Prototyping 101

When most people picture innovation, they think of the lightbulb moment—that flash of inspiration when someone is struck by a new, world-changing idea.

At X, innovation is a collaborative, iterative, often messy process. We start with a huge problem, one that, if solved, would transform society. Then we fail to solve it. Over and over and over again, because that failure is how we learn. Only by testing many different ideas (most of them bad!) can we learn as fast as possible, in hopes of identifying the radical solution we’re after.

The key to that process is prototyping.

Prototyping is how ideas make contact with the real world. It is how concepts are tested, assumptions are upended and the future is invented.

While there is no one way to prototype, here are a few lessons we’ve learned after putting countless moonshot ideas through their paces in our Design Kitchen.

"What am I trying to learn?"

At X, a prototype is anything you create to learn, so every prototype should begin with the question: “What am I trying to learn?”

Are you trying to understand a fundamental technical aspect of your project? Then you probably need an experimental rig just to see if the physics will work. These “works-like” prototypes are the farthest thing from a finished product.

The world’s first transistor invented at Bell Labs was formed out of gold foil, a germanium crystal, some plastic, a metal base and some wires. Today, tens of billions of transistors fit inside of your phone, but the very first one was larger than a deck of cards.

Prototypes don’t all need to be physical, either. If you’re trying to solve a problem with software rather than hardware, then creating a digital prototype is essential. Prototyping software used to require specialized knowledge, which meant hardware experts prototyped hardware, and software engineers prototyped software. Vibecoding, Arduinos and 3D printers have completely blurred that line, so that experts from any domain can now prototype with software and code an app almost as fast as you can think.

In just a few weeks, we built a working software demo for an early-stage moonshot focused on eliminating the massive waste in harvesting and processing raw materials, even though it was trained on messy multimodal data and involved complex industrial systems.

Maybe you’ve landed on a solution, but now you need to learn whether your solution will ever be profitable. In that case, you’d need a financial prototype to work through a techno-economic analysis.

Each stage of an innovation may require a different prototype, and before you get started, it’s essential to think through what you first hope to learn, what kind of prototype you need and how you’ll plan to test it once it’s built.

V0.crap

If your first prototype is pretty, you’re doing it wrong. The path to a moonshot begins with duct tape (but actually, you should use Gaffer’s tape).

Google Glass began with a backpack and some safety goggles.

X’s Project Loon, which aimed to beam the internet from high-altitude balloons, started with a foam cooler tied to a weather balloon.

And Mineral, our moonshot to dramatically increase our understanding of plant life, began with two bicycles and a few cameras bolted together.

We describe these prototypes as “V0.crap”—minimum viable prototypes that you can make to learn what you need as fast as possible. Something that works well enough, just long enough, for you to test something out.

Don’t be distracted by the shiny new products that line store shelves. They all started out half-baked on somebody’s workbench.

Don't Reinvent the Wheel

Phil Yee, one of X’s longtime engineers, starts every new hardware prototyping project the same way: with a trip to Home Depot and an electronics store. Billions of dollars of development went into the products that sit on those shelves, and you should be the beneficiary of their hard work. At X, we’ve bootstrapped prototypes from chicken fryers, dog toys and even a remote-controlled toy tank. Only invent the things that need inventing. Everything else you should hack together and go.

Prototyping with AI

Beyond vibecoding, AI tools have foundationally changed how we prototype both in the digital and the physical world, in non-obvious ways.

Often the questions we want to answer with our prototypes involve learning about how humans will interact with them, use them or break them. But learning from humans takes a lot of time and planning. What if you could generate a few hundred users (personas) and let them give feedback on your prototype in a few hours? You can do that with AI. You can generate a new app or software tool with a few prompts in Google Anti Gravity, generate some personas in Gemini and then have them give you feedback on the prototype.

Is the feedback as good as feedback you will get from real people in the real world? No. But it can teach you something fast, which is the goal.

I use AI to generate prototypes, but also as a prototyping partner. It has become as integral to my process as a sketchpad, calipers or CAD tools. Asking Gemini important questions—What supplier has this weird material I need? What academic labs are doing bleeding edge research on this tech? I have these electrical components, how can I solve this problem with them?—allows me to focus on the aspect of prototyping that brings me joy and helps me learn.

Prototypes are not Products

A prototype is something you build to answer a question. Ideally, you’d build one prototype per question, since trying to create one that answers them all will only bog you down and likely fail. Products, on the other hand, are what you build and sell once you’ve answered all of your questions.

That may seem obvious, but I’ve often seen people and teams try to “jump scales”: to take something that works on the lab bench and scale it into production with few changes. Whether it’s a process invented in a chemistry lab, or the first prototype of a SaaS play, it’s easy to get excited when a concept in your head starts to work in the real world. That is a magical feeling. But that feeling is just the start of the journey, not the end.

If you’re interested in more prototyping tips, check out our MasterClass Course on Breakthrough Innovation Strategy. In this course Joe and others from The Moonshot Factory share the tools and frameworks we use to drive innovation and turn big ideas into real-world impact.