Toolpath’s browser-based AI platform aims “to make it 10x faster to go from digital design to high-precision machined part.”
This is Engineering Paper, and here are the latest design and simulation software updates.
Sometimes I worry that I’m going to run out of AI startups to write about. Thankfully, they keep popping up like heads on a Hydra. Today I draw my pen on Toolpath, an AI startup focused on CAM automation.
“We’ve built an AI engine to remove the things users do in CAM software today,” Toolpath CEO Al Whatmough told me.
Namely: Toolpath automates toolpath creation.
In a manual CAM workflow, Whatmough explained, users specify their toolpaths tool by tool and setup by setup. It requires know-how and patience. Even for a simple part, generating a toolpath could take an experienced user 30 minutes or more.
Toolpath does the whole thing for you in a couple minutes.
It works like this: Users upload their CAD part to Toolpath’s browser-based interface. Toolpath’s AI, with knowledge of the user’s available set of tools, generates a full 3-axis milling strategy along with a cost and cycle time estimate.
“A part that’s manufacturable for one person is not manufacturable for another. And so this looks a lot like DFM [design for manufacturability], but it’s not really. It’s DFM that’s specific to the context of the capabilities a given shop,” Whatmough said.
If a part can’t be machined with the user’s existing tools, Toolpath will suggest some from a database of “tens of thousands” of tools from different vendors, according to Whatmough.
Whatmough showed me a brief demo of the AI platform in action. He pulled in a part (the bike clamp top shown above), and without doing anything else Toolpath began analyzing it. About a minute later, it had generated the full machining strategy, with four setups and six tools from Whatmough’s library, with a cost estimate of $176.95.
A toolpath is just that, and to turn it into a real-world part you still need a CAM system to simulate it and generate the CNC machine code. Toolpath lets you send the generated toolpath directly to Autodesk Fusion, where users can take it to production in the Fusion CAM workspace. In the demo, Whatmough imported the bike clamp’s machining strategy into Fusion by copying and pasting a key into the Toolpath add-on, which recreated the setups natively in Fusion.
Whatmough knows Fusion CAM well—he was Autodesk’s director of product management for manufacturing until he left in 2021. But there’s nothing about Toolpath that requires Fusion, he said, and Toolpath plans to support other CAM systems too. Eventually, it may outgrow them entirely.
“When we’ve done our job right, we will have a closed loop system that goes the whole way to the machine,” Whatmough said.
Toolpath is already raising eyebrows and capital in the CAM world. Whatmough says the platform has a couple hundred paying subscribers (the base subscription is $1,500 per year), and last week it closed its latest investment round with funds from toolmaker Kennametal, CAM developer ModuleWorks, and venture capitalist firm Leaders Fund. Toolpath didn’t disclose the value of this round, but to date the company has raised approximately $20 million USD across three rounds, according to a company spokesperson.
I found this particular head of the AI Hydra to be less hyperbolic than most. Toolpath is focused on automating something with reproducible steps and a clear solution. Whatmough says the AI is trained to “play the game of machining a part, similar to the way you train an AI to play the game of chess.”
A chess-bot might make a bad move, but it won’t ever make an illegal move, and likewise Whatmough says the Toolpath AI won’t hallucinate. “It’s a more explicit [AI]… you have to remove all the geometry of this part, and these are the tools you get to use. It’s how I do it as a programmer today.”
I was impressed with Whatmough’s demo, but I always take demos with a grain of salt. The end users get the final say, so if you’re one of them, please share your thoughts on Toolpath in the comments or message me directly at malba@wtwhmedia.com.
Quick hits
- Siemens Digital Industries Software has entered into an agreement to acquire Excellicon, an EDA software developer focused on developing, verifying and managing timing constraints. Siemens didn’t disclose the terms of the deal.
- AMD announced new GPUs and CPUs at Computex 2025 in Taipei last week. They include the Radeon RX 9060 XT and Radeon AI PRO R9700 graphics cards and Ryzen Threadripper 9000 Series processors.
- Reality capture company XGrids announced a new plugin for Autodesk Revit that integrates XGrids’ Lixel CyberColor (LCC) technology with the popular BIM platform. LCC combines lidar data with 3D Gaussian splatting and “automatically generates spatial models that capture both visual fidelity and structural accuracy,” according to the XGrids press release. LCC for Revit will be available on the Autodesk app store by the end of May 2025.
One last link
Engineering.com senior editor Ian Wright writes about his latest obsession in I love lattices.
Got news, tips, comments, or complaints? Send them my way: malba@wtwhmedia.com.
