Agentic Systems Engineering.
Move fast. Sound decisions.
Grunetal's Ref is a collaboration platform for model based systems engineering, covering the full development cycle.
We designed Ref for the AI R&D age, where teams iterate fast with agentic execution, and virtual development throughput is massive.
Ref is built for emerging industries, to keep development coordinated as product and processes change. This also means that you can start right away, without installation.
Book your demo
We listen.
You talk directly to the people building the product. We design for easy use and generic utility. We ship features in days, tested with tough edge cases. Our architecture is designed for fast extension.
Full trail.
From first idea to mature product.
Agentic automation is multiplying development throughput. More variants, more test runs, more data, faster. Teams that can't keep up with their own output lose track of what's current, what's been decided, and why.
Requirements shift, designs branch, campaigns produce results, variants compete. Under extreme complexity, Ref keeps the full development trail intact.
Requirements & architecture
Start with what you have. Connect your repositories, upload model directories, experiment data. Write requirements and derive your initial architecture.
Development
Models, software, supplier inputs, design variants, organized in one place: engineering data management without a PLM project. Ref works with MBSE artifacts, SSP, FMI, to give you a few examples.
Test campaigns
Campaigns of hundreds of runs in containers on Azure Batch, started by you or by an agent swarm. Ref references exact inputs, code, and environment. Comparisons stay meaningful. Results are reproducible. Read how simulation data management works in Ref.
Evaluate and decide
Compare variant results against your criteria, with traceability from every result to its inputs.
Every decision linked to its evidence. Every change varianted.
Know immediately what is affected when something changes. No surprises across teams.
Know which data is latest. Get notified when dependencies change. No more stale baselines.
AI-native
AI agents that know your project.
Ref was designed for AI agent swarms from the ground up. Swarms of agents run your tools in containers, from one interactive session to HPC campaigns of hundreds of runs on Azure Batch, and file every result with its trail. We would love to show you in person how we do this.
Open by design
Bring your own tools. Keep your workflows.
Ref is built open. Keep using the tools and standards your teams already use, from MBSE toolchains and simulation environments to FMI/SSP and scenario testing on HPC. No lock-in. Your processes stay yours.
Our roots
Made by industry veterans. Ref runs in EU data centres.
Ref grew out of decades of R&D collaboration consulting and project management, designing and managing global corporate platforms. We are a German company, working across Europe for best cost and compliance. Your data stays in your country, or wherever you want it to be.
Consulting and platform work by our founders for these companies, before Ref. Ref is for small and medium enterprises who are using AI automation to scale up by orders of magnitude.
We listen. We adapt. We deliver.
The teams who organize their development for agentic throughput set the pace in their industry. Free pilot. See where you stand.
Tell us what your team needs.