Hardware Podcasts

How Protolabs turns CAD files into parts in under 24 hours

The Robot Report Podcast · 14 Aug 2026 · 1h 12m

AI in practice: 4 real uses of AI described

What this episode covers

Marc Kermisch, CTO and AI Officer at Protolabs, explains how the company uses AI and simulation to turn CAD files into physical parts in under 24 hours across CNC machining, 3D printing, injection molding, and sheet metal. He details how ProDesk flags manufacturability issues like tight tolerances and ejector locations, while internal AI tools optimize print box placement and visual quality inspection. Kermisch also discusses the critical tradeoffs between prototyping and production, highlighting common design mistakes such as non-uniform wall thickness and missing draft angles that engineers must address before tooling.

AI in practice

Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.

Engineering & simulation

Protolabs uses machine learning models to assess part manufacturability and simulate tool paths, allowing them to generate G-code directly from CAD files without skilled labor programming the machines.

“a lot of it is really around how we've built machine learning models to assess the manufacturability of the part and then to simulate the tool path that is required to mill that part in particular. … We then take that tool path and we generate the what's called the G code, which goes right into the controllers of these mills.”

Marc Kermisch · Hear it at 35:01

Manufacturing

Protolabs uses AI capabilities to optimize 3D print box placement, determining the best part orientation and scaffolding to maximize quality and part counts for cost-efficient printing.

“We've created AI capabilities to help us optimize that part placement and maximize for quality as well as part counts so we can deliver the most cost efficient print experience for the customer.”

Marc Kermisch · Hear it at 39:36

ManufacturingClaude Code, OpenAI Codexspent 40 hours vs hundreds of thousands of dollars for commercial solution

Protolabs engineers used agentic coding tools like Claude Code and OpenAI Codex to build an internal visual quality inspection system, simulating defects and training an AI model to compare parts against CAD designs to determine quality.

“they designed with clawed code, uh, an AI model. They downloaded an open weight model to a laptop. … simulating all the different variants of defects that we could see on particular parts. … using the trained AI model and all the simulated defects that we had created. To determine is that part of quality or not of quality.”

Marc Kermisch · Hear it at 48:51

Manufacturing

Protolabs uses AI coding agents to allow non-programmers to program cobots by describing tasks in natural language, reducing the need for specialized application engineers on every cobot.

“we can apply a traditional AI coding agent, um, where now they don't have to know Python or the machine code to to control that. They can just kind of say, hey, I need the robot to do X, Y, and Z. … Now all of a sudden, I don't need to have a application engineer with that skill set necessarily on every Cobot…”

Marc Kermisch · Hear it at 48:51

Most relevant to

companies scaling up

Topics

Mentioned

Companies
Protolabs, Serve Robotics, Starship, No Tips, Tesla, Locusts
Products
ProDesk
Also
CMMMC, ISO, ITAR, CUI, MJF, CNC, 3D Printing, Injection Molding, Sheet Metal, Cobots, AI

Related episodes from other shows

Other shows on the same ground, closest first.

  • The Builder Circle · 1h 06m

    S2 E5: Rapid Prototyping Deep Dive with Greg Paulsen

    Greg Paulsen, Director of Applications, Engineering and Marketing at Xometry, discusses the complexities of rapid prototyping and the transition to mass production. He details the pros and cons of various 3D printing technologies, including filament-based FDM, powder bed fusion (SLS, MJF), and resin-based methods (SLA, DLS), explaining how each affects design constraints and scalability. Paulsen highlights common hardware horror stories, such as designing for 3D printing without considering injection molding requirements, leading to prohibitive costs and tooling issues. He emphasizes the importance of clear communication with manufacturers and using AI-driven tools to optimize designs for manufacturability early in the process.

    Prototyping & iterationMaterials & manufacturing processes
    8 chapters

  • Being an Engineer · 46m

    Dylan Lundberg | Protolabs

    Dylan Lundberg, Manufacturing and R&D Engineering Manager at Protolabs, explains how the company reinvents manufacturing through automation, software, and standardized processes to deliver prototypes and low-volume production parts via 3D printing, CNC machining, and injection molding. He details the company's evolution from a garage startup by Larry Lucas into a global enterprise with nearly 3,000 employees, highlighting the strategic acquisition of Hubs to expand their service offerings. Lundberg discusses the operational challenges of maintaining speed and quality at scale, including limitations in handling super-tight tolerances or highly unique geometries, and outlines Protolabs' future focus on expanding into high-volume production capabilities.

    Manufacturing partners & offshoringDesign for manufacturing & NPI
    9 chapters

  • Product Startup · 19m

    182: Consumer Product Hardware Trends in Mid 2023

    Brian Thompson, head of Creo CAD software at PTC, discusses how modern hardware companies are integrating simulation, additive manufacturing, and model-based definition directly into the core CAD environment to accelerate time-to-market. He explains that staying within a single native CAD model, such as Creo, eliminates context switching and data errors, allowing startups and enterprises to iterate rapidly from concept to production. The conversation highlights specific capabilities like ANSYS solvers for structural analysis and Oxetic lattices for energy absorption, which enable designers to validate performance and manufacturability without leaving the design tool. Thompson emphasizes that this unified approach de-risks product development by allowing teams to test with real customer feedback on short-run additive prototypes before committing to expensive full-scale tooling.

    PCB design & layoutMechanical engineering
    6 chapters

  • The Manufacturing Executive · 37m

    A New Era of Design Engineering Productivity w/ Hugo Nordell

    Hugo Nordell, founder and CEO of Encube, explains how his software addresses the gap between design intent and manufacturability by leveraging AI to predict CNC machining costs and climate impact at the point of design. He argues that decades-old engineering software paradigms are failing modern manufacturers due to a severe talent scarcity, where new engineers lack shop-floor experience, and rapidly changing supply chains driven by reshoring and sustainability legislation. Nordell details how Encube’s platform uses machine learning to simplify complex simulations and provide real-time feedback on manufacturability, aiming to unlock 10x productivity growth for companies like Volvo and major aerospace contractors.

    Design for manufacturing & NPIAI & machine learning
    6 chapters

  • Product Startup · 26m

    163: 2023 Trends in Product Development for Hardware Products

    Brian Thompson, head of Creo CAD software for PTC, discusses three key 2023 product development trends for hardware startups: using design simulations earlier in the process, integrating design for manufacturing directly into CAD to avoid context switching, and leveraging additive manufacturing for short-run production. He explains how these tools reduce risk and cost by allowing designers to explore more alternatives and get real user feedback before committing to expensive tooling. The conversation highlights how software advancements make simulation and DFM accessible to non-experts, enabling faster iteration and better product quality.

    Prototyping & iterationDesign for manufacturing & NPI
    5 chapters

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by The Robot Report Podcast. The episode itself belongs to its publisher.