Ursa Major, a Colorado aerospace and defense company, is pairing AI coding assistants with EOS's documented machine interfaces to write its own build-preparation and fleet-monitoring software for metal additive manufacturing. According to an EOS case study, one part-specific thermal strategy for an EOS M 400 that once meant days of manual programming now takes 10 minutes.
The work centers on Polaris, Ursa Major's custom slicing repository, where engineers connect large language models to the code. An engineer can hand an AI assistant a newly published academic paper on laser scanning strategies, ask it to write the scanning function, and apply custom parameters to individual parts on a build plate. In the demonstration, the tool sorted hatch tiles from the outside in based on each component's centroid. It also updated the configuration files, built the support structures, and prepared the slice file.
The second project is fleet monitoring. Tom Pomorski, Director of Additive Manufacturing at Ursa Major, fed EOS CONNECT Core API documentation into an AI coding assistant and asked for a database that polls the machines at set intervals. The result is a web-based system built in one week with 100 percent AI-generated code. It logs build data to a local SQL database, projects live camera feeds from active printers, and sends automated status reports. If it detects a short feed or a recoater snag, it emails engineers an image of the build chamber. "The real unlock in the future is going to be the companies that have really good documentation for how these tools interface with the real world or with hardware," Pomorski said.
For automation engineers, the example shows how much integration work now depends on interface documentation. Ursa Major used the EOS REST API and EOS CONNECT, along with EOS's laser toolpath API.
The printers behind this are at Ursa Major's Youngstown, Ohio facility. The fleet totals six EOS-powered systems: an AMCM M 290-1 FLX, two EOS M 400 machines, and three AMCM M 450-4 FLX units added according to a September 2025 EOS release. Ursa Major's Hadley hypersonic engine is roughly 80 percent 3D-printed by mass and has flown more than nine missions. For solid rocket motors, its LINX process prints different-sized motor cases and internal tooling on the same machine. Printed hardware replaces forged parts that can carry 12-month lead times for exotic steels, which lets the team design, print, test, and iterate in weeks.
EOS's case study draws a line at production. Flight-critical parts still require deterministic verification, closed-loop process validation, and strict quality control, and EOS describes the AI-generated code as a tool for process development. For customers who need standardized production, the company also sells validated, out-of-the-box software.
