The platforms our
research is built on.
Studying how machines understand a real operation means working across the full stack of a modern industrial site: cloud infrastructure, frontier AI models, and the industrial software that runs the floor.
These are the platforms our systems are built on.
The partners behind our work.
Each of these is a program we applied to and were accepted into.
Anthropic (Claude)
The reasoning layer on top of a deterministic engine.
Nexus resolves the exact rung, the citation, the reproducible answer. Frontier models sit on top of that engine rather than underneath it, which is why the system gets better as the models improve and why its answers do not drift when they do.
Inductive Automation (Ignition)
The supervisory layer our systems read.
Ignition is one of the platforms Nexus reads directly. Holding the credential is how we learn that layer well enough to reason over it: the tag structures, the historians, and the conventions real plants actually use rather than the ones the documentation assumes.
Microsoft (Azure)
Cloud infrastructure for the work that runs off the floor.
Nexus deploys on premise and air gapped, but the research around it does not have to. Azure carries the development and evaluation infrastructure behind the systems, and it is the environment most of the enterprises we work with already standardize on.
Google Cloud
Where the research runs at volume.
Studying how a system reasons over an operation means running the same questions across a lot of real data, repeatedly, and measuring what comes back. Google Cloud carries the data and evaluation workloads that make those results worth trusting.
Amazon Web Services (AWS)
Meeting operations on the infrastructure they already trust.
Plenty of industrial operations already run on AWS, under security review and audit their teams have done once and would rather not repeat. Being in the network means our systems can land there instead of asking anyone to make an exception for us.
Why working across every
layer matters.
Most tools live in one layer. The questions we study only show up where the layers meet.
Working across all of them is what lets our systems:
- Understand the physical system, not just its data
- Reason over real operational signals as they happen
- Act in ways that hold up when they meet reality
A factory, a logistics operation, an internal platform. The research holds only when it runs in production, not in a slide deck.
See the research running.
Nexus is the first place our research meets a live operation. Trying it is the clearest way to see what we are building toward.