Use Cases
ULTIMATE
Common products built on the server: what each needs, which pieces provide it, and what you write yourself.
A spreadsheet inside a SaaS product
The need. Customers work on spreadsheets together inside your application, and each customer only sees its own data.
The pieces.
- The core with an adapter on the database the product already uses.
- Authentication hooks that read your JWT: the subject is the user, a claim is the tenant, and
beforeLoadrefuses documents of other tenants. - Sharing for invitations inside a tenant, with read-only and editor levels.
- Scaling by document guid when the customer base grows past one process.
What you write. The tenant check in beforeLoad, the document list in your own UI, and the mapping from your users to document owners.
Why it works. Tenancy lives in your hooks and your database, so it follows the rules your product already enforces, with no second permission system.
Internal tools that replace spreadsheet files
The need. Teams email spreadsheet files and versions diverge. The company wants one live copy, with audit and backups.
The pieces.
- The core and an adapter on an internal database; the Intrasheets interface if users expect Excel.
- Snapshots for point-in-time copies and restore.
- Comments for review threads on cells.
- The REST API for the jobs that used to email the file: nightly imports, exports to the data warehouse.
What you write. Single sign-on in the hooks, and the jobs that feed or read the documents.
Planning, budgeting and forecasting
The need. Finance, sales and operations build plans together: each department fills its part, a consolidated view sums them, and the numbers must be right.
The pieces.
- Server-side formulas, so a consolidated total read through the API is the calculated value, not a stale cache.
beforeChangeto lock regions: a department edits its own rows, the totals are read-only.- The REST API to push actuals from accounting systems into the plan every night.
- The AI agent for "explain the variance in Q3" or "add a scenario with 10% lower revenue".
What you write. The region rules in beforeChange, and the actuals import.
Data collection
The need. Field staff, customers or suppliers submit data that should land in a structured sheet the team works on.
The pieces.
- Formify to publish a form from a worksheet: submissions are written as rows.
- Validations and dropdowns in the worksheet itself, enforced on every client.
- The REST API for systems that submit without a form.
What you write. Who may submit, and a rate limit on public forms.
Regulated and air-gapped environments
The need. Finance, healthcare, defense or the public sector need a collaborative spreadsheet, and data cannot leave their network.
The pieces.
- A self-hosted Node.js server and the database they already certify.
- The browser libraries served from their own hosts.
- AI, when allowed, through the PROMPT formula pointed at a model on their own network; nothing reaches an external provider.
What you write. Deployment, and the identity integration with their directory.
Why it works. Nothing in the architecture requires a cloud service: the server, the storage, the identity and the model are all theirs.
AI-assisted analysis
The need. Users ask questions and give instructions in plain language: summarize, clean, restructure, chart.
The pieces.
- The AI agent for a chat inside the product, editing the live document with tools.
- Usage tracking and budgets to control spend per user or plan.
- The MCP server so power users can work on the same documents from Claude Code or Claude Desktop.
What you write. The budget policy and the chat UI, or use the one in Intrasheets.
Why it works. The AI works on the live document under the user's permissions, and every edit reaches the other users as it happens. See AI with Jspreadsheet Server.
Automation and integration
The need. Data comes from other systems: prices update from an ERP, a CRM feeds a pipeline sheet, a script reconciles two tabs every hour.
The pieces.
- The REST API: every worksheet module, from values and rows to styles, validations and media.
- The client extension for scripts that need the realtime channel instead of request and response.
- MCP for automations built on AI agent frameworks.
What you write. The integration jobs, with a service token your hooks recognize.
Why it works. A write from a job is an ordinary operation: users with the document open see it arrive, and it persists like any edit.
Choosing the pieces
| If the product needs | Add |
|---|---|
| Several people on one document | The core and an adapter |
| Backend access | @jspreadsheet/server-api |
| An assistant for your users | @jspreadsheet/server-agent |
| AI clients such as Claude Code | @jspreadsheet/mcp |
| Excel-like interface | intrasheets in the browser |
| More than one process | Guid-routed instances |
Start with Getting started, or read Why Jspreadsheet Server for what the server takes off your plate.