Keep the same semantics.
Desktop and server share Grid’s Rust execution engine. Deployment choices change how work is hosted and connected while preserving the meaning of your formulas.
Best to run code.
Your code.
A world of
possibility.
Put computation where your work lives. Compose open-source libraries, proprietary systems, and live data around one model—with a Rust runtime you can run on your own infrastructure.
A model, connected.
DesktopServerCloud
Grid runtime / RustYour model
State · Computation · Reactions
BackboneNodes & jobs
ExtensionsLibraries & tools
ConnectorsData & events
One model. Reach in every direction.
The runtime advantage
One model. More ways to put it to work.
Own the deployment.
Keep model state and computation on infrastructure you control. Expose the inputs, outputs, and live updates your application needs through the public API.
Select the power you need.
Grid’s architecture keeps optional networking, assurance, and specialized capabilities at explicit deployment boundaries, keeping their additional work out of ordinary calculation.
01 / Portable execution
Your laptop. Your servers. Your cloud.
Start a model on your desktop. Put it behind an API. Give people a browser app. Grid’s Rust runtime owns the model’s state and execution across these deployment shapes, so your business logic stays in the model as the way you deliver it changes.
What you can buildBuild a local tool that can become a shared service, with the same model at its core.
Choose where to run02 / Backbone networking
Your model is part of a bigger system.
Backbone connects the runtime to durable jobs, reactive deliveries, peer synchronization, and connector events. It gives models a way to participate in a network of data and work, beyond a single process.
What you can buildConnect a live operations model to incoming events, background work, and other Grid nodes.
Explore Backbone topologies03 / An extensible runtime
Bring the code the world has already written.
Open-source libraries. Your team’s proprietary algorithms. Specialized tools and services. Grid’s extension system lets you bring capabilities into a model through typed packages, with Luau, WebAssembly, and selected native execution profiles. Reuse compatible packages from ecosystems such as npm, PyPI, and crates.io, or connect to remote services through external functions.
What you can buildWrap a proven calculation or a private service once, then compose it with the rest of your model.
Connect external functions04 / Connectors & live data
The data you need is already out there.
Databases, APIs, webhooks, event streams, and the systems unique to your business. Query through external functions or feed changes into declared model bindings through connector ingress. Build an adapter for a new source and let Grid carry its updates into dependent calculations.
What you can buildCombine database results, live telemetry, and a SaaS API in one reactive application.
Connect a source to a model05 / Disconnected deployments
Build for the places the network doesn’t reach.
A factory floor, a remote site, a field operation. Put the runtime beside the work and provision the code and data it needs locally. For deployments with denied, disrupted, intermittent, or limited connectivity (DDIL), selected Backbone configurations support bounded authority exchange and convergence of declared input state when peers reconnect.
What you can buildDesign a field-planning tool around local computation and explicitly supported synchronization. Remote APIs and live feeds still need their own connectivity strategy.
Plan a disconnected deploymentPut it together
A live product,
built around your expertise.
Imagine a demand-planning app: sales arrive from your database, inventory changes stream in, and your forecasting library computes what to order. Grid keeps the dependent model values current and exposes them to your app.
- 01 / ConnectSales + inventory
Queries and connector events feed the model.
- 02 / ComputeYour forecast
An extension or external function adds your expertise.
- 03 / DeliverA live application
Serve results through the API and subscribe to changes.
From possibility to production
Start small.
Build outward.
Run one model, connect one source, then add the capabilities your product needs. Runtime features and extension support depend on the selected build and deployment. The guides cover the exact setup, including optional Backbone topologies.