Keep the cells. Add a language.
Start with typed formulas, then add trust and optimization without giving up the spatial model.
Learn Grid
Build the model.
Check the result.
Follow a complete guided build with runnable source, exact checkpoints, and a next step. Use the canonical reference when you need the language contract behind it.
Continue learning · this browser
Finding your next unfinished build…
Completed builds stay in local browser storage. Your recommendation appears as soon as that progress is ready.
All guided buildsProgress stays in this browser
Browser-local progressChecking completed builds
Guided builds
One useful outcome at a time.
Every build includes prerequisites, complete steps, expected values, a challenge, and related reference.
01020304050607080910111213141516171819202122
Build your first reactive model
Turn typed source cells into a live profitability model, then prove the dependency graph by changing one value.
Runtime setup: Grid source editor or model authoring surface · No connector or network access required
Developer quickstart: model to dashboard
Load a portable revenue model, follow its named outputs into a Layout surface, and prove the UI stays synchronized after one source change.
Runtime setup: Grid source editor or model authoring surface · Built-in Layout surface · No connector or network access required
From arrays to data pipelines
Turn paired series into cumulative and scenario analytics, then choose native collections when the job needs keyed, unique, or queued state.
Runtime setup: Grid source editor or model authoring surface · No connector or network access required
Debug and harden a model
Separate failed business checks from error values and compiler diagnostics, then add explicit recovery and guarded write contracts.
Runtime setup: Grid source editor or model authoring surface · Diagnostics or Problems view · No connector or network access required
Build reusable domain logic
Extract repeated calculations into inferred named functions, then model a closed order lifecycle with payload-bearing choices and exhaustive matching.
Runtime setup: Grid source editor or model authoring surface · Basic familiarity with MAP, REDUCE, and MATCH · No connector or network access required
Build your first Grid app
Turn an approval model into a routed App surface, bind explicit state, add a guarded action, test real UI states, and publish the result as App Home.
Runtime setup: Grid 0.65.0 candidate source with the App surface enabled by deployment policy · Permission to edit and deploy a model · A host that exposes App runtime and Publish Grid App actions · No connector or network access required
External enrichment, safely
Combine API-backed signals and remote scoring while keeping downstream decisions computable through pending work and failures.
Runtime setup: Grid runtime with async workers enabled · Network access for live FX results · Configured ml.scoring route for live ML_SCORE results
Run the complete model API lifecycle
Move one model from editable authoring to protected operation, with exact evaluation, compare-and-swap replacement, stale-write diagnosis, source-lock proof, rename, and cleanup.
Runtime setup: Running Grid Server Build with Node and Runtime Host ready · Public API bearer token · curl and jq
Follow one model over realtime events
Subscribe over SSE, adopt the initial snapshot, order revisions, detect stale events, and reconnect through explicit resynchronization.
Runtime setup: Running Grid Server Build with a ready model API · Public API bearer token · curl, jq, and two terminal windows
Carry a connector event into a model
Send pinned ingress fixtures through an operator-owned binding and prove one declared input drives two downstream outputs.
Runtime setup: Running Grid Server Build and operator-configured delivery binding · Separate public API and connector-ingress bearer tokens · curl and jq
Build a live revenue dashboard
Filter, join, group, and present changing order data as KPI cells and a purpose-built Layout surface.
Runtime setup: Durable model tables · Layout surface · Table editor, import, connector, or mutation path
Take a live dashboard through connector handoff
Continue the live revenue build at its source boundary: admit one connector event, prove the existing dashboard settles exactly once, and prepare its production handoff.
Runtime setup: Completed live-revenue-dashboard build deployed under a known GRID_MODEL_ID · Operator-configured Orders upsert binding · Separate public API and connector-ingress bearer tokens · curl and jq
Automate operations with rules and schedules
React to conditions, retain deliberate operational state, and choose explicit timing and missed-run behavior.
Runtime setup: Grid source editor or model authoring surface · Resident Grid runtime for wall-clock schedules · No connector or network access required
Verify a canonical example
Move one portable model through source identity, compile/deploy, actual assertions, and a provenance-bound runtime receipt.
Runtime setup: Node.js 20 or newer and jq for source-only checks · Running Grid Server Build, curl, and a public API token for runtime checks · Operator-attested identity for the exact runtime build
Explain and validate a decision
Trace a decision to its inputs, compare revisions, quantify uncertainty, and reject invalid writes.
