Autodesk Platform Services (APS) is moving toward AI-assisted development, with tools intended to help developers discover capabilities, build applications, manage software portfolios, and distribute solutions. At Autodesk University 2026, the company presented a future in which developers describe their goals and AI helps assemble the services, tells Beyond PLM blog.
APS connects design and project information through APIs, supporting viewing, data access, and cloud automation. Autodesk Assistant helps developers identify APIs and assess business value. The APS Workflow MCP is designed to understand API dependencies and help coding agents produce working applications, allowing developers to concentrate on reviewing and improving them.
Availability matters. Workflow MCP remains in private beta, while conversational administration and deeper integration of marketplace applications are future plans. Existing APIs and Assistant capabilities provide the starting point.
In Beyond PLM, Oleg argues that the evidence for APS comes from customers organizing fragmented information. One engineering consultancy connected project records across Autodesk services, SharePoint, and Outlook. During a controlled test, its agent identified conflicting generator capacities in calculations and an electrical diagram. The example illustrates the value of bringing records together before asking AI to interpret them.
Other examples reinforce that sequence. A substation engineering firm established structured project data before adding automation, reporting a 25% improvement in initial design delivery. JE Dunn reduced project update times from 12 hours to 5 minutes after investing in connected enterprise data.
The article also raises an ownership question: Autodesk may connect information while another platform supplies its business context and meaning. Manufacturing examples remain limited, although a Vault-to-ERP connector demonstrates proposed bill-of-materials changes requiring human approval.
For engineering organizations, Oleg’s argument is practical: structure and connect data first, then apply agents with people responsible for decisions. Easier integration can reduce programming effort, but understanding relationships, checking proposals, and exercising engineering judgment remain essential.
