Según su CEO, ChipAgents busca llevar IA agéntica a la verificación de chips
ChipAgents no diseña chips: los verifica con ia agéntica para chips, un enfoque que cambia los bancos de pruebas escritos a mano por agentes que generan y corrigen esos tests solos. La startup, fundada por el investigador William Wang, detalló el método en una entrevista con Pulse 2.0. El dato importa porque la verificación, no el diseño, es la fase que más tiempo consume en el desarrollo de un…
ChipAgents is an AI-driven startup based in Santa Barbara, California, founded by research scholar William Wang. Their core mission is to revolutionize the semiconductor industry by automating the verification process of chips, rather than designing them. This shift in focus targets the time-consuming phase of chip development where bugs often slip through and cause costly delays.
ChipAgents' unique approach involves using autonomous agents that generate test benches, run simulations, and debug errors without human intervention. This automation not only streamlines the verification process but also enhances the accuracy of the final product, as it reduces the chances of bugs going unnoticed. The startup's founder, William Wang, explained this innovative methodology in a recent interview with Pulse 2.0, highlighting how their agents can generate test benches, run simulations, and fix failures without any manual input from verification engineers.
The company's agents work in a loop, continuously generating test benches, running them through the RTL simulator, and interpreting the results. If a test fails, the agent attempts to pinpoint the specific line of RTL code responsible for the error and proposes a correction. This process significantly reduces the time engineers spend on repetitive tasks, allowing them to focus on more critical aspects of chip design.
Wang emphasizes that ChipAgents does not replace the role of hardware verification engineers. Instead, it complements their work by handling repetitive, time-consuming tasks, thus enabling engineers to dedicate more time to refining and perfecting the chip design. This AI-driven approach is a novel application of natural language processing techniques, similar to those used in software agent development, but adapted to the unique challenges of hardware verification.
By automating this critical phase of semiconductor development, ChipAgents aims to accelerate the time-to-market for chips and improve the overall quality and reliability of electronic devices.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
