Why Fast AI Prototyping Isn't Helping Your Team Build the Right Things
Speed and fidelity are worthless if you don’t know what question you’re trying to answer
Your product team likely never tests. Anything. You just build the next thing you’ve been told to build on your stakeholder-dictated roadmap.
And when you do test, it’s only to “validate,” to look for approval from customers and stakeholders to show you were right all along.
AI tools like Figma Make, Lovable, and Replit let you build faster and at higher fidelity than ever before. That reduced friction has created a new problem. The prototypes Product, tech, and UX are building now compete with each other for customer review and stakeholder approval.
And none of them answers the question of whether you’re building something customers will pay for.
The Phoneless Future That Never Happened
Humane AI had $230 million in funding, Apple-pedigree founders, and a stubborn insistence on a vision they never properly tested.
Critics expressed skepticism as soon as their first commercial dropped, even as the team worked around the clock to ship. Behind the scenes, founders Bethany Bongiorno and Imran Chaudhri, both Apple veterans, enforced a toxic positivity culture, dismissing employees who raised hardware or software concerns for “violating company policy.”
Humane’s glitzy launch and celebrity endorsements were followed by scathing reviews just a week after its long-awaited wearable “AI pin” device shipped in April 2024.
Just four months later, in August, daily returns started outpacing new sales, with over $1 million in returned products piling up and thousands of orders canceled even before they shipped.
By February 2025, Humane’s flameout was complete. HP acquired most of the company’s assets for $116 million, roughly half of what investors had invested. The founders and part of the team were absorbed into HP’s new AI lab.
But by far the biggest losers? The thousands of customers who paid $699 to be part of the next-generation “phoneless future.” All they were left with was a bricked, worthless device as of early 2025.
Two decades of Apple design and engineering experience weren’t enough to save the founders from their failed vision.
The Humane pin turned out to be something they wanted to build, but not something people wanted to own.
Humane’s Four Failures
While we might call Humane a “failure,” they failed against four different risks that threaten every product effort:
Value Risk: Not enough customers saw the value or were willing to pay for it.
Usability Risk: Customers who ordered the pin couldn’t get what they hoped for from it. Returns outstripped new shipments within months.
Feasibility Risk: Despite a roster of Apple veterans, the team couldn’t solve the hardware and software problems they refused to acknowledge. Quality suffered because internal feedback loops were shut down.
Business Viability Risk: Humane’s business model rapidly fell apart once it hit reality. A $699 device plus a $24/month subscription, with unresolved safety concerns that forced a charging case recall.
What if instead of attacking its own people, Humane’s founders had remained open to internal and external feedback, and continued testing and prototyping with their industrial and software engineers and end users?
What could the Humane AI pin have become?
Fail in Small Ways First
Humane failed in one spectacular, expensive, public way.
The alternative is to intentionally fail in many small, cheap, low-risk ways first. And that’s exactly what prototyping and testing do. It shows you where you’re wrong early so you can fix it before the stakes get that high and you can still change direction.
The different prototyping approaches below are how you do it with intention instead of luck.
The Three Ways To Use Prototyping
Prototyping and testing serve three distinct functions, each reducing a different kind of risk.
Build to Think to try out new ideas and open up possibilities
Build to Learn to run experiments and disprove assumptions of how users might react
Build to Align to share a clearer picture with extended team members and leadership
These aren’t deterministic steps in a linear path. Teams move in between them in waves as their understanding grows.
Pair these functions with the stakeholder mapping from Issue 4, and you know who to test with, what to test, and how much to invest in each test.
Paying subscribers get the frameworks, fidelity levels, a 15-minute exercise to put all three functions to work, and my exact prompts to walk you through the exercise for additional support.
The Prototyping Challenge
Run these two questions against your current work before you keep reading.
Even if you’re not aware of it, you’re currently following a strategy. How many uniquely different possibilities did your team generate before you decided to follow that strategy? How did you prototype and test your choices?
Pull up the last three features your team shipped. How did you prototype and test each one? With whom? Which of the three functions did you use?
Surprised? Good.
Your Move
The only way to pressure-test the assumptions underlying your strategic and feature choices before your customers do is through prototyping and testing them first.
The paid section walks you through four frameworks to apply all three functions.
You’ll learn why fidelity matters in testing, and why higher fidelity isn’t always better. You’ll also learn how to match the right stakeholder ask to the right prototype.
And, no, “What do you think?” is never the right question to ask.
By the end, you’ll have a Test Design Plan, pre-committed assessment criteria, and a Money Story you can bring to your next planning cycle.
Let’s Walk Through How You’re Prototyping & Testing
Reply TEST and tell me about the prototypes your team has built recently, which approaches you used, and what you learned.