Grist

@Grist show me all records in the 'Customers' table where status is 'Active' and sort by last_contact_date

I've retrieved 47 active customer records from your Grist document. The most recently contacted customers include: Acme Corp (contacted 2 days ago, $45K ARR), TechStart Inc (contacted 5 days ago, $28K ARR), and BuildRight LLC (contacted 1 week ago, $52K ARR). The oldest contact was DataFlow Systems at 89 days ago. Would you like me to flag customers who haven't been contacted in over 60 days, or export this filtered view?

This example demonstrates Grist's read capabilities for structured data queries. Requires read access to the specific document. Grist excels at spreadsheet-like filtering and sorting, but complex joins across multiple tables may require multiple tool calls. Response time depends on table size—documents with 10K+ rows may take several seconds.

@Grist add a new project to the 'Projects' table: name 'Website Redesign', owner 'Sarah Chen', budget $85000, start date today

I've added the new project record to your Projects table (row ID: 142). The record includes: Project Name: 'Website Redesign', Owner: 'Sarah Chen', Budget: $85,000, Start Date: 2025-01-15, Status: (set to default 'Planning'). The record is now visible in your Grist document and any linked views. Should I also create related task records in your Tasks table, or set up a webhook to notify your team channel?

This write operation shows Grist's ability to insert structured records with proper data types. The MCP will validate column names and data formats before insertion—mismatched types will fail. Side-effect: this immediately modifies your live Grist document. If your table has formula columns or triggers, they'll recalculate automatically. Always verify table and column IDs first.

@Grist review all tasks in the 'Sprint_Tasks' table due this week and draft a status summary grouped by assignee

I've analyzed 23 tasks due this week across your Sprint_Tasks table. Here's the breakdown by assignee: **Marcus** (8 tasks, 5 in-progress, 2 blocked, 1 complete), **Priya** (9 tasks, 7 in-progress, 2 complete), **Jordan** (6 tasks, 4 in-progress, 2 not-started). Blockers: Marcus has 2 tasks waiting on API documentation; Priya's tasks are on track. Recommendation: prioritize unblocking Marcus's dependencies and check in with Jordan on the 2 not-started items before Friday standup.

This synthesis example combines Grist's data retrieval with AI reasoning to produce actionable insights. The MCP fetches raw records, then the AI aggregates and interprets patterns. Accuracy depends on your table's status field values matching common conventions. For real-time dashboards, consider Grist's native summary tables instead—this approach is best for ad-hoc analysis.

When Grist beats Airtable for support triage at 10-person scale

A 10-person support team routes 200 inbound tickets daily across three product lines. They need a shared view where anyone can reassign, tag, and update status without fighting over a spreadsheet. Grist wins here because the 32-tool MCP gives you full CRUD on records, tables, and webhooks—your AI can add tickets from Slack, update priority mid-standup, and trigger alerts when SLA thresholds hit. The API_KEY auth means one shared credential for the whole team, no per-seat OAuth dance. The trade-off: if you're already deep in Airtable's automation ecosystem, migration cost outweighs the MCP advantage. But if you're starting fresh or your current tool can't handle programmatic writes at this volume, Grist's MCP is the faster path to a working triage board.

Why Grist's MCP works for async backlog refinement

A 5-person product team grooms 40 backlog items weekly, splitting work across two time zones. They need to bulk-update story points, reassign owners, and archive completed epics without scheduling a synchronous call. Grist's MCP handles this because the Create Table and Add Records tools let your AI scaffold new sprint views on the fly, while Delete Grist Table Records cleans up stale rows as you go. The 32-tool surface means you're not hitting API limits when processing batch updates—critical when you're refining 15 stories in one conversation. The boundary: if your backlog lives in Jira and you need two-way sync with engineering tickets, Grist becomes a shadow system. Use this when your backlog is lightweight and your team prefers a spreadsheet mental model over heavyweight project tools.

When Grist's webhook tooling accelerates CRM hygiene

A 3-person sales team maintains 120 active deals in a homegrown CRM built on Grist. Every Friday they deduplicate contacts, archive dead leads, and flag stale opportunities older than 60 days. The MCP's Create Document Webhook tool lets your AI listen for new deal entries and auto-tag them by source, while Delete Column and Delete Attachment prune obsolete fields as your pipeline schema evolves. The 32-tool count matters here because sales ops is messy—you're constantly tweaking table structure, not just reading data. The catch: if your CRM is Salesforce or HubSpot, this MCP is irrelevant; it only works if Grist is your system of record. But for teams running lean with a Grist-native pipeline, the MCP turns weekly cleanup from a 90-minute slog into a 10-minute AI-assisted sweep.