I watch my dad worry.

My grandmother lives with my parents. She's mostly fine. But every time my dad leaves the house, even just to run errands - I see it. The quick check before he walks out the door. The slightly shorter trips than he planned. The "I should probably head back" that cuts conversations short.

He doesn't say it explicitly, but I know what he's thinking: What if something happens while I'm gone?

And here's what terrifies me: I'm going to be him in 20 years.

My parents will age. They'll need checking on. And I'll be the one cutting trips short, feeling guilty about living my own life, carrying that low-grade anxiety everywhere I go. The "are they okay?" question that never fully goes away.

This isn't about emergency detection. My dad isn't worried about worst-case scenarios. He's worried about the normal stuff: Did she wake up? Is she wandering confused at night? Did she eat? Is everything... okay?

Most monitoring solutions miss this entirely. They're built for crisis response-fall detection, emergency buttons, medical alerts. But that's not the daily anxiety. The daily anxiety is simpler and more persistent: "I just want to know she's okay so I can actually live my life."

And there's no good answer right now. Cameras feel like you're spying on your own parent. Calling multiple times a day feels intrusive and creates its own burden. Wearables get forgotten or rejected. So you just... carry the worry.

I'm building this because I refuse to accept that "constant worry" is the price of being a good daughter.

It's basically an invisible guardian system - though I'm still figuring out the actual name (hit me up if you have some good name ideas)

Here's how it works: non-camera motion sensors scattered throughout the home. No cameras. No microphones. No wearables. Just passive sensors that detect movement and learn daily routines over time.

The system learns when someone typically wakes up, when they move around, when they sleep. It builds an individual baseline - not based on population averages, but on this specific person's patterns. Then it only alerts caregivers when something meaningfully deviates from that baseline.

Right now, I'm limiting the MVP to exactly two alert types:

That's it. No alert spam. No false alarms about someone sitting still for 20 minutes. Just high-confidence signals that something might actually be off.

The goal is peace of mind without stripping away dignity. The older adult doesn't feel watched. The caregiver doesn't feel anxious. The system just sits quietly in the background until it actually matters.

I've built the MVP. Sensors are deployed in two places: my apartment in California and my parents' home in Bangalore.

And I'm learning that "just learn a routine" is harder than it sounds.

The system threw a false positive when I came home late on a Saturday night - flagged as unusual night activity. It threw another alert when my parents traveled and left early one morning, which skewed the baseline it had built.

Turns out a 7-day baseline gets confused easily. One outlier day and the whole model drifts. Weekday routines look different from weekend routines. Travel throws everything off.

So I'm debugging. Figuring out how to make the system smarter about what's actually concerning vs just... life happening.

But here's what I know: the core problem is real. My dad's anxiety is real. And I'm convinced that privacy-first, routine-based monitoring is the right approach - I just need to make it actually work reliably.

Next milestone: deploy to 2-3 more homes of friends and family for real validation beyond my own testing. And pray my 9-5 doesn't discover I've been thinking about baselines and false positives during standup.

Yes, I see the irony.

I'm building a privacy-first product... while publicly documenting the process. I'm literally calling this newsletter "Building Quietly" while being anything but quiet about it.

But here's the distinction: I'm documenting the messy process, not hyping the outcome. The false positives. The "why didn't I think of that?" moments. The gap between what I want to build and what actually works today.

That's the transparency part. The "quietly" part is what I'm building: something that sits in the background and doesn't demand attention unless it matters.

So this newsletter is my forcing function. Every two weeks, I'll share what I'm learning, what I'm struggling with, and what I'm deciding. Sometimes it'll be technical. Sometimes it'll be strategic. Sometimes it'll just be honest founder doubt.

If you've ever worried about aging parents, if you've ever felt that anxiety gap, or if you're just curious about what it takes to build something from scratch - this is for you.

See you in two weeks

— Shwetha