AI-guided repair for complex machinery

Describe the fault. MACH-AI finds the part, plans the access, and guides the repair.

It reads a machine's 3D CAD, bill of materials, technical manuals and service history — so a technician can ask in plain language and get a step-by-step, validated repair.

How it works See what it does

Runs on-premise or air-gapped · works with STEP, glTF and your PLM exports · built for defense, aerospace and heavy industry

Built for Aerospace MRONaval & shipbuildingSubmarines Defense vehiclesMining & heavy equipmentOEM service teams

Try it

Ask like you'd ask a colleague.

Pick a fault. MACH-AI matches it to the subsystem and component, then builds the access and repair path.

What it does

One model of the machine, six things it can tell you.

MACH-AI connects the engineering record into a single machine model, then reasons across it.

Knows the machine

CAD geometry, BOM lines, manuals and service history — linked. What a part is, where it sits, what it touches.

Finds the component

Matches a plain-language symptom to the subsystem and the exact part, and shows the reasoning with its sources.

Plans the access

Works out what has to come off to reach it, in what order, with time and tool per layer — from the assembly's own dependencies.

Guides the hands-on work

Step by step against the real 3D model — tool, caution, torque — with the part highlighted and the motion shown.

Validates the fix

Checks post-repair readings against the service limits and returns a confidence figure with the maths behind it.

Learns every repair

Each completed job becomes a structured, machine-readable task plan — for the next technician, or an autonomous cell.

The product

From a sentence to a signed-off repair.

How it works

Three steps to a machine that can guide its own repair.

Register the machine

Attach the CAD, BOM, manuals and service exports. MACH-AI parses the geometry, links the BOM to it, extracts the procedures, and builds the component graph.

Describe a fault

A technician types or speaks what they see. MACH-AI matches it to the subsystem and component, and shows how it got there.

Follow the guided repair

Access plan, step-by-step guidance on the 3D model, then a validated result — logged to the maintenance record.

Why it matters

Maintenance moves from "read the manual and figure it out" to "ask, see the machine, follow the procedure."

Every repair taught to MACH-AI today becomes part of a growing machine-intelligence layer — the foundation for increasingly autonomous physical action tomorrow.

−58%

mean time to repair on the demo unit (195 → 82 min)

94%

post-repair confidence, computed from the check margins

1,240

components linked across 4 subsystems from one CAD + BOM ingest

The model

Understand. Reason. Guide. Act.

Each stage builds on the last. The first three help a maintenance team today; the fourth is what that same understanding enables next.

Understand

Structure the machine — components, geometry, relationships and function.

Available today

Reason

Connect the failure, access constraints and procedure into a repair plan.

Available today

Guide

Step-by-step assistance for the technician, with a confidence check.

Available today

Act

Executable tasks for an autonomous maintenance system.

Industries

Built for complex, high-consequence machinery.

AerospaceAircraft and engine MRO teams.

Naval & shipbuildingComplex ship systems and dockyard maintenance.

SubmarinesConstrained environments, limited specialist access.

Army & defense vehiclesTanks, armour, artillery and field systems.

Mining & heavy equipmentRemote maintenance where OEM expertise is hard to reach.

OEMsIntelligent maintenance for internal teams and MRO partners.

See MACH-AI work a real fault end to end.

How it works