Limitless Labs Raises $20M to Put AI Agents Inside CAD and CNC Programming

CNC programming is one of manufacturing’s most persistent skilled-labor bottlenecks. Limitless Labs has raised $20M to automate it — not with traditional CAM rule sets, but with AI agents trained on the physics of metal cutting.

What Happened

Limitless Labs has closed a $20 million Series A, bringing total funding to $27.3 million. The round was led by Dell Technologies Capital with participation from Square Peg, Grove Ventures, Meron Capital, and Kinetica. The company’s platform integrates natively with Creo, Siemens NX, and Mastercam — the dominant CAD and CAM tools in aerospace, defense, and industrial manufacturing.

The core product is a Physical AI Foundation Model trained on a combination of metal cutting physics, CAD geometry, and real machine constraints. The CAM Agent built on this model recommends tooling selections and generates complete toolpaths autonomously from part geometry and material specifications. Early pilots at Blue Origin and Sandvik report 50% reductions in CNC programming workload, per the company. The platform is ITAR-compliant for defense applications.

Why It Matters

CNC programming is a domain expertise problem. A skilled CNC programmer reads a CAD model, understands the material behavior, selects cutting tools from a library of thousands, sequences operations to minimize fixturing changes, and generates toolpaths that don’t crash the machine or violate quality tolerances. This process can take days for complex parts and requires experience that takes years to develop. The shortage of experienced CNC programmers is a recognized manufacturing capacity constraint.

Existing CAM automation — parametric templates, feature-based machining in tools like Mastercam — reduces some of that burden but still requires a programmer to configure inputs and verify outputs. What Limitless Labs is attempting is qualitatively different: an agent that understands enough of the underlying physics to make autonomous recommendations rather than applying rule templates.

The training data approach is the key technical bet. Most AI coding or design tools fail in manufacturing contexts because they’re trained on text and code rather than on physical constraints — toolpath strategies that work in simulation can violate material properties or machine dynamics. Training on actual metal cutting physics and real machine constraint data is meant to close that gap.

Implications

The choice of pilot partners is deliberate. Blue Origin operates precision manufacturing at aerospace tolerances; Sandvik is one of the world’s largest cutting tool manufacturers. Both have deep domain expertise and would be able to validate whether the 50% workload reduction claim reflects genuine automation versus shifted work. A sustained relationship with either suggests the model is producing outputs that meet professional review.

ITAR compliance is a prerequisite for meaningful defense and aerospace penetration, and it’s unusual to achieve this early in a startup’s lifecycle. It signals that the founding team is building for regulated industrial markets rather than consumer or general manufacturing.

The broader shift toward AI-integrated CAD and design tools has been more about generative geometry than manufacturing process automation. Limitless Labs is working the other side of that problem — not “how do you design for additive” but “how do you program the machining of what you’ve designed.”

Content created with AI assistance and reviewed for accuracy.