First Principle
First Principle
We are five people working together in Dubai, United Arab Emirates. We care about AI agents, robotics, manufacturing, human-machine integration, nanotechnology, and new ways to compute. Some of this is active work now. Some of it is where we are building enough depth to work seriously later.
Current product
It is for people trying to reach a real outcome while the boundary between what they should do and what machines can do keeps changing.
The system keeps a live model of what the person knows and can do, what current AI models and tools can do, the outcome, the deadline, and the constraints. It builds a path, delegates executable work to agents and tools, gives the person learning or practice when human capability is the bottleneck, checks what actually happened, and updates the next step.
We are using it ourselves and building the first usable version. Its name is not public yet.
Research
We work on agents that can inspect their tools, build and modify workflows, coordinate specialized agents, keep state across long tasks, and check their output against explicit requirements. A central problem is keeping the system useful while the models and tools beneath it keep changing.
We are building depth in dynamics, control, robot learning, and simulation. We want to understand how learned models can control real systems when sensors are noisy, actions are delayed, the world is only partly known, and mistakes have physical cost.
We are interested in reducing the time between designing something, simulating it, running an experiment, fabricating it, and measuring what happened. This includes the software, robotics, and production systems that could let small teams build serious physical technology.
As machines take over more reasoning and action, we want to understand how work, learning, decisions, and control should be divided between the human and the machine. Longer term, this also includes closer physical integration between both.
We want to work toward atomically precise engineering, molecular machines, and nanoscale robots. The real problems include how these systems are designed, powered, controlled, manufactured, and made reliable enough to operate outside a laboratory.
We are interested in computation beyond the current chip stack and in physical structures that can sense, adapt, repair, or reproduce. This is longer-term work, and we are still developing the physics, materials, and engineering foundations needed to approach it seriously.
Public work
We are working toward our first technical reports and useful open-source contributions. We will put the actual papers, code and results here when they exist.
Who we are
First Principle is the group through which we build products, run research, and develop the technical depth needed to take on harder problems.
We are early. One product is being built now, our work on AI agents is active, and the physical fields require months of learning and experimentation.
For research, collaboration or engineering work, email mahdi.ettehad85@gmail.com.