Puppeteer was built to tackle a recurring issue in healthcare: communication gaps that too often result in real patient harm. Working on early Generative AI projects showed us just how powerful conversational agents could be—but also how risky they are in sensitive contexts like healthcare. Yet that was exactly where the need was greatest. Patients were missing follow-ups, misunderstanding instructions, and slipping through the cracks—not because clinicians didn’t care, but because the system didn’t scale. So we set out to build an AI-powered platform that could support healthcare teams at scale—without compromising safety, empathy, or accuracy.

Focusing on Patient Communication Just Made Sense

Healthcare is a high-impact, high-stakes domain where Generative AI holds incredible potential, but only if applied responsibly, with accuracy and empathy as top priorities. We knew safety had to be more than just a feature—it needed to be the foundation. Patient communication isn’t like customer service or tech support; these are vulnerable people in sensitive situations, and one wrong word can erode trust. That’s why we built robust guardrails to keep conversations accurate and safe, embedded comprehensive knowledge bases for the right context, and added human-in-the-loop capabilities so professionals can intervene when needed. Our goal wasn’t just to showcase advanced technology; it was to make it trustworthy.

By focusing on patient communication and implementing these safeguards, we could make interactions safer, more reliable, and more empathetic. At the same time, we free doctors and nurses from repetitive communication tasks, allowing them to concentrate on what really matters—caring for patients. That was our moment of clarity: this was how we could truly make a difference.

Why the Name 'Puppeteer'?

Naming it Puppeteer just made sense. Imagine an intricate puppet show—behind the scenes, there’s a master puppeteer skillfully pulling the strings, coordinating every movement to create a seamless, lifelike performance. That’s exactly how our system works. It’s not just one AI model handling a conversation; it’s an entire network of models orchestrating complex interactions in real time. One model interprets patient sentiment, another ensures medical terminology is accurate, and yet another keeps the context consistent throughout the dialogue. Each one works independently yet harmoniously, just like puppets on a stage controlled by a skilled puppeteer. This dynamic, modular architecture is what makes Puppeteer flexible enough to handle any healthcare communication scenario.

Listening to the People Who Matter

Our clients have shaped Puppeteer just as much as we have. We’re constantly listening and iterating. That’s how we ended up with features like detailed performance analytics and patient-specific metrics.

One of our earliest clients was a mid-sized telehealth clinic that was overwhelmed with patient follow-ups. Nurses were spending countless hours on repetitive phone calls for appointment reminders and medication check-ins. During a demo, the clinic director said something that struck a chord: 'If the agent could handle phone calls too, that would be a game-changer.'

That feedback clicked. We realized it wasn’t just about messaging patients; it was about removing communication bottlenecks. We got to work, building a voice-capable AI that could handle natural, meaningful conversations over the phone. When we presented it, the director’s reaction said it all: 'This changes everything.' Now, they could focus on patient care instead of being tied to the phone all day.

The Right Team

We knew from the beginning that the team would make or break us. So we kept it simple: A-players only. We use a simple philosophy to identify the best talent—a framework that categorizes people into five levels. We only hire at the top two levels: Level 5, the 'Superstar,' who identifies the right problem and gets it solved, and Level 4, the 'Systems Thinker,' who sets up efficient systems to solve the problem. It’s not just about experience—it’s about mindset and the ability to own challenges. Building that kind of team meant every person brought something exceptional to the table, and it shows in everything we build.

Why We Do This

Puppeteer isn’t just a product—it’s a journey. From the early days of exploring AI at Meta to building LangAI, and finally shaping Puppeteer with a team I trust, it’s been about solving a real problem. We’re not just innovating for the sake of it. We’re building tools that actually help people—tools that make patient communication safer, more accurate, and more human. And we’re just getting started.