Xometry upgrades AI models for custom manufacturing
Xometry, Inc, Bethesda, Maryland, USA, a global AI-native marketplace connecting buyers and suppliers of custom manufacturing services, has announced a series of upgrades to its artificial intelligence model architecture. The company states that the new generation of interconnected AI models supports the manufacturing workflow from quotation through production by drawing on proprietary data covering geometry, manufacturability, pricing, supplier capability and production outcomes.
According to Xometry, the enhanced models are intended to improve the speed and accuracy with which engineering data is converted into manufacturing recommendations and pricing decisions.
Improved manufacturing process recommendations
Xometry has upgraded its context-aware AI process recommender to better interpret a part’s intended application, material requirements and geometry. The company states that the system can infer application-specific requirements, such as tighter tolerances and aerospace-grade materials for aerospace components, enabling it to provide recommendations even for customers with no previous order history.
When a CAD model is uploaded, the system recommends the most appropriate manufacturing process from more than twenty supported manufacturing technologies. Xometry reports that customers accept the AI-generated recommendation in more than 85% of cases, representing an improvement over the previous model, particularly among first-time users.
Updated cost-prediction models
The company has also introduced a new generation of cost-prediction models designed to generate pricing based on the manufacturing requirements of individual parts.
According to Xometry, the models consider a wider range of variables, including geometry, material, surface finish and whether a component is produced as an individual part or within a larger production batch. The company states that this results in more accurate pricing estimates.
The new models have initially been deployed for CNC machining. Trained using Xometry’s proprietary manufacturing data, they reportedly improve CNC cost-prediction accuracy by approximately 15% based on live testing. The models also generate a calibrated confidence level for each quotation before it is returned to the customer.
Xometry plans to extend the updated cost models to additional manufacturing processes available through its Instant Quoting Engine (IQE) and expand the range of CNC materials supported by the platform.
Enhanced supplier matching
The company has also updated its sourcing models to improve job allocation across its supplier network.
According to Xometry, the adaptive models use supplier capability data to match manufacturing jobs with suitable partners. Matching criteria include machine dimensions, process capabilities, historical quality performance and on-time delivery records.
The company states that this approach improves partner matching by presenting suppliers with manufacturing opportunities better suited to their capabilities, rather than broad job notifications. Xometry believes this will improve pricing responsiveness and strengthen the overall supplier network.
“Across the millions of parts quoted through Xometry, our AI is constantly learning,” stated Vaidy Raghavan, Chief Technology Officer at Xometry. “Our new models move beyond treating that knowledge as separate skills, where one model recommends how to build something, another prices it, another finds who should build it. Now, they are an intelligence layer that gets sharper every time a part moves through it. That’s what makes Xometry the digital infrastructure for custom manufacturing.”
He explained, “The data layers are what’s actually deciding how a part gets made, what it costs, and who builds it. And the smarter that data gets, the more directly our customers feel it – more optimal recommendations, accurate pricing, better-fit suppliers – every time they upload a part.”
