Technology should accelerate great ideas, not stand in their way. That is why AskDona exists.
Researchers, engineers, and HPC teams spend valuable time searching for information, understanding complex computing environments, and resolving technical uncertainty — time that could be devoted to research and discovery.
AskDona is an AI-powered knowledge platform that helps scientific organizations make trusted institutional knowledge easier to find, understand, and use.
Behind every computation
is a person asking a question.
High-performance computing is one of humanity's most powerful tools for scientific discovery. Behind every simulation is a researcher seeking an answer to an important question. Behind every computation is a team working to expand what humanity can understand and achieve.
AskDona does not replace HPC systems or scientific expertise.
It supports the people who make them meaningful.
In production at the
supercomputer Fugaku.
AskDona powers the first-line inquiry support chatbot on the official Fugaku support site at RIKEN Center for Computational Science (R-CCS) — adopted in July 2024 and fully adopted in February 2025 — answering from roughly 16,000 pages of specialist documentation without additional tuning.
4h → 5s
First-response time for user inquiries that previously took specialists up to four hours.
−61%
Peak year-on-year reduction in support tickets to human staff (April 2025).
6,558
HPCI and JHPCN research reports searchable in natural language via the R-CCS Supercomputer Outcome Navigator.
+20pts
Accuracy advantage over competing RAG products, verified jointly with RIKEN.
Our journey with RIKEN R-CCS
From first deployment to a published operational report
Trusted by leading research organizations
Questions don't wait
for systems to fail.
HPC users need support throughout the entire research process — from learning how to use the environment and confirming technical specifications to installing software, submitting jobs, transferring data, and optimizing performance.
Classifying the HPC user inquiries received by AskDona, 37% were technical support for troubleshooting and code or scripts, and 25% concerned basic operation, learning how to use the system, and understanding concepts.
Questions about specifications, usage conditions, and operational rules also accounted for 26% of all inquiries.
What HPC users need help with
Share of analyzed inquiries by category
Nearly two-thirds of all inquiries relate to troubleshooting, technical support, and understanding how to use the computing environment.
Where one answer isn't enough
Highest similar follow-up rate by area
Software build and library questions stand out — a similar follow-up question appears in approximately 45% of cases, indicating that a single answer is often not enough to resolve the underlying uncertainty.
AskDona helps make the knowledge behind these questions easier to find, understand,
and apply —
so researchers can resolve uncertainty faster and continue their work with confidence.
The figures presented on this page are based on our independent analysis of inquiry data collected from a specific population during a defined period. They may not be representative of all HPC users or computing environments, and the results may vary depending on the classification methodology, scope of the analysis, and timing of the analysis.
With the power of scientific computing, advancing the next generation of research.
AskDona helps researchers, engineers, HPC users, and operations teams spend less time navigating complexity and more time advancing science. We help organizations:
Access trusted institutional knowledge across technical documents, manuals, research materials, and operational resources.
Understand complex computing environments more quickly — from schedulers and libraries to site-specific policies.
Reduce the time spent searching for information and resolving recurring questions.
Preserve specialist knowledge across generations of researchers and engineers.
Accelerate onboarding for new users, researchers, and technical teams.
Improve knowledge sharing across laboratories, departments, and institutions.
Establish consistent, transparent AI-assisted assessment workflows in areas where expert review remains essential.
Every hour spent searching for information is an hour that could have been spent discovering something new.
AI for Science
Generative AI and high-performance computing are not competing technologies. They address different challenges.
Model, simulate, investigate.
High-performance computing enables people to model, simulate, and investigate complex physical and scientific phenomena — the systems and the physical world that research seeks to understand.
Find, organize, understand.
Generative AI helps people find, organize, understand, and interact with knowledge — the documentation, code, experience, and expertise that surround every computing environment.
Scientific progress increasingly depends on both. AskDona supports the connection between people, institutional knowledge, and advanced computing — making trusted information available wherever research happens.
