DAC 2025: Towards Multi-Agent Systems In EDA

By Sally Ward-Foxton, EETimes | July 7, 2025

SAN FRANCISCO, Calif. – At DAC, EDA vendors showcased their visions for future multi-agentic systems that can handle greater portions of the EDA workflow autonomously.

During his Monday keynote, William Chappell, vice president and CTO for the strategic missions and technologies divisions at Microsoft, showed advancements in agentic workflows for scientific research and development from the company.

His example showed a task manager agent asked to come up with an approach to developing a vaccine for a particular disease. This agent was able to understand the assignment, reflect on whether the task was executable (if not, it could break it down into further steps), and call other agents to reason and implement parts of the scientific method. These agents were able to make a hypothesis, call tools to gather relevant information, generate a shortlist of possible proteins and analyse their efficacy. The task manager agent collected output task logs from the other agents, and if it judged that they failed their tasks, it could ask them to start again.

The tools being called could be existing software tools, or could be generated by the agents, which have the ability to write code.

To read the full article, click here