Sizzle and Steak...

We’ve all heard the phrase: “Sell the sizzle, not the steak.” And sure—sizzle matters. You need attention. You need a story that resonates.

But if the steak isn’t any good—if the product doesn’t actually deliver—people feel duped. Overpromised. Burned. They won’t come back. And they definitely won’t evangelize on your behalf.

So yes, sell the sizzle. But you better make sure the steak lives up to the hype.

You’ve probably also heard the expression “technology in search of a problem.” I’ve been there. It rarely works. People get excited about what they’re building—the creativity, the process, the tech. But sometimes they get so deep into how they’re building that they lose track of why. They chase solutions to problems that don’t really exist—or ones that aren’t painful enough for anyone to change behavior to fix.

And with the surge in generative AI tools, we’re going to see a lot of that in the near future.

This story is the opposite. It’s about a real problem—clear, recurring, and underserved—and how we found it through a tool we quietly ran together for years.

My wife is a physician. She spent over a decade at UCLA and the VA working as a psychiatrist, primarily with U.S. veterans—arguably one of the most complex patient populations in the country. She was trained in a therapy called CBT-I—Cognitive Behavioral Therapy for Insomnia.

In 2018, she started a private practice to focus on insomnia and other sleep disorders. A core part of CBT-I involves daily sleep journaling—collecting detailed, structured data on sleep timing, quality, and disruptions. Over time, this data becomes surprisingly robust. It can be correlated with medications, therapy adjustments, substance use (like alcohol, caffeine, or nicotine), stress levels, and other lifestyle behaviors. It forms a clinical baseline that helps her understand what’s working, what isn’t, and how to tailor treatment in a measurable way. The problem was, no simple tool existed to support her approach.

She described what she needed, and we built a simple, fast tool tailored to her private practice. It was a basic Google Form connected to a spreadsheet. Quick to fill out, easy to manage. On the backend, we added lightweight scoring logic to help her spot trends and guide the conversation during weekly patient encounters.

The application worked great for her and her patients from day one. Sure, there were tweaks here and there—small changes to improve usability or refine the scoring—but overall, it did exactly what she needed it to do. Patients engaged with it consistently, and the data gave her a clear view into what was helping and what needed to change. And they didn’t need any nudging. Because it was part of the therapy. And when your doctor says, “Do this,” and you want to sleep again—you do it. “Doctor’s Orders” are powerful words.

After COVID, her practice shifted to 98% telehealth. Patients still used the same simple Google Form to log their sleep each day—but now, they weren’t sitting in the room watching her flip through papers or screens. Instead, the results were already synthesized and ready when the session began. The data quietly shaped the conversation without distracting from it. Ironically, by removing the visibility of the tool, the journaling became even more effective. The session felt more natural, and more informed.

Over time, the data in the spreadsheet ballooned. More patients, more rows, more logic. Sorting got slow. Calculations dragged. On nights and weekends, I became her part-time IT department and the product needed to handle the scale better.

But here’s what stuck with me: the data wasn’t just abundant. It was rich. Some fields were open-ended. Patients wrote paragraphs. They weren’t just checking boxes—they were trying to get better. And it showed. In fact, some patients continued using the journal for months, even years, after their therapy had ended. They found value in the structure it provided—an ongoing framework that helped them maintain progress and extend the benefits of treatment over time.

A few years ago, she mentioned that her colleagues had started asking about her tool. And not just for sleep. They saw potential applications for pulmonary conditions like asthma and COPD, for diabetes and vascular disease, and across a wide range of mental health therapies. The structure and consistency of journaling—when done well—could help clinicians track symptoms, behaviors, and responses over time in ways that traditional check-ins often missed. That got our attention.

When generative AI became more accessible, we started talking seriously about how to take this to the next level—automated, conversational, more scalable. I’ve spent the better part of my career as a Chief Product Officer in healthcare tech, AI, and NLP. So the vision came together fairly quickly.

And now in 2025, we are—just weeks away from launching flagship V1 product, PulseGen, that brings all of this together. More on that soon...

Thanks for reading. Leave a comment, send a note, or dispatch a raven—we’d love to hear from you.

Mark