Right now, stop reading and look at your phone’s home screen.
Count how many apps are built specifically for AI—not regular apps that added AI features, but products designed from the ground up for the AI era. ChatGPT probably makes the list. Maybe a few others.
But for most of us, the answer is surprisingly close to zero.
This observation comes from Andrew Chen’s recent piece on how AI will change startup building, where he introduces what he calls the “Home Screen Test.” It’s a deceptively simple way to measure how far we’ve actually come in the supposed “golden age of AI.” (That was just a lead-in, but read his entire article to think about the paradigms that are shifting)
The results reveal something startling: despite all the AI hype, we’re still in the very early innings of building products that truly leverage AI’s potential.
The paradox is real. We’re living through what everyone calls the AI revolution, yet it’s virtually invisible in the place where we spend most of our digital time—our phone’s home screen.
This isn’t a sign that AI is overhyped. It’s a massive signal that the biggest opportunities in product management are still wide open.
What This Test Actually Reveals
The Home Screen Test exposes a weakness in how most companies are approaching AI product development. We’re stuck in “AI feature mode” instead of “AI-native mode.”
Think about the difference between mobile websites and mobile-native apps in the early smartphone era:
- Old approach: Take a newspaper → put it online = website, or take existing websites → make them work on mobile = mobile-friendly sites
- Breakthrough approach: Build something totally new for mobile = Instagram, Uber, Snapchat
We’re seeing the same pattern with AI today:
- Current AI approach: Take existing products → add AI features = most “AI” products today
- Future AI approach: Build something completely new for the AI era = ???
Most “AI products” are really just existing products with AI features sprinkled on top. A chatbot here, some automation there, and some predictive text.
But very few companies are asking the harder question: “What would this product look like if we designed it from scratch for an AI-first world?”
This creates enormous white space for product managers who can think differently. While everyone else is adding AI features to existing workflows, the biggest opportunities lie in completely reimagining what those workflows could become.
How to Use the Home Screen Test as a PM
Smart product managers can turn Chen’s observation into a practical framework for spotting opportunities and evaluating strategy.
Do a quick review of your mostly used apps:
- Count AI-native apps on your phone
- Analyze which categories have been reimagined vs. untouched
- Use your home screen as real-time market research
Apply it to your product strategy:
- Ask: “Could our product exist exactly as it is without AI?”
- If yes, you’re probably building in feature mode, not native mode
- Follow-up: “Are we reimagining the experience or just digitizing existing workflows?”
Use it for competitive intelligence:
- Track how fast truly AI-native products gain distribution
- Compare against incumbent apps that add AI features
- See whether the market rewards bold reimagining or incremental improvement
The most important insight: Great product managers develop pattern recognition for what’s missing before everyone else sees it. Your phone screen is literally showing you the future product landscape—you just need to know how to read it.
What “AI-Native” Actually Looks Like
To understand the difference between AI features and AI-native products, let’s explore what that vision might look like in healthcare—a space ripe for reimagining. Yes, they might sound crazy today.
B2C Example: Health OS
Today’s health apps are fragmented and reactive:
- Separate apps for fitness, symptoms, prescriptions, and appointments
- Even with AI features, they’re just digitized versions of analog health management
- You still have to actively manage your health across multiple tools
An AI-native approach would create something like a “Health OS”:
What it does: An AI-native approach would create something like a “Health OS”—an AI that continuously learns your body’s patterns from passive data and proactively manages your entire health ecosystem.
It might detect that you’re getting sick two days before symptoms appear by analyzing subtle changes in your voice patterns, walking gait (via phone sensors), sleep quality, and heart rate variability.
Instead of waiting for you to feel bad and then help you find a doctor, it would automatically reschedule your meetings, order groceries optimized for recovery, and book the right type of appointment with the right specialist.
Why it’s AI-native:
- Requires real-time pattern recognition across thousands of health signals
- No human could process voice changes, walking gait, sleep quality, and heart rate simultaneously
- It’s not a health app with AI features—it’s an AI-enabled health infrastructure
B2B Example: Clinical AI Teammate
Current healthcare software just digitizes analog workflows:
- Electronic medical records = paper charts on computers
- Diagnostic tools = reference books in software form
- Administrative systems = paperwork automation
An AI-native approach would create a “Clinical AI Teammate”:
What it does:
An AI-native approach would create a “Clinical AI Teammate” that attends every patient interaction, knows every patient’s complete history across all providers, and actively participates in care decisions in real-time. During a patient visit, it would simultaneously cross-reference symptoms with the patient’s 10-year health history, current medical research, similar patient outcomes from across the health system, and suggest treatment adjustments on the fly.
Why it’s AI-native:
- Requires instant synthesis of massive medical knowledge with individual patient data
- Impossible for humans to hold simultaneously in their heads
- Makes doctors superhuman rather than replacing them
The Opportunity for PMs
These examples illustrate why being early to spot AI-native opportunities matters so much. The companies that build Health OS or Clinical AI Teammate won’t be competing with existing health apps—they’ll be creating entirely new categories.
The mindset shift:
- Old question: “How can we add AI to our existing product?”
- New question: “What experiences become possible in an AI-first world that couldn’t exist before?”
What you need to develop:
- “AI-native opportunity radar”—the ability to look at any workflow and imagine a complete reimagining
- Pattern recognition for what’s missing before everyone else sees it
- Thinking “AI-first” instead of “AI-added”
The advantage goes to PMs who can create fundamentally new experiences while everyone else is incrementally improving existing ones.
Your Action Plan
Here’s how to start building your AI-native opportunity radar (I am doing it along with you):
Do the Home Screen Test routinely. Track which categories remain unchanged and which are being reimagined. Look for patterns in what’s missing.
Apply the “pre-AI impossibility test” to your current product. Ask: “What would this experience look like if we could do things that were impossible before AI?” Then work backward to what’s possible now.
Study workflows in industries you’re not familiar with. The best AI-native opportunities often come from applying AI-thinking to spaces that haven’t been touched yet.
Start building your catalog of “AI-native patterns”—the fundamental ways that AI enables new types of experiences rather than just optimizing old ones.
The apps that will dominate your home screen five years from now probably don’t exist yet. The question is: will you be the PM building them, or watching from the sidelines as someone else does?
Thanks to Andrew Chen for the provocation that sparked this thinking. The biggest opportunities in product management are often hiding in plain sight—we just need to know how to see them.