Cloud quickstart

Cloud quickstart

Cloud quickstart

Go from a provisioned tenant to a live, recomputing model in about five minutes, using nothing but curl. No repo checkout, no local toolchain — your Grid is already running.

If you want to learn the language itself, the language quickstart is the next stop after this page.

What you need

  • A provisioned tenant — visible in your dashboard with status ready.
  • Your tenant's hostname and an API token (step 1 below).
  • curl, or any HTTP client. The dashboard also generates equivalent Node.js and Python snippets.

1. Collect your connection details

Open your tenant in the dashboard and find the Connect to your Grid card. Click Reveal & paste tokens — the snippets on that card are pre-filled with your real hostname and token, and the .env tab gives you values you can export directly:

export GRID_API_BASE="https://<your-tenant-host>"
export GRID_API_TOKEN="<your-api-token>"

Token reveals are recorded in the operations log, so access stays auditable.

2. Smoke-test the tenant

Liveness needs no auth, which makes it a good first request:

curl -sS "$GRID_API_BASE/healthz"

If this responds, your Grid is up and the hostname is right. Everything below authenticates with the bearer token.

3. Deploy your first model

A model is deployed by POSTing plain Grid source — the same text you would write in a .grid file. Deploys are idempotent on the model id, so re-running this replaces the model rather than duplicating it:

curl -sS "$GRID_API_BASE/api/models" \
  -H "Authorization: Bearer $GRID_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "id": "hello",
    "source": "A1 = 10\nA2 = 20\nA3 = SUM(A1:A2)"
  }'

The model is live as soon as the request returns: three cells, one of them (A3) derived from the other two.

4. Push a new input

Change an input and the recompute fans out automatically — there is no separate "run" step:

curl -sS "$GRID_API_BASE/api/models/hello/input" \
  -X PUT \
  -H "Authorization: Bearer $GRID_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "symbol": "A1", "value": 99 }'

5. Read the result

Resolve any cell by name to read its current value:

curl -sS "$GRID_API_BASE/api/models/hello/resolve/A3" \
  -H "Authorization: Bearer $GRID_API_TOKEN"

This returns { "value": 119 } — the sum of the new A1 (99) and the untouched A2 (20). That round trip — deploy source, push an input, read a derived value — is the whole working loop; everything else builds on it.

Where next

  • Language quickstart — build your first real model and learn the shape of Grid source.
  • Concepts — the vocabulary: model, binding, cell, symbol, alias, region.
  • Rules & schedules — make models react to time and change, not just inputs.