SEMAA: AI Agent Assistant for MBSE

Ardalan Aryashad
USC IMPACT Laboratory

Overview

SEMAA is an AI agent assistant designed to support model-based systems engineering workflows. The system integrates with Visual Paradigm as a plugin, adds diagram import and export tools, and enables context-aware chat so the agent can reason about the current model state. We also built our own plugin that acts as the MCP server for remote LibreChat servers at semma.site, enabling web-based access for collaboration and review.

System Design

The system connects a chat interface to a model interpreter that understands diagram context. It supports exporting diagrams to text, reasoning over the current system state, and pushing updates back into the modeling tool to accelerate MBSE workflows.

Demo