AI-native design-to-manufacture
ForgeLab puts a Design, DFM, Materials, Cost and Matching agent on your brief. They critique against a real process rulebook, revise the geometry in the open, and price it with a parametric model — you approve production-ready before anything is treated as manufacturable.
Free · no account · real DFM rules, not a guess
Brief in · revision pack out · ~4 minutes · under a $4 cap per run
Manufacturing feedback usually arrives after tooling assumptions are already wrong. A render isn't a part. “Looks printable” isn't a wall-thickness rule with a number attached.
How it works
01
Design
3-5 candidate concepts from your brief, checked against the process rulebook before you see them.
02
DFM
Every stated dimension run through check_feature — wall, overhang, radius, holes — with the rule cited.
03
Iterate
A blocking or major issue goes back to Design, which applies the smallest fix that clears the rule, then re-checks.
04
Materials & cost
A ranked shortlist and a parametric unit-cost range with a full breakdown — never a firm quote.
05
Matching
Pilot manufacturers ranked against process, material, MOQ and envelope — refused outright past a blocking issue.
06
Gate
Production-ready, release, and export are blocked until a human approves. ForgeLab never ships.
The roster
Design
Brief → candidates
Generates 3-5 manufacturable concepts and revises the geometry when DFM sends one back.
DFM
Rule-grounded critique
Critiques the selected concept against the process rulebook — FDM, SLA, or 3-axis CNC — never a guess.
Material Advisor
Shortlist + trade-offs
Ranks materials for the process against service temperature and toughness, and names what to avoid.
Cost Estimator
Parametric range
Machine time from geometry and volume, material by mass, amortised setup, finishing — with the breakdown shown.
Matching
Pilot manufacturers
Ranks ForgeLab's pilot-shop directory by process, material, MOQ and envelope, and explains every match.
Outcome Logger
Learn from the result
Records a confirmed production outcome and writes reusable lessons to the design's provenance log.
Human authority
Marking a design production-ready, releasing an order package, sharing outside the org, and any IP or licensing change all stay with a person. ForgeLab prepares the revision pack and the RFQ — it doesn't send either.
Why you can trust the output
Grounded
DFM findings cite a rule and a measured value. Upload an STL and the wall thickness and overhang angle come from the mesh, not a description.
Costed
Every run has a hard $4 cap. A live token meter shows exactly what each agent step spent.
On the record
Typed messages between agents, a full audit log, a per-step inference trace, and a labelled eval suite scoring the DFM verdict against ground truth.