Prove It Before the Building Sees It

Automatry Core is our next-generation, deterministic controls-execution engine — the longer-term execution layer for our agentic direction. It gives engineers and AI agents a structured way to create controls code, exercise it across operating conditions and deliver an exact signed programme.

The goal is controls authored at software speed and executed with the consistency buildings require.

Start with readable source

We've selected TypeScript — a language agents understand well, with a clear route into Core's deterministic execution engine.

Core uses a focused controls contract around the language: known inputs, declared outputs, bounded resources and explicit timing. The source remains human-readable and diffable, so an engineer can see what changed between versions and an agent can work on the same programme without translating it into a private format.

The delivery bundle keeps source, build identity and permissions together. That gives the programme a clear route from authoring to operation.

The Builder carries the workflow

The Core Builder can take a project from context and source through verification, compilation, simulation and signing. An AI agent can drive that pipeline while each stage remains visible to the engineer.

Verification checks the programme against its controls contract. The build prepares it for a deterministic execution engine — the same engine that runs on the gateway. Simulation exercises the programme against defined inputs and expected behaviour. Signing identifies the exact artefact approved for delivery.

The same structure makes changes easy to follow. A new source version produces a new build identity and a fresh set of simulation results, keeping the programme, its scenarios and its delivery bundle aligned.

Test demanding conditions on purpose

Core's simulator uses the same execution engine with a virtual clock. It can move through hours of operating conditions quickly while retaining deterministic inputs and outputs. Engineers and agents can exercise the conditions that define dependable control:

What happens when the main sensor fails? What happens when communications disappear? Does equipment restart safely after a power cycle? Does the fire-alarm interlock take authority immediately? Are repeated starts prevented when plant proof is missing?

Each scenario records the programme version, inputs, outputs and any suppressed or rejected write. That gives fast AI-assisted development a precise feedback loop and lets engineers concentrate on operating intent, safety behaviour and the unusual cases that deserve attention.

From the Builder into the field

The Builder and lab testing are proving out well; field adoption is the next step. We're looking forward to deploying our take on what will help the industry.

Core's deterministic engine, signed delivery model and gateway integration are designed to make that adoption repeatable. The same programme can be simulated quickly, admitted under a clear policy and understood later through its recorded inputs and outputs.

A stronger controls toolchain

Core makes control programmes easier to create, test, understand and improve. AI can help an engineer turn a sequence into source, construct scenarios and explain results. Deterministic execution makes behaviour repeatable and gives incident replay a consistent foundation.

This changes where engineering time goes. Engineers can spend more of it defining the operating contract, reviewing demanding conditions and improving how the building behaves.

The ambition is straightforward: combine the speed of modern software development with the dependable execution expected from building controls.