Agentrys builds Agentic Design Automation (ADA), a framework of self‑improving AI agents and agent‑native EDA tools that automate the entire chip design flow—from architecture and RTL to verification, physical design, analog, and packaging. By continuously learning from design data and expert feedback, the agents reduce reliance on senior engineers, accelerate design cycles, and increase throughput for semiconductor companies developing AI accelerators.
Funding
Funding not disclosed
Founders
Product
Problem
Chip design for AI accelerators requires deep expert judgment accumulated over years, but senior engineers are retiring faster than new talent can be trained. The shortage of qualified designers limits the ability to meet the rapidly growing demand for advanced AI chips.
Solution
Agentrys is developing Agentic Design Automation (ADA), a framework of self‑improving AI agents and agent‑native EDA tools that operate autonomously across the full chip design flow—from architecture and RTL implementation to verification, physical design, analog, and package co‑design. By encoding expert feedback and continuously learning from design data, the agents iteratively enhance both their reasoning and the underlying tools, creating a compounding improvement loop. This approach aims to reduce reliance on human judgment, accelerate design cycles, and increase the throughput of high‑performance AI chip development.
Target Audience
Primary customers are semiconductor companies and design houses developing AI accelerators that need to scale chip design productivity and reduce dependence on scarce senior engineering talent.
Features
- Agent-native toolchain built from the ground up for high‑frequency, automated interaction with EDA components
- Self‑improving agents that learn from existing design datasets, expert feedback, and each design run
- End‑to‑end automation covering hardware‑software co‑design, architecture, RTL, verification, physical design, analog, and packaging
- Continuous refinement loops that update both agent reasoning models and tool implementations
- Scalable decision‑making capable of split‑second optimizations and bug resolution without manual intervention