Founder’s note

Software has much left to explore, and boundless room for innovation. Yet we mostly treat its shape as settled: files and folders, editors and terminals, and apps that work about the same as they did 20, or even 50, years ago.

Software’s basic pieces, like buttons, lists, canvases, terminals, visuals, sounds and performance, even the hardware underneath, have barely been explored.

Think of a calendar. You can view your appointments as a long scrollable list or as a grid, and both hold the same appointments, but we all intuitively know the grid is better. The calendar-as-grid lets you take in a whole month at a glance and move a meeting with a simple drag. Better ways of working like that are waiting to be found almost everywhere we look.

Software should also be a joy to use: beautiful, powerful, simple and fast. You might finish a day’s work tired, but you should finish it satisfied. Unfortunately, software today often leaves us frustrated or, increasingly, burnt out.

I founded Isomorphic explicitly as a “deep tech software” company, because finding those better ways of working with computers takes what deep tech takes in any field: advanced engineering, R&D and first-principles thinking. Only instead of applying them to build complex physical machines, we apply them to reimagining the software primitives of today and to developing new ones.

However, you can’t reimagine all of software at once, so we started with one domain: building backends. We asked what it would mean for a backend’s architecture to be the interface you develop it through. People and AI could then look at a whole system, from its production infrastructure to the local development environment, and understand and change it as a single thing. We are trying to turn the mental models developers carry in their heads into something you can see, touch and manipulate, because more intuitive and powerful interfaces are what will let the next generation of software be built.

Hold on to the picture that paragraph just put in your head, and compare it with a pile of files or an AI chat window. You’ll quickly realize how anemic our current software interfaces are.

Isomorphic is our current answer. It has barely scratched the surface of what we believe backend development can be, yet it already makes agents clearly faster, cheaper and more capable, to the point that a small model can now do production work. At the same time, it gives the developer a better experience, more control, and a clear understanding of the system they are building.

We love backends. They are critical systems that hold surprising depth and complexity, but they are also an intentional starting point. What we learn and build while reimagining backend development gives us the raw materials that, once understood deeply, will insha Allah let us extract the general principles underneath how people, software and AI interact, and underneath the nature of software and of complex systems.

That is the real goal behind Isomorphic: a new way to think about, build and interact with software, one that can represent all of it while giving us a new perspective from which to design and use it. That perspective is the key to entirely new software primitives, and from there to new computer hardware and new software experiences. In other words, Isomorphic is looking for a new interface for software.

Goals this ambitious are what make deep tech what it is, and progress toward them is often measured in milestones. A milestone is a noteworthy step toward the goal that sends a strong signal to the market and makes the next milestone possible. In hardware a milestone can be a rocket engine on a test stand, but software milestones are often invisible, so it’s important to put effort into showing them.

Isomorphic’s first milestone was a proof-of-concept agent that, on very narrow test cases and in a highly unusual “development environment”, performed much better than the best agent at the time. That led to our pre-seed, on the bet that we could scale this to production backends.

The next milestone is our public release, which includes benchmarks, demos, and our advanced backend development environment. It will show that the benefits do scale to a complete backend, further validate our thesis, and position us for a greater milestone. We are doing all of this with extreme capital efficiency, a trait critical to success in deep tech.

Whether you build, teach, invest in, or use software, I hope this invites you to reflect on the nature of software, where it is today, and what it can be.

— Omar