student working through an asthma case might encounter a worried mother asking rapid-fire questions. Another case might involve a teen quietly downplaying symptoms. The AI agents began to respond authentically, adjusting their delivery to sound like natural human speech.

Soon enough, overwhelmingly positive feedback began to pour in:

“This was a ton of fun… I can’t wait for more cases.”

“I loved the patient’s personality and the lifelike conversation she was able to have.”

Students were doing more than just learning: they were connecting in real-world, interactive practice. And that’s what inspired the creation of Sketchy in the first place.

Clinical Learning Extends Beyond “Right Answers”

Sketchy wasn’t chasing novelty for novelty’s sake. Their content and product teams rigorously tested how voice interactions shaped students’ confidence, recall, and long-term readiness. What they found was encouraging.

By replacing passive multiple-choice with active dialogue, students practiced what to say, and how to say it. 

Since the format was interactive, the team could also study how students learned: what they paused on, where they hesitated, and how quickly they adapted.

For Sketchy, these insights weren’t abstract. They fed directly into the design of future simulations, helping refine both the educational content and the AI logic behind it as a beneficial learning development flywheel. 

How ElevenLabs Facilitated a Platform Shift

From a business angle, integrating ElevenLabs made sense. 

Voice AI interactions have become a clear differentiator, particularly for institutions seeking more immersive and repeatable learning formats. The voice-powered cases keep learners engaged longer and encourage more frequent return sessions (a metric the team tracks closely). 

But the real value was harder to quantify. Students came away from simulations better prepared and also more curious, reflective, and willing to try again and improve.

That shift in mindset toward active, self-directed, human-centered learning isn’t easy to engineer. But with the right voice tech, it became possible for Sketchy. 

What Comes Next?

Sketchy now sees voice AI as a pillar of its future platform, not a one-off innovation. 

New use cases are already in development, encompassing scenarios that extend beyond clinical facts into areas such as ethics, teamwork, and conflict resolution.

As diagnostic AI tools improve, Sketchy’s team expects their role to evolve. It won’t be 

enough to train students to deliver the correct diagnosis. They’ll also need to teach how to interpret AI outputs, manage uncertainty, and communicate risks to patients with clarity and empathy.

In that context, human connection becomes a competitive edge.

And voice, genuine, imperfect, human-sounding voice, plays a central role.

Sketchy’s Advice for Other EdTech Teams

If there’s one lesson Sketchy would offer peers in the EdTech world, it’s this: Don’t just digitize what already exists. Use AI to build what couldn’t exist before.

Voice AI unlocked new capabilities to simulate patients who interrupt. Emotions that catch you off guard. Characters that shape how a patient’s case is perceived emotionally. These aren’t gimmicks. They are tools that train students to navigate the real world, rather than a linearly scripted one.

As Girish Krishnaswamy, Sketchy’s CEO, put it: “The goal was never to replace instructors or automate empathy. It was to give educators superpowers and to make learning unforgettable for students.”