About RunMat
Every generation of engineers has had a tool for doing math. The abacus. The slide rule. The graphing calculator. MATLAB/Fortran.
Each one made calculation faster, enabled larger scale problems, and put it in more hands. Each one has, as a result, enabled new engineering breakthroughs.
At RunMat, our goal is to build the best tool for running math: the best calculator of our generation. If you use computers to run math, we're building RunMat for you.
Our north star is simple: make running math fast, accessible, and collaborative for everyone who uses computers to run math.
Math per second
Calculation speed has always set the ceiling on what we can build. Ancient Egyptian builders used mechanistic movements with knotted ropes to make makeshift calculators, managing a few operations a minute to do their calculations. With that, they raised the pyramids. Every advance since, from mechanistic tools like the Abacus, to the advent of micro-processors and personal computers, unlocked a new tier of engineering.
The latest set of advances are the largest: dedicated math hardware (GPUs), modern ML models, and large human teams spread out over the earth. GPUs in particular are now in every laptop and desktop and can execute orders of magnitude more computations per second, yet most compilers and runtimes were built for the CPU and treat the GPU as an afterthought; the result is that practically, GPU speeds are out of reach for typical engineering calculations.
RunMat treats the GPU as a first-class compute device, detecting independent chains of work and dispatching them in parallel to saturate your hardware. The result is the same math, running orders of magnitude faster.
We believe computation speed is the single biggest constraint on what we can build. We want to help you push the limits of what you can do with math.
What RunMat is today
Runtime. At the core is an open-source runtime written in Rust. It executes MATLAB syntax code with a fast-starting virtual machine and a JIT compiler, while an acceleration engine fuses array operations into GPU kernels and keeps data resident on the device between them. The result is math written in simple MATLAB syntax, running blazing fast on Apple, NVIDIA, AMD, and Intel GPUs. It ships with a large standard library and a GPU-accelerated 2D and 3D plotting engine, and it's released under the Apache 2.0 license.
Desktop. RunMat Desktop turns the runtime into a daily driver: a MATLAB-aware editor and language server, interactive GPU-rendered 2D and 3D plots, a live workspace inspector, project history and collaboration, and an agent harness that works directly with your files, plots, variables, and runtime. It runs natively on macOS, Linux, and Windows or entirely in the browser, where you can work in a Temporary Sandbox or save projects to an organization.
Collaboration. Serious math is rarely a solo effort, so RunMat is built for working together. Projects are shared workspaces: when a teammate saves a file, you see the change in real time. Every save is versioned automatically, recording who changed what and when, and a single click captures a snapshot of your whole project so you can always return to a known-good result. There's no version-control ceremony, no "which copy is current," and no emailing files around.
You can dig into how all of this works in the documentation.
Where we're going
RunMat is a platform for running math, and we're building it out in every direction.
We want running math to be effortless at any scale: the same code that runs on your laptop should run on a far bigger machine the moment you need more power, and eventually on the very hardware you're designing for. Scaling up and down should be effortless. We want the hardest parts of high-fidelity work, like simulation and math over real geometry (aka FEA), to feel approachable.
We want to allow anyone to simulate reality in high fidelity, and help you push the limits of what you can do with math.
The team
Nabeel Allana
Self-taught programmer since age seven and a Waterloo Mechatronics Engineering grad. Before RunMat, Nabeel was a technical lead on Apple's autonomous-vehicle program and has worked across engineering teams at Apple Product Design, Toyota Manufacturing, and BlackBerry.
Julie Ruiz
Julie co-founded the company behind RunMat and leads the work that turns RunMat into a product people can find, understand, and adopt, drawing on a background in marketing, communications, and strategic accounts.
Fin Watterson
Fin is a technical marketer and storyteller who has spent more than a decade bringing engineering software, industrial technology, and deep-tech products to market across startups, scaleups, and global manufacturing companies.
Company
RunMat is made by Dystr Inc. We're a US based company backed by some of the best investors in Silicon Valley.
Get in touch
You can get in touch with us by emailing team@runmat.com or filling out this form.