AMD expands its software stack with Rocm.ai
AMD has expanded its Rocm software stack with a new AI-native component, Rocm.ai.
Rocm.ai, announced at AMD Advancing AI 2026, combines AI-assisted development, intelligent deployment, and AI-powered software optimization.
AMD claimed that AI software development is becoming increasingly agent-driven, with developers using AI coding agents and natural language processing to build and deploy software. AMD believes that Rocm.ai – and in particular the inclusion of AMD Skills in it – will reduce the barriers to adopting AMD Rocm and also reduce the amount of document crawling needed for set-up.
Speaking at AMD’s annual Advancing AI conference, Vamsi Boppana, SVP of AI at AMD said: “Rocm.ai … brings the capabilities of AI-assisted GPU programming to developers. It starts with a simple idea of giving developers access to the power of an AI ecosystem through the AI coding agents they already use. Whether you are using Cursor, Cloud, [or] Codex, Rocm.ai helps those agents understand AMD platforms, understand Rocm, and helps [developers] build and optimize [their] workloads.”
“In other words, we’re making those popular coding agents into Rocm superusers – you should be able to describe the workload you want to run, the performance target you want to hit, and then let the agents help you get there,” he said.
Breaking down Rocm.ai
Drilling down into what actually constitutes AMD Rocm.ai, there are three main components: Rocm CLI, AMD Skills, and Hyperloom.
Hyperloom is an AI-assisted optimization system that can, according to Boppana, analyze workloads, choose configurations, select and choose kernels, adjust parallelism strategies, and initiate towards performance goals.”
Boppana gave the example of a lead developer he recently spoke to who “pushed through a suite of 14,000 models through Hyperloom, optimizing them and creating valuable insights”.
“This would have been impossible to imagine, even with a large team of developers, before,” he said.
Rocm CLI, meanwhile, provides a unified command-line for installation, validation, servicing, updating, and troubleshooting of AI workloads on AMD platforms. AMD Skills integrates AMD-authored expertise into coding assistants like Claud, Cursor, and Codex.
“Let me make this concrete for you,” Boppana said on stage. “Imagine you’re a developer who wants to optimize Vmax… with PLLM on MI355s. Today, there are many things you need to get right. You need the right PLLM optimizations, the right model recipes, the right environment variables. You may need to write new kernels or optimize kernels – and then you need to tune all of this.
“With Rocm.ai, this interaction becomes much simpler. You can say ‘optimize Vmax entry with Hyperloom’ [hit] enter and behind the scenes, the agent does what an expert would do: pulls a known good recipe, runs and profiles the workload, creates and tests a bunch of kernel and runtime configurations, and it keeps iterating towards target.”
“Developers shouldn’t have to manually reason through every layer of model architecture, kernel selection, prompt strategy, and rack-level deployment,” said Boppana. “AI is transforming this experience and that’s a very, very big shift for AMD hardware and software.”
Rocm.ai be available from August 2026.
