AMD has introduced AMD Ross, an agentic AI assistant designed to accelerate the embedded development lifecycle, spanning hardware and software design, optimisation, debugging, AI implementation and deployment.
Purpose-built for AMD Embedded development environments, AMD Ross combines direct access to AMD tools, an AMD-validated Knowledge Base and expert-authored agent skills. The platform enables engineers to use natural language to search documentation, execute tool commands, generate code, troubleshoot errors and optimise designs.
Unlike generic AI coding assistants, AMD Ross is designed around the tools, methodologies and engineering knowledge used across AMD’s embedded portfolio. Its capabilities span hardware design and debug, embedded software and algorithm development, edge AI implementation and system-level design.
The platform brings together four core components: Model Context Protocol (MCP) servers that connect AI agents to AMD Embedded tools; an AMD Knowledge Base containing validated technical documentation; expert-authored agent skills that capture repeatable engineering practices; and ready-to-run design examples.
This architecture is intended to help engineering teams reduce repetitive work, shorten debugging and optimisation cycles, accelerate onboarding and improve the reuse of institutional knowledge.
“AMD Ross brings AMD Embedded tools, trusted knowledge and expert-authored workflows together in a single agentic AI experience grounded in the technologies and methodologies our customers use every day,” said Salil Raje, senior vice president and general manager, AMD Embedded. “AMD Ross brings the power of agentic AI to embedded developers to move product innovations from design intent to deployment faster by accelerating the entire life cycle.”
Among its use cases, AMD Ross can help engineers with hardware and software partitioning, high-level synthesis-based hardware designs, silicon optimisation and debugging, embedded software development, machine-learning optimisation, power estimation and optimisation, and system schematic review and board-layout optimisation.
The assistant is also client-agnostic, allowing developers to work with their preferred large language models, IDEs and command-line environments while maintaining connections to AMD Embedded development tools.
AMD said the platform can accelerate time-to-prototype through workflow automation, reduce debug and optimisation cycles through validated guidance, automate repetitive tool interactions and help new engineers access established engineering practices.
An example cited by AMD is the use of Ross to analyse timing issues, identify violations and root causes, and recommend implementation improvements. It can also assist with Vitis HLS optimisation through recommendations around pipelining, loop optimisation, pragma insertion and architecture refinement.
AMD Ross is available now, with additional AMD Embedded tools and workflow capabilities planned on a monthly cadence.