NEWIntroducing MADES v1.0

The World Model for everything that is made.

From a sketch to a shipping part, MADES is the first industrial foundation model that understands geometry, physics, materials, machines, cost and manufacturing constraints — end-to-end.

7
Core Model Families
6
Industrial Modalities
100ms
DFM Latency
99.2%
Feature Recognition
LIVE · MADES CONSOLE
v1.0.2
Part under analysis
Bracket-Ti-047
Process
5-axis CNC · Ti-6Al-4V
DFM score
98.7 / 100

Trusted by pioneers in advanced manufacturing & research

Nanyang Technological University
Singapore Economic Development Board
Advanced Remanufacturing and Technology Centre
National Additive Manufacturing Centre
Agency for Science, Technology and Research
Singapore Manufacturing Federation
Beyond CAD Copilots

MADES is not a CAD copilot.
It is an Industrial World Model.

Traditional copilots merely autocomplete the screen. MADES builds a latent understanding of the physical world — how forces travel through steel, how a tool cuts aluminum, where a mold will fail — so it can design, plan, simulate and manufacture at a level of reasoning previously reserved for teams of senior engineers.

MADES understands

Geometry
Physics
Process
Materials
Machines
Cost
Constraints
1
Understand products
READY
2
Design products
READY
3
Make products
READY
4
Optimize products
READY
PHYSICS ENGINE
Real-time sim
Six-Layer Architecture

One model. Six reasoning layers.

From human language to machined part — every reasoning modality a senior engineer uses, distilled into a single foundation model.

Layer 01

Industrial Language

Multimodal semantic understanding

Requirements parsingIndustrial semantic groundingEngineering parameter extractionKnowledge-grounded reasoning
Layer 02
CORE

Industrial Geometry

Deep B-Rep understanding

Topology graph generationParameter graph constructionFeature recognition across 30+ primitivesConstraint graph learning+1 more
Layer 03

Industrial Physics

Materials & mechanical reasoning

Steel, aluminum, titanium, polymers, compositesStress, thermal deformation, vibration, fatigueCutting, melting, solidification, forming dynamics
Layer 04

Manufacturing Reasoning

DFM · Process planning · Cost

DFM: risk, cost, tolerance diagnostics & redesign hintsProcess routing: machines, fixtures, tools, sequenceCost reasoning: material, machining, finishing, logistics
Layer 05

Manufacturing Execution

CNC program generation & CAM

Automatic tool selectionToolpath generationCollision-free sequencingMachining simulation
OUTPUTS · CNC programs · G-code · M-code · Toolpaths
Layer 06

Design Generation

End-to-end industrial design

"Design a drone arm optimized for 2kg payload, CNC-machined from 7075 aluminum."

Concept generation from requirementsParametric geometry synthesisMaterial & process co-selectionEnd-to-end manufacturability assurance
OUTPUTS · Concept · Parameters · Materials · 3D model · Process plan · Quote
End-to-end from requirement → geometry → physics → process → machine → outcome
Capability Matrix

Built for the physical world — not just pixels.

Text-first AI breaks down when geometry, physics and tolerances matter. MADES models are trained from the ground up on CAD, CAE, CAM, and process data.

Capability
Generic LLM
CAD Software
CAM Software
MADES
Understand natural language
Understand 2D drawings
Understand 3D CAD models
DFM & manufacturability analysis
Process planning
Engineering cost estimation
Generate ready-to-run G-code
End-to-end new product design

✓ Full native support · ○ Limited / plugin · ✗ Unsupported

Foundation Models

A constellation of specialist models.

MADES isn't one monolithic LLM. It is a carefully orchestrated fleet of foundation models — each built for a distinct industrial modality — composed into one world model.

Industrial LLM

70B – 200B

Language, reasoning, process, knowledge

General intelligence foundation for industry.

Inputs
Natural languageTechnical documentsProcess manuals
Outputs
Process plansReasoned reportsTechnical briefs
MODEL_v1

Geometry Model

30B (custom)

CAD understanding & generation

The "Geometry GPT" — reads and writes 3D.

Inputs
B-Rep graphsPoint cloudsMeshes
Outputs
STEPParasolidFeature graphs
MODEL_v1

Industrial Vision

12B vision-language

Drawings, PDFs, sketches

The eyes of the industrial stack.

Inputs
Engineering drawingsPDFsHand sketches
Outputs
Structured parametersGeometric features
MODEL_v1

Planning Model

Decision-transformer

Process & NC planning

The planning network of the factory floor.

Inputs
Part modelsManufacturing specs
Outputs
Process routesNC programs
MODEL_v1

Simulation Model

Multi-physics latent

Physics & cost prediction

Industrial-grade physical simulation in milliseconds.

Inputs
Design variantsMaterial parameters
Outputs
FEM resultsCost estimatesDefect risk
MODEL_v1
+

More models soon

Robotics · Inspection · Quality

See research roadmap
Industry Scenarios

Built for the industries that build the world.

From titanium airframe brackets to medical implants — MADES is shipping in production across six of the world's most demanding manufacturing verticals.

Aerospace

Intelligent design and manufacturing optimization for complex structural components in titanium and high-temperature alloys.

SHIPPINGv1.0

Automotive

DFM analysis and cost optimization across full vehicle part portfolios, from powertrain to chassis and body-in-white.

SHIPPINGv1.0

Mold & Tooling

Automated mold design, gate placement, cooling circuit optimization and process planning for high-volume tooling.

SHIPPINGv1.0

Consumer Electronics

Rapid design iteration and instant quoting for precision structural parts — from concept to DFM in minutes.

SHIPPINGv1.0

Industrial Machinery

Intelligent production of general mechanical parts with adaptive process planning across mixed CNC fleets.

SHIPPINGv1.0

Medical Devices

High-precision design and manufacturing for surgical instruments, implants and regulated medical components.

SHIPPINGv1.0
M
AIMADES · HQ SINGAPORE1.3521°N · 103.8198°E
About AIMADES

Born from Nanyang Technological University.

AIMADES is an AI research company headquartered in Singapore, staffed by a core team from NTU with decades of combined industrial domain expertise and leadership experience at top global AI labs. Our mission is to build the cognitive layer for the world's physical production.

NTU
Core Research Roots
10+
Years of Industrial Depth
3
Continents Served
24/7
Engineering Partner Network

Begin the next intelligent manufacturing. of intelligent manufacturing.

Experience MADES today. From first sketch to production-ready G-code, in one model.

No credit card requiredFree tier availableSOC2-ready infrastructure