The Fight He’d Already Had

This story grew from my Healthcare UX podcast chat with Smit Doshi of MyHappy.Health.

I showed up ready. That’s the best way to put it. I’d spent days getting ready—a question for every gentle answer, a deeper follow-up tucked behind each one, keeping my toughest points in reserve so they’d land naturally instead of giving the game away. I even had a label for the trap I wanted to avoid: a guest who offers a polished script and moves on, leaving me with a nice-sounding but empty episode. I know I can be too quick to accept smooth answers, so I designed my whole approach to keep myself from doing that.

The product gave me lots to explore. A free, voice-first AI health companion for people who often don’t have a doctor. A friendly face on the screen. A tool that reads X-rays and suggests possible conditions. Each of those could easily cross the line from helpful to practicing medicine—especially for users who might not catch a mistake. I’d lined up questions for every sensitive spot.

Then Smit beat me to them all.

Not by avoiding them—but by having already thought them through. The point I wanted to reach—that a confident but wrong AI answer could harm or even kill someone—he said on his own, without being prompted. The data concern I’d sharpened—where does a scared person’s information really go?—he opened up, naming the issue before I could even ask. And then he gave me something I hadn’t even known to request.

His team created an anemia screener: just a photo of the inner eyelid to estimate hemoglobin. It worked perfectly from start to finish—and then they turned it off. Why? Because a city woman with a low reading can head straight to a clinic for a blood test, but a village woman with the same result has no backup. She’d have only the app to go on. There’s not enough data yet to be that sure. So the feature sits there, ready but dark.

I base my whole approach on that idea: map the failure path before the success path. A confident wrong answer hurts the least-resourced person the most. I’ve written about it more than once. And here was a builder who had already arrived at it on his own and accepted the cost—a finished feature, ready to ship, turned off—before I even showed up.

I should have been thrilled. Mostly I was. But underneath, I felt a little uneasy.

Because this is exactly what the trap looks like.

The overly agreeable person is the one you learn to question. In research and interviews, a smooth session is the one that worries you: they agreed with everything, so you learned nothing—no real signal. A conversation with no pushback usually means you missed the truth. My instinct, shaped by years of this, saw the smoothness as emptiness—soft tape, no edges, nothing pressed because there was nothing to press against.

And that instinct was wrong. That’s what I’ve been thinking about.

It took me a bit to realize why this agreement felt different. He didn’t dodge the tough questions—he answered them before I even asked, already including the counterexamples. That’s the key. Someone who’s just deflecting gives you a neat frame and stops there. Someone who’s really done the work gives you the frame and then hands you the sharp edge: the blind spot, the missing feature, the honest admission that they don’t know exactly where the line is—and that not knowing is the answer. He offered the very counterexamples I’d planned to pull out of him. Deflection doesn’t do that; it hides the weak spot. He led me straight to his.

So here’s the craft issue I’m left with—and it’s mine, not his:

From where I sit, in the moment, both the deflector and the person who’s done the work can sound equally smooth. Both are calm and fluent. Both leave you feeling there’s nothing left to push on. The real clue isn’t the smoothness—that’s the same. It’s where the counterexample comes from. Did the guest bring the complication themselves, or did I have to dig it up? That’s the only reliable sign I’ve found. And you can only catch it if you’re prepared enough to know what the counterexample should be.

That’s the part I can’t quite settle, and I won’t pretend I can. Because “he brought his own counterexamples” is exactly what a very smooth operator would do, too. A polished pro plants the objection and answers it, so you leave thinking you saw behind the curtain when you really saw just what they wanted you to see. I don’t think that happened here. I believe he did the work. But “I believe he did the work” is just a feeling I had from my chair. And the whole point of my prep is that I don’t trust feelings alone.

So I’m keeping both. He put in the effort, and I feel good enough about it to share it. I still keep a little doubt about my own confidence—because the moment I let that go is the moment a slick pitch might slip by me.

The prep didn’t catch him in anything. Instead, it did something I didn’t expect but really appreciate: it showed me exactly what he’d already worked through in himself. You need those tough questions even if you never use them. They help you gauge the size of the battle the other person already fought—the one they finished long before sitting down with you.