Built at NSMLab · Sogang University
ForgeIQ · by NSMLab

The forging
workbench
re-imagined.

Three intelligent programs in one workbench — predict cogging energy, map processing windows, and generate optimal 3D preforms in minutes instead of weeks. No data required to get started.

0
Programs
< 0 min
Time to first result
0 MB
Built-in U-Net model

Why this exists

Designing forging schedules
used to take weeks.

Manual calculation of pass schedules, hand-built processing maps, and trial-and-error preform design consume thousands of engineer-hours every year. This platform compresses the loop into seconds.

ML-driven

Pre-trained neural networks predict process outputs (ENE, preform geometry) from your design parameters. No fitting required.

Quick mode

Don't have your own dataset? Every program runs with built-in reference data — enter parameters, get results.

Lab-grade physics

Backed by validated models from the Net Shape Manufacturing Lab — published methods, real-world validated.

New · Auto Pipeline

Define your target.
Press play.
Get a full design pass.

Instead of running each program separately, the Auto Pipeline chains training-data correction, pass-schedule optimisation, and 3D-preform generation in one run — surfacing a single result page you can iterate on like a feedback loop.

  • 1
    WorkpieceYou enter the cross-section, length and cut limit.
  • 2
    TargetYou set the grain-size goal and weight.
  • 3
    Auto-runCorrection → Pass Schedule → 3D Preform (sequential).
  • 4
    EvaluateAggregated KPIs: void closure %, passes, preform Δvolume.
  • 5
    IterateTweak dimensions or target, re-run; previous iterations stay in the table.
pipeline.log
[13:24:01]doneCorrect training data
BQI recomputed → ASTM=6 · 12 KB
[13:24:03]doneCompute pass schedule
7 passes · min void closure 96.1%
[13:24:08]doneGenerate 3D preform
STL ready · Δvolume −1.03%
96.1%
void min
7
passes
target
Iteration #1 · 7.4 s totalRe-run with new target

See it in action

Six views from the platform

From the closed-loop Auto Pipeline to the Attention-U-Net 3D preform — here is what you get the moment you sign in.

Click any image to enlarge.

Live walkthrough

Watch a 62-second demo

A full Auto Pipeline run — from raw cogging Excel to a watertight 3D preform STL — captured end-to-end.

What's inside

A workbench that respects
your time

Quick mode

Run any program with bundled reference data.

History

Every run saved. Compare any two side-by-side.

Bookmarks

Save parameter sets and reapply with one click.

PDF export

One-page report for Pass Schedule results.

Live support

Chat directly with the team — real-time replies.

AI assistant

Quick answers about every feature, in-app.

Trilingual

English · O'zbekcha · 한국어 interface.

Model caching

Heavy U-Net loaded once, reused across requests.

Who built this

NSMLab · Sogang University

The Net Shape Manufacturing Laboratory researches data-driven forging and forming processes. This workbench is the operational front-end for years of published research on cogging optimization, processing-map analysis, and preform prediction.

Research lead

Net Shape Manufacturing Laboratory

Sogang University · Republic of Korea

Engineering & deployment

Platform team

Full-stack web + ML infrastructure

Open a program. See a result.
It really is that fast.

Sample data is bundled with every program — there's nothing to download before your first run.