From JuliaCon 2025: How Dyad Is Bridging Software and Engineering to Build the Next Generation of Physical Systems
What happens when code meets CAD? When software engineers team up with hardware engineers? When digital twins aren’t just simulations but self-improving, adaptive systems?
You get Dyad.
Unveiled at JuliaCon 2025, Dyad is a next-generation modeling environment built on Julia that merges physics, data, and code into a single platform. More than just a tool, Dyad represents a design philosophy, one that reimagines how engineers create everything from electric vehicles to spacecraft.
Watch the full session here:
The word dyad means ‘something that consists of two elements or parts’ and that duality is built into the Dyad DNA.
With Dyad:
For decades, these groups have worked side-by-side but often in separate environments. Dyad’s aim is to give them one place to meet. It bridges the gap between disciplines, allowing advanced simulation tools to be accessible to people who don’t want to write code all day — while still giving full power to those who do.
One of Dyad's core innovations is its bidirectional interface. You can build a system diagram visually; drag a motor here, a controller there, wire up the feedback loop. Instantly, Dyad generates the Julia code for that exact model.
Want to tweak the code? Change a parameter or add a term to the equation? As soon as you save, the GUI updates to reflect your edits.
This makes Dyad feel natural to both kinds of engineers; those who think in circuits and mechanical linkages, and those who think in algorithms and source files.
Dyad sits on top of Julia’s powerful DifferentialEquations.jl ecosystem, so it can represent the kinds of complex physics you find in real-world systems, from ODEs to PDEs.
But the real magic is what happens after you build a model:
It’s not only building a digital twin but also making a digital twin that learns, adapts, and improves over time.
Dyad is already being used to model things like a NASA HL-20 lifting body, active suspension systems, and electric vehicle drivetrains, all with realistic degradation over years of operation.
That means you can ask questions like: What will the range of this EV be after a decade of use? And get an answer based on a high-fidelity simulation, not a back-of-the-envelope guess.
Even more exciting: Dyad models aren’t trapped inside the design software. They can be deployed directly into embedded systems that run inside vehicles, aircraft, and industrial machines.
Dyad enables software-defined physical products, where performance improves over time.
Can a ten thousand dollar vehicle feel like a Porsche because of the software running it? With Dyad, it's possible.
In a Dyad world, a product isn’t frozen the day it leaves the factory. Its digital twin keeps learning, the control software keeps improving, and the performance keeps getting better.
Dyad wouldn’t exist without the Julia ecosystem. Every package, solver, and library that’s been built over the years is part of what makes this possible. And Dyad gives something back — a way for any Julia package to be pulled straight into a modeling workflow and run in production systems.
If you’ve built a Julia package, Dyad can directly interface with it, meaning your work could be driving the next generation of drones, EVs, or robotics.
Dyad isn’t just about merging software and engineering. It’s about bridging the gap between design and operation, between simulation and reality, between the physical and the digital.
It’s about building systems that get better over time, and giving engineers, scientists, and developers a shared language to make it happen.
The future of engineering design is here. And it’s written in Julia.