Ref: Work with agents on dynamic data.
Ref is a cloud-based data system designed for cooperation
with AI in research and development:
- Ref keeps track of your development data and lets you automate processes of agentic environments.
- Designed for minimal organizational overhead,
but with enterprise data governance. - Optimized end to end for minimal costs
across transfer, storage, and token usage.
Ref is built on novel principles derived from decades of industry experience that create decisive advantages for your R&D.
If you work in an emerging industry, and need to manage, automate and audit on large scale research and development throughput, Ref is a likely candidate for your data and automation platform.
We can't publish our strategic advantages on the open web, but we would love to show you in person.
Book your demo
Full trail
Everything connected.
From first idea to mature design.
Product development is iterative. Requirements shift, models branch, simulation campaigns produce millions of results, variants compete. Ref keeps the full data trail intact in massive development cycles.
Requirements & architecture
Start with what you know. The repositories you have, the model directories, the experiment data — just connect and upload. Define requirements and derive an initial architecture.
Development
Your software, models, supplier inputs, variants find their place in Ref. Integrate your own tools and workflows. Ref is built open.
Test Campaigns
Interactive sessions and large-scale HPC campaigns. Every run references its exact inputs, code, and environment. Comparisons stay meaningful. Reproducibility is built in.
Variants, evaluation, decisions
Compare variant results against criteria. Document the discussion, capture the decision, and keep the evidence connected. Then start the next cycle.
Every decision linked to its evidence. Every change versioned.
Exact inputs, exact environment, exact code. Every run can be reproduced.
Know which data is latest. Get notified when dependencies change. Trace change impact. No morestale baselines.
AI-native infrastructure
Context and skills
for AI agents at enterprise scale.
AI agents are only as effective as the context they can access and the skills they can execute. Ref provides both — structured, permissioned, and auditable.
Management oversight
Development status at a glance.
Project leads and managers see what matters:
Variants evaluation state, pending decisions, change impact.
Supporting Standards
Ref supports major research and systems engineering MBSE methodologies and standards. The open design of the system helps us to accomodate most connectors and standards in days to weeks.
Our roots
Made with decades of enterprise data management experience.
Ref grew out of R&D consulting and project management for global engineering organizations. We built this system and pivoted our company for the AI economy, where white collar service cost is essentially hardware, utilities, financial risk and the living expenses of a minimal set of service owners, ensuring you get the lowest possible cost at high quality and compliance within the European geopolitical region.
Cloud-native, as a service
Open your browser and start building.
No infrastructure to maintain. Agents, automated processes and interactive sessions run on managed infrastructure — secure and scalable to extreme data sets. Operations, IT compliance, insurance are not your problems anymore. All while we give you full control and financial tools to manage your costs. You can still utilize your tools and tailored workflows, with no vendor lock-in.
Built for researchers. Built for engineers. Built for you.
We listen to integrate Ref into your environment. Free demo & pilot:
Reach out. Save cost with Grunetal.