A letter from our Founder and CEO
Some of the most important code in the world wasn’t written last year. It was written over decades, by engineers who understood something hard and wrote it down precisely enough for a machine to carry it out. Flight dynamics, grid balancing, signal processing, braking systems. That knowledge still works.
Most of the time, no one notices it. Until it stops. On November 13, the FAA halted departing flights across the United States after a communications failure at air traffic control centers. For a few hours, the whole country saw what usually stays invisible: a handful of systems hold everything else up.
Some software is allowed to fail. This software isn’t. So for years, the safest move was to leave it alone.
But the world around it is changing faster than ever.
AI has made code easier to produce than at any point in history. What once took weeks can now take minutes. But more code doesn’t necessarily mean better code. Bugs, security flaws, and unintended behavior still slip through. And when software can be generated faster than humans can review it, the traditional ways we establish trust begin to break down.
At the same time, AI is giving bad actors new capabilities. Vulnerabilities can be discovered and exploited faster. Systems that have operated reliably for decades face threats they were never designed to withstand.
The cost of standing still is growing. So is the risk of moving forward.
We can no longer afford to leave critical software untouched. And we cannot afford to change it without confidence.
For decades, computer scientists have worked on a solution to this problem. Formal methods use mathematics to reason about software and establish rigorous guarantees about its behavior. The science is established, but applying it to complex, real-world systems has historically been slow, expensive, and difficult to scale.
Now AI is changing what’s possible. For the first time, we have the opportunity to combine the speed of AI with the rigor of formal methods. To automate engineering work that once required enormous human effort, while independently verifying that the results preserve the behavior we depend on.
We believe these two advances belong together. And that’s why we built Code Metal.
We start with the code engineers have already tested, certified, and trusted for years. We take that trusted context and use AI to translate, modernize, and optimize it for new languages and hardware. And we apply independent verification to establish evidence that the behavior that matters survives the transformation.
This means giving organizations the ability to move at the speed of AI without sacrificing the confidence their systems demand.
That takes a particular kind of team. Ours has spent careers in compilers, formal methods, hardware, and mission-critical systems. They know what it means when software can’t fail, because they’ve built software that couldn’t. Some hold security clearances to work on the most sensitive of these systems. They’ve chosen some of the hardest and least visible problems in engineering, because the consequences of getting them right reach far beyond the code itself.
Our mission: make AI trustworthy for systems that can’t fail.
We made a film about the work we do and the missions we serve. Planes that land. Robots that hold steady. Cars that stop. Every day, because someone got the code right.
The code changes. The mission doesn’t. That’s Work of Consequence.
Peter Moralesco-founder and CEO