Runtime setup: Saved model revisions for history · Cell inspector or equivalent Why action · No network access required
Build an explainable policy graph
Combine typed graph structure, recursive policy, contested evidence, and dependency witnesses into one auditable release decision.
Runtime setup: Predicate and Graph V2 runtime · WHY inspector or gridctl · Complete source bundle for the imported governance module
Turn a model into an app
Extend one CRM model with Table, Kanban, and Layout, then compare Map and custom Component patterns honestly.
Runtime setup: Built-in workbook surfaces · Trusted custom UI only for inspecting the Component pattern · Basic JSX helps for that comparison
Optimize a pricing decision
Use target solving and bounded optimization to recommend a price and production run without mutating inputs.
Runtime setup: Grid solver functions · No connector, network, or surface required
Run a warehouse simulation
Advance model time through seasonal demand, smoothed memory, delayed arrivals, stock flows, and replayable trajectories.
Runtime setup: Simulation runner and model history · No external data source required
Promote a model to production
Turn process, credential, storage, health, logging, capacity, compatibility, and rollback evidence into one fail-closed promotion gate.
Runtime setup: Existing staging Grid Linux Server deployment · Operator access to configuration, state, logs, metrics, and traffic controls · Node.js 20 or newer, curl, and jq
Take an App Builder contract to React or Vue
Replace the generated renderer while preserving the App's exact model reads, writes, routes, error behavior, and capability boundary.
Runtime setup: A completed and published Grid App Builder package · A package workspace or host workflow that can build and republish the model-owned UI · React or Vue build tooling for the chosen entry
Publish a signed Library to grid ns
Scaffold and test a Luau Library, prove control of a publisher namespace, sign the exact release locally, and publish an immutable public coordinate.
Runtime setup: Grid 0.61.0 or a currently supported Grid Core release with Luau extension tooling · OpenSSL with Ed25519 support · A unique lowercase publisher namespace · A private directory outside version control for the publisher key and token · HTTPS access to publish.grids365.org and grids365.org · Authority to create a durable public namespace and release
Watch Grid work
See one edit become a live model.
40 release-pinned films connect each visible result to a guide, runnable model, or documentation. Every player appears only after its media, captions, and source evidence pass verification.
Explore the film seriesFrom build to distribution
Sign it. Verify it. Publish one exact release.
Take a real Luau Library from generated fixture to an immutable grid nscoordinate—with offline signing and the current publisher-review boundary made explicit.
Test locallyDry-run offlineEnroll safelyVerify the digest
Learning tracks
Bring the mental model you already have.
Treat the model as the backend.
Learn the model, data, error, and integration contracts before choosing built-in App surfaces or a custom interface.
- Build your first reactive model
- Developer quickstart: model to dashboard
- From arrays to data pipelines
- Debug and harden a model
- Build reusable domain logic
- External enrichment, safely
- Run the complete model API lifecycle
- Follow one model over realtime events
- Carry a connector event into a model
- Build a live revenue dashboard
- Take a live dashboard through connector handoff
- Build your first Grid app
- Turn a model into an app
- Promote a model to production
- Take an App Builder contract to React or Vue
- Publish a signed Library to grid ns
Generate against one exact contract.
Move from canonical source and diagnostics to repair, verification, narrow bindings, and evidence-backed generated applications.
Move from evidence to action.
Build an auditable decision, automate its operating rules, then optimize and compare repeatable experiments.
Carry one model through every boundary.
Deploy and change a protected model, adopt its event stream, deliver connector data, and verify the resulting behavior.
Make model behavior observable and recoverable.
Harden diagnostics and state changes, verify claimed behavior, then rehearse health, capacity, deployment, and rollback checks.
Go deeper
Canonical concept paths.
The exact reference remains close when a build makes you curious about the underlying capability.
Language center
Values, references, names, types, and reactive evaluation.
Open reference →Inspection and explanation
Follow dependencies and ask an evaluated model why.
Open reference →Rules and schedules
Add deliberate state changes with explicit conditions and ownership.
Open reference →Live models
Connect changing data while preserving the model dependency graph.
Open reference →Model time
Work with history, schedules, delay, state, and simulation trajectories.
Open reference →Build a Grid app
Author, test, publish, and hand off a model-bound App.
Open reference →Prefer the full map?