Presentations · 2026
Software-quality and functional-safety guardrails for AI agents
为 AI 智能体建立软件质量和功能安全护栏的探索
Overview
The presentation proposes task decomposition, modeling rules, traceability and human review for AI-generated engineering models. Reusing established MBD verification and CI APIs supplies process guardrails; examples also show gaps when a basic agent is asked to deliver ISO26262 work products.