Research
Research and results from the Agentrys team — building Agentic Design Automation for chip design.
June 2026 · Research
Composable Silicon: an autonomous multi-agent workforce for chip design
An autonomous multi-agent workforce takes a chip from spec to sign-off-clean GDS on a fully open-source flow — designing AgentCore, a 32-bit CPU, and optimizing it for XLOPS/W with no human at the wheel.
June 2026 · Research
Cracking CVDP's hardest tasks: autonomous verification at scale
On CVDP's three Design Verification categories — testbench stimulus, checker, and assertion generation — Agentrys' self-learning agent lifts pass rates to 90–96%, a 4–15× improvement over prior state-of-the-art, with no human in the loop.
February 2026 · Research
Self-Improving Agents Solving CVDP RTL Coding Tasks
Agentrys achieves 95.8% overall pass rate on four RTL coding tasks of the Comprehensive Verilog Design Problems benchmark, including a perfect 100% on Code Debug, through seven generations of self-improving agents.