Every major industry built a digital brain.
Metals never did. We're changing that.
METALLAI turns weeks of metallurgical trial-and-error into overnight answers. Physics-aware AI for alloy prediction and inverse design — so engineers stop guessing and start designing, on the first try.
to hit target properties
with quantum-hybrid AutoForge
validated 42CrMo4 case
& rework globally
Pharma has computational discovery. Semiconductors have process simulation.
Metals still run on trial & error.
Every alloy, every heat, every process window — still figured out one melt at a time. Decades of metallurgical intuition are retiring out of the industry with nowhere to go.
— The Cost of Trial & Error
— What METALLAI Changes
- Multiple trials to hit spec
- Each off-spec heat → scrapped
- Weeks of delay per order
- Relies on senior metallurgist intuition — no digital record
- 1–2 high-confidence trials
- Off-spec heats rescued by adjusting heat treatment
- Predictions in seconds
- Expertise captured as computable knowledge
Three modules. One intelligence layer.
From instant property prediction in your browser to autonomous alloy design overnight — METALLAI is built for every stage of the metallurgical workflow.
Web Predictor
Composition + process → mechanical properties in seconds. Right in your browser. No installation. Confidence bounds included.
- Yield, UTS, hardness & elongation
- Steel, stainless, Al, Ti, Ni & Cu alloys
- Confidence intervals & nearest-alloy explainability
- Metallurgical warnings (CE, Pcm, weldability)
Inverse Design & Process Optimization
Flip the question. Give AutoForge your target properties and get back feasible composition + process routes — with trade-offs made explicit.
- Target-driven process recipes
- Pareto-ranked candidates
- Fatigue & weldability gated in
- 17× faster via quantum-hybrid search
R&D Acceleration
Virtual experiments at the scale of a supercomputer, the speed of a browser tab. Explore compositions that don't exist yet — physics-aware and uncertainty-bounded.
- Millions of virtual experiments
- Physics-aware candidate validation
- Trade-off landscape visualization
- Faster time-to-certification
Give us the targets.
Get back a family of feasible routes.
Engineers don't ask "what are the properties?" — they ask "how do I achieve them?" AutoForge inverts the problem: from target window to composition + process, in seconds.
Define Targets
Set yield, UTS, elongation, hardness — plus composition & process constraints.
Virtual Exploration
Quantum-hybrid search scans millions of composition + process combinations autonomously.
Physics Validation
Every candidate gated by thermodynamic, kinetic & weldability checks. No hallucinations.
Pareto-Ranked Results
A family of feasible routes — balanced, aggressive, cost-optimal. You pick the trade-off.
From our LinkedIn series:
real alloys, real labs, real numbers.
Two case studies showing METALLAI in action — one for forward prediction, one for inverse design. Both experimentally validated or benchmarked against industry practice.
Can AI Reduce Alloy Development to a Single Trial?
Target window: 285–340 HB hardness, 880–1080 MPa UTS. Traditional approach would need 5–10 iterations.
Same Alloy. Same Composition. Different Performance.
Targets: 1150 MPa YS, 1300 MPa UTS, 13% elongation. AutoForge returned two legitimate process routes — balanced vs. aggressive.
Strength. Manufacturability. Reliability.
All on the table — at the same time.
METALLAI doesn't just predict numbers. It surfaces the three-way trade-off that every metallurgist knows exists but rarely sees quantified.
Fewer experiments
To hit a target property window — validated on 42CrMo4 and benchmarked on 300M aerospace steel.
Faster design search
Quantum-hybrid annealing reaches the same Pareto front in 92 s vs. 1,611 s classically.
Prediction accuracy
Validated on 100+ data points at a major Turkish metal manufacturer's R&D facility.
Can we design alloys that don't exist yet?
Not magic — the same physics-aware search, pointed at a blank slate instead of a known grade.
Predict & Optimize
Hit a target window inside a known alloy family. Composition + process → properties.
Inverse Design
Feasible composition + process routes — with fatigue, weldability & uncertainty gated in.
Novel Alloys
The inverse engine proposes compositions outside the catalogue — from targets alone.
From prediction to design — bridging physics, data, and decisions.
Backed by & built with
See METALLAI on your own alloy.
Launch the browser predictor, or request a tailored demo with our team.
