In brief
- AMD introduced HPC Assistant as an open-source AI assistant and said its code is available on GitHub.
- The first version was developed with CSC, which operates LUMI, and connected to LUMI’s knowledge base.
- AMD lists file transfers, job submission, GPU profiling and energy-consumption analysis among the workflows the assistant supports.
- AMD said it has begun working with Oak Ridge National Laboratory on a Frontier integration; the source does not say that use on Frontier has begun.
On November 17, 2026, AMD announced the open-source HPC Assistant, designed to help researchers manage workflows on high-performance computing (HPC) systems. According to the company, the assistant provides center-specific information and automation through a natural-language interface; its code and documentation are available on GitHub. AMD said it will also demonstrate the tool in Chicago during SC26.
Works with center-specific information
HPC centers can differ in their file systems, storage quotas, job schedulers and software policies. In AMD’s approach, the assistant uses a center’s approved documentation to provide answers in the right context. According to the company, this can give users system-specific guidance—for example, on where to store a large dataset or under what conditions to submit a job.
AMD says the assistant can turn natural-language requests into steps such as transferring files, preparing and submitting batch scripts, tracking jobs and retrieving results. These are capabilities described in the company’s product announcement; the source does not provide independent measurements of deployment or results at different centers.
The first version was developed with CSC, the Finnish IT Center for Science, and connected to the knowledge base for LUMI, which CSC operates. AMD also says it has begun working with Oak Ridge National Laboratory to bring HPC Assistant to Frontier. This does not mean that use on Frontier has begun.
GPU performance and energy visibility
According to AMD, the assistant can help analyze GPU profiling data, providing information on areas such as compute utilization, cache efficiency and memory traffic. The company aims for these data to help identify performance bottlenecks and assess where optimization may be possible.
According to AMD, the energy profiling feature uses system metrics measured with Omnistat and makes job-level consumption visible through the assistant and Grafana. AMD also says it supports optimization of matrix-heavy kernels for AMD GPU architectures with GEAK. The product announcement also lists the ability to submit and track these optimization tasks as a feature of the assistant.
AI workflows for scientific research
AMD says that, among the features it introduced as previews at SC26, users can launch AI agents for scientific workloads and AMD Inference Microservices (AIM) solutions. The example AMD gives for LUMI is OpenFold3 AIM for protein structure prediction. According to AMD, in this scenario the assistant launches the GPU server, downloads model weights, submits the job and provides guidance during visualization.
Another preview feature, AIM Agent, is designed to help teams prepare to run their own models in an HPC environment. According to AMD, the tool reviews model repositories and documentation; identifies requirements; creates build recipes and container images; runs tests and benchmarks; and prepares a deployment report.
What it means for architecture and archviz workflows
AMD HPC Assistant was not introduced as a product specifically for architecture or visualization software. The announcement also provides no information about direct connections to CAD, BIM or rendering tools. As a result, any potential use for architecture teams would depend on those workloads being run on an HPC center’s infrastructure and the center’s knowledge base being connected to the tool.
Offices and visualization teams should first assess compatibility with their existing compute infrastructure, the GPUs they use and center-specific policies. AMD says the open-source system can be adapted by centers, but the source gives no information about compatibility with specific architecture software or costs.
Sources
1 sourceSource texts are not republished; short quotes are marked, everything else is our own summary and commentary.
The main promise of AMD HPC Assistant is to make technical steps involved in using supercomputers more accessible through natural language. Processes such as job submission, profiling and energy-consumption monitoring may be worth exploring for research teams using HPC infrastructure; however, the announcement does not show direct integration with architecture or rendering software.
For offices and visualization teams in Turkey, the practical value depends on whether they already have access to this kind of computing infrastructure. Open-source code offers the possibility to review and adapt the tool, but integrating center documentation and checking hardware compatibility will require separate consideration. This source provides no information about the product’s cost.
Frequently asked questions
What does AMD HPC Assistant do?
According to AMD, the assistant provides answers based on HPC centers’ own documentation and helps with processes such as file transfers, preparing job scripts, job tracking, GPU profiling and analyzing energy consumption.
Can AMD HPC Assistant be used on LUMI and Frontier?
The first version was developed with CSC, which operates LUMI, and connected to LUMI’s knowledge base. AMD said it has begun working with Oak Ridge National Laboratory on Frontier; the source does not say that use on Frontier has begun.
Is AMD HPC Assistant open source?
Yes. AMD says its code and documentation are available on GitHub.



