1a is building AI software for hardware engineering, starting with rocket engines.

Our thesis

Engineering tools have improved enormously.

The workflow around them has changed much less.

A design change still creates work downstream. Models need to be updated. Analyses need to be rebuilt. Results need to be interpreted. The reasoning behind the change needs to be recorded.

We are building 1a to work across that loop.

The engineering loop

Engineers define the requirements, develop the geometry in CAD, prepare the analysis, run simulations, interpret the results, and change the design.

Change one fillet radius and the mesh may need to be regenerated, boundary conditions reapplied to changed faces, and parts of the analysis rebuilt before the next run.

That work is spread across several engineering tools.

What we're building

1a starts from the requirements for a part or system.

It creates and modifies geometry, prepares engineering models, configures and runs simulations, interprets the outputs, and proposes the next design change.

The engineer reviews the work and makes the engineering decisions.

1a also records why the design changed: the requirement behind the decision, the analysis that supports it, and the result that led to it.

If someone asks why a wall is 3 mm thick, the answer should already be there.

The same record can support manufacturing and design review with drawings, dimensions and tolerances, analysis results, and verification evidence.

Why rocket engines

Rocket engines combine fluid flow, combustion, heat transfer, structural loads, manufacturing constraints, and tightly coupled design decisions.

A change in one part can affect several others.

They are a demanding place to build 1a, and they are the hardware we know best.

Why we're doing this

We want to build ambitious hardware faster.

Too much time is lost rebuilding analysis and moving work between engineering tools after every design change.

1a is the software we want to use ourselves.