AI Sheets App Overview

AI Sheets

1. What it is

AI Sheets is a spreadsheet app where users create, import, and collaborate on structured data — and an AI assistant answers questions about it, proposes edits for review, fills columns in bulk, and generates formulas from plain-language descriptions. Questions that involve counting, totals, or averages run as exact SQL over every row, not as model estimates.

• Enablement: AI Sheets requires a license. An admin enables it from the Apps Marketplace, then configures AI features and limits from Settings inside the app.
• What it is not: AI Sheets is not a database or a BI tool. It stores flat rows of data per sheet. It does not replace external spreadsheet software — it is for structured data that lives inside your organization, with AI layered on top.

2. Standing it up

1. Go to Admin → Apps Marketplace and enable AI Sheets.
2. Open AI Sheets → Settings (visible only to app admins). Confirm that Enable AI Features is on. Review the individual AI toggles — AI Edits, AI Column Fill, and AI Automation Actions are all on by default.
3. Decide whether to allow Public Sharing Links (off by default). When enabled, sheet owners can generate a link that anyone with the URL can view, optionally requiring authentication.
4. Set the Max Rows Per Sheet limit if the default of 5,000 is not appropriate. The value must be between 100 and 50,000.
5. Grant access to the users or groups who should see the app. Users who have access see AI Sheets in their app launcher and can create their own sheets immediately — no further configuration is required.

There is no prerequisite setup beyond enabling the app. Users can create sheets, import data, or pick a template the moment access is granted.

3. How it fits together

Sheets and rows. A sheet is a named grid with a defined column schema and a set of data rows. Each sheet belongs to the user who created it. Sheets support up to 100 columns and a configurable maximum number of rows (default 5,000). Each row’s data is capped at 64 KB.

Column types. Every column has a type: Text, Number, Currency, Date, Email, URL, Boolean, Select (dropdown with defined choices), or Formula. Formula columns are computed at read time from an expression and are never directly editable. The formula engine supports arithmetic, comparisons, logic, string functions, date functions, and aggregate functions like column totals and lookups across rows.

Collaborators and permissions. The person who creates a sheet is the owner. Owners can share sheets with other users as Viewer (read-only) or Editor (can edit rows and data). Only the owner can delete the sheet, manage sharing, create automation rules, or save it as a template. Viewers can comment on cells. The collaborator’s default permission is Viewer.

Example: Alex creates a “Q3 Hiring Pipeline” sheet with columns for Candidate, Role, Stage (a Select column with choices “Applied”, “Phone Screen”, “Interview”, “Offer”, “Hired”), Salary, and Start Date. Alex shares the sheet with two recruiters as Editors so they can update rows, and with the hiring manager as a Viewer. The hiring manager asks the AI panel: “How many candidates are at the Offer stage?” — the system runs an exact count over all rows and returns the number.

Templates. A sheet can be saved as a template by its owner. Templates appear in the Templates tab alongside built-in system templates (Project Tracker, Budget Tracker, Contact List, Inventory, Event Planning) and pre-populated example sheets (Sales Pipeline and others). Creating a sheet from a template copies the column schema and any existing rows into a new sheet.

Views. Each user can save personal views on a sheet — a named combination of filters, sort column, and sort direction. Views are per-user; they do not affect other collaborators.

Form bridge. When the Forms app is enabled, a user can create a sheet from a form template. The sheet’s columns are derived from the form’s fields. Existing submissions are backfilled as rows, and new submissions are added automatically as they come in.

Real-time collaboration. Multiple users editing the same sheet see each other’s row changes, new rows, deletions, and comments in real time. A presence indicator shows who is currently viewing the sheet.

Tags. Sheets can be tagged with free-text labels. Tags appear as filterable pills on the sheet list, scoped to sheets the user owns or that were shared with them — a tag on someone else’s private sheet is never visible.

Starring. Only the sheet owner can star a sheet. Starred sheets appear in the Starred tab for quick access.

Soft delete and trash. Deleting a sheet moves it to the Trash tab. Sheets in trash can be restored within 30 days. After 30 days, they can no longer be restored. The owner can also permanently delete a sheet from trash at any time.

4. Running it

Creating a sheet

Users can create sheets in four ways from the main sheet list:

1. Blank sheet — enter a name and optionally describe it with AI to auto-generate columns.
2. From a template — pick a system template or a saved user template.
3. Import from CSV or Excel — upload a .csv or .xlsx file. The system infers columns and imports rows.
4. Import from Google Sheets — paste a public Google Sheets URL.

When importing into an existing sheet, the AI can suggest how the file’s headers map to the sheet’s columns.

Asking questions with AI

From the AI panel on a sheet, users type a natural-language question. The AI classifies the question and responds with one of these actions:

• Aggregate (count, sum, average, min, max) — runs exact SQL over all rows and returns the computed number, optionally grouped by a column.
• Filter — shows only rows matching the criteria.
• Sort — reorders rows by a column.
• Chart — renders a bar, line, pie, or doughnut chart from the data.
• Summary — a text analysis for qualitative questions.

Proposing and applying AI edits

When a user asks the AI to change data (“set all overdue tasks to Blocked”, “add 10 rows of sample contacts”), the AI proposes the edit first — showing which rows are affected and a before/after preview. The user confirms before anything is written. Every applied AI edit is logged in the activity history and can be reverted.

Edit types include updating rows, deleting rows, adding rows, adding a column, and updating a column’s label or options.

Enriching a column with AI

AI Column Fill lets users fill or classify a column’s values using AI. The user describes what value to derive for each row (for example, “categorize the support ticket by urgency” or “extract the city from the address”). The system previews results on a sample of rows, and the user confirms. The fill then runs in the background in batches. It can target all rows or a filtered subset, and can skip rows that already have a value. Enrichment supports both existing columns and new columns created on the fly.

Managing automation rules

Sheet owners can create automation rules from the Automations tab on a sheet. Each rule has a trigger, optional conditions, and one or more actions. Up to 25 rules per sheet are allowed.

Triggers: Row created, Row updated, Row deleted, Field changed, Rows imported, or Schedule (runs at a recurring interval, minimum every 5 minutes).

Actions: Notify collaborators, Notify a user, Send email, Update fields, Create a row, HTTP request, AI fill column (on row triggers only), AI summary digest (on schedule only), or AI anomaly alert (on schedule only).

AI automation actions are gated by the AI Automation Actions admin toggle. When the toggle is off, AI action types cannot be added to new rules or saved on existing ones.

Sharing a sheet publicly

When Public Sharing Links is enabled in Settings, sheet owners can generate a read-only public link. The owner can optionally require viewers to authenticate (sign in to the organization) before viewing. The public link can be regenerated to invalidate previous links.

Duplicating a sheet

Any user with edit access can duplicate a sheet. Duplication copies the column schema and all data rows into a new sheet owned by the duplicating user. The original owner is notified. Duplication is also the mechanism behind “Use template” — using a template duplicates it and strips the template flag.

Exporting data

Sheets can be exported as CSV or Excel (.xlsx) from the sheet view. A template CSV (headers only, no data rows) can also be downloaded to use as an import template.

Cell comments

Any collaborator — including viewers — can comment on a specific cell (a row-and-column intersection). Comments support @mentions by email or name. Mentioned users, the sheet owner, and prior participants on the same cell thread are notified. Comments appear in real time for other users who have the sheet open.

Viewing activity and version history

The Activity tab on a sheet shows a log of actions: creation, edits, imports, exports, shares, duplications, column changes, AI edits, and more. The Versions tab shows a change history tracked by the version system. Editors can revert a sheet to a previous version.

Mobile

AI Sheets has a dedicated mobile view at /m/apps/ai-sheets. On phones, users see their sheet list, open a sheet in a simplified table view, star sheets, export to CSV, and read or post cell comments. The full spreadsheet grid and AI panel are desktop-only; phones redirect automatically.

5. Settings

All settings are in AI Sheets → Settings (app admin only).

Setting	Default	What it changes
Enable AI Features	On	Master switch for all AI capabilities — questions, suggestions, edits, enrichment, and AI automations. Turning this off disables every AI feature while leaving the spreadsheet fully functional.
AI Edits	On	Whether the AI can propose row and column changes (update, delete, add rows/columns). Requires the master AI switch.
AI Column Fill	On	Whether AI enrichