Hearth

Concept

My thesis started with a question that kept bugging me: when a device in your home sends data somewhere, do you actually know about it? Most of the time, no. And it's not because you don't care, it's because consent, where it exists at all, is a checkbox you sign once at setup, in legal language, before any data was even flowing. That's not meaningful consent. That's compliance theater.

Hearth is my answer to that gap. It's a privacy-first smart home system made up of three parts working together, a physical hub that sits quietly between your router and your devices, a mobile app that shows you what's actually happening, and an AI agent that explains it all in plain language instead of burying you in settings.

The Consent Gap

I pulled four academic sources and synthesized five recurring problems in how smart homes handle data: collection that's invisible by design, consent that's opaque and granted once then forgotten, an ecosystem so fragmented every manufacturer has its own app and policy, data literacy so low that even visible data can't be interpreted, and no user-owned audit trail, so data leaves and leaves no receipt.

These compound each other. Invisibility makes consent meaningless. Fragmentation makes unified control impossible. Low literacy makes even visible data unactionable. I couldn't fix this incrementally, it had to hit all five at once.

Why We Built This

I benchmarked five real products against those five problems, three consumer ecosystems and two network-level tools. The pattern was consistent: consumer platforms are accessible but opaque, network tools are transparent but require technical skill nobody outside this field has. That empty middle, accessible and transparent, became Hearth's design brief.

Four principles governed every decision from there: frictionless by default, explain before you act, never make someone choose between safety and privacy, and the data belongs to whoever lives in the house. The structural answer underneath all of it is a tiered consent model. Safety-critical transmissions pass through automatically and get logged for later review. Everything else queues for a plain-language decision before it leaves the home.

How It Works

Keeping these short, read the full walkthrough in the case study ↗

Step 1. Dashboard

One glance answers “is my home safe?” Six risk levels collapse down to three, Safe, Caution, Risk, and every connected device sits in one card regardless of brand.

Step 2. Consent Settings

Five data categories, each with three rules: always ask, auto-approve, always block. Early on, everything asks. As you promote judgment calls into rules, Hearth gets quieter as trust builds.

Step 3. Review Queue

Each request states who wants what, and why, in one sentence, then offers Block or Allow. Guided Review walks anyone unsure through three questions before it earns them the option to defer.

Step 4. AI Agent

Ask it anything and it explains the trade-off honestly, including the upside of allowing, then hands the decision back. It never decides on your behalf.

Step 5. Audit Trail

Every transmission reads like a sentence: device, category, what happened, when, outcome. Blocked entries get a “what was protected” field, logging not just what happened, but what didn't.