Agentrys has successfully raised $24.5 million to integrate intelligent AI agents into the microchip development automation process. This funding comprises a $19.1 million seed round that was oversubscribed, and a $5.4 million pre-seed round.
Etna Labs led the seed round financing, while MediaTek spearheaded the earlier pre-seed investment. The capital raised will be used for hiring personnel, developing agent-specific tools, and expanding client engagement.
Founded by Mark Ren, Agentrys develops a concept it calls Agentic Design Automation (ADA). This approach goes beyond traditional EDA software, which only automates individual engineering tasks. Instead, it enables semiconductor teams to create AI-based workforces capable of managing and refining entire design cycles.
Chip development traditionally relies heavily on highly skilled engineers and thousands of hours of manual labor. Agentrys aims to reduce this burden by integrating autonomous agents directly into existing engineering environments. The company's platform is compatible with commercial EDA tools, proprietary software, and infrastructure.
The company asserts that its open architecture allows clients to build and own agent workforces tailored to their unique processes. Furthermore, the intelligent design layer learns from client data, workflow activity, and evaluation signals.
Agentrys demonstrated its platform using an autonomous multi-agent workflow for a 32-bit central processor. The system progressed from specification to final GDS layout with verification, requiring no human intervention. The company also reported achieving over 90% accuracy when tested against NVIDIA's public CVDP benchmark.
The new funding will provide Agentrys with additional resources to advance its engineering and research capabilities. Plans include hiring specialists and creating agent-focused tools for semiconductor workflows. The company also intends to expand collaboration with clients in verification and physical design areas.
Mark Ren, founder and CEO, noted that 'creating production agents that reliably automate real engineering work is far from simple.' He emphasized that Agentrys helps engineering teams transform manual operations into continuously improving agent workforces.
Representatives from Etna Labs noted that chip design is particularly suited for recursive improvement via AI, as engineering outcomes can be objectively assessed. MediaTek also highlighted the potential for extracting and systematically reusing semiconductor knowledge during the chip design process.
Agentrys currently works with fabless design companies, global foundries, and emerging chip startups. Their current projects cover digital and analog design schemes, while system design is planned for the company.
The company's long-term goal is to make AI systems increasingly competent throughout the entire semiconductor design lifecycle. Each workflow execution can generate engineering insights that are then used for subsequent evaluation and improvements.
This model potentially allows companies to retain specialized knowledge while reducing dependence on scarce engineering expertise. It also empowers design teams to customize AI agents to their specific processes and infrastructure. Agentrys believes this approach can create a cumulative advantage for semiconductor companies, allowing clients to develop systems perfectly aligned with their workflows rather than renting generic AI capabilities.
Ren previously worked in AI and EDA research at NVIDIA Research and IBM Research. He also led the ChipNeMo initiative, an early industrial effort to create large language models focused on chip design. This creates a system designed to enhance its capabilities with repeated use by engineers.
