Human-Centered Agentic AI Workflows for RTL Verification

Introduction

Design and verification productivity are increasingly constrained not by individual tool performance, but by process-level complexity across design creation, verification, and iteration as designs evolve. While continued advances in engines and solvers remain essential, modern teams face growing friction in planning, steering, and re-steering work across tightly coupled workflows.

The Questa™ One Agentic Toolkit extends the Questa One solution with human-centered agentic workflows that embed intelligence directly into these workflows, rather than isolated tools. By assisting engineers with design creation, verification planning, execution steering, and analysis across tools and iterations, the toolkit helps reduce manual coordination while preserving the trust, accountability, and rigor required for RTL sign-off.

^{Figure 1: Agentic AI learning model.}

Rather than replacing engineers or enforcing full autonomy, the toolkit is explicitly designed around human-in-the-loop operation. Agents propose actions, execute bounded tasks, and summarize outcomes, while decision authority remains with the engineer, enabling more adaptive, scalable productivity without compromising verification integrity.