Multi-Agent AI Systems for Automated CAD-to-Print Workflows

While Additive Manufacturing hardware has advanced, the digital pre-production workflow—encompassing CAD model preparation, optimization, and slicing—remains a significant, expert-driven bottleneck. This process is often manual, time-consuming, and difficult to scale. This presentation introduces a novel framework that employs a multi-agent system (MAS) of specialized AI agents to automate and optimize this complex digital pipeline.

In this paradigm, the workflow is managed by a team of collaborative software agents, each with a distinct specialization. These include: a "Design Analysis Agent" that assesses part printability and identifies topological features; a "Support Generation Agent" that intelligently designs and places minimal, effective support structures; an "Orientation Agent" that determines the optimal part orientation for build success and material properties; and a "Slicing Agent" that selects ideal process parameters and generates machine-specific instructions.

We will detail the coordination strategy that allows these agents to communicate, negotiate, and converge on a globally optimized solution. For example, the Orientation Agent negotiates with the Support Agent (to minimize support needs) and the Slicing Agent (to ensure feature resolution) simultaneously. This collaborative decision-Making process surpasses the capabilities of monolithic, sequential software.

This presentation will outline the MAS architecture and the machine learning models underpinning agent decision-making. A case study will be presented, comparing the time, material usage, and engineering effort required for this automated workflow. The results demonstrate a reduction in pre-production lead time and a significant improvement in build-setup consistency, offering a scalable strategy to unlock high-throughput, autonomous AM production.