Bringing Conversational AI to PTC Windchill
The AI Assistant Admin Settings UI gives administrators a central place to configure Model Context Protocol (MCP) connections, enabling end users to connect the assistant to the apps and tools they use. The design organizes connection setup, access controls, and availability into a guided workflow, helping administrators understand which integrations are configured and how they become available to users.
For end users, the experience translates that configuration into a clear path to discovering available apps, authenticating their accounts, and confirming connection status. Attention to setup guidance, permissions messaging, and error recovery helps users understand what they are connecting and resolve issues along the way. Together, these workflows support a consistent experience from administrator setup to everyday use.
Designed PTC Windchill’s LLM-powered AI assistant, with chat, MCP connection workflows, and dedicated settings for end users and administrators. The design process explored how each role connects tools, configures the assistant, and interacts with AI within existing workflows. Iterative prototyping and user testing helped evaluate navigation, clarify connection steps, and refine the settings experience around users’ needs.
Additional workflows addressed credit usage and messaging, helping users understand consumption and available limits. Multi-thread chat, context-setting controls, and thread history supported switching between tasks, grounding conversations in relevant product information, and returning to previous work.
I used AI tools to build interactive prototypes, quickly explore design directions, and iterate on workflows for user testing. We gathered feedback from these sessions and synthesized it in Dovetail to develop user personas and identify feature changes. These insights guided refinements to the prototypes, connecting design decisions to user needs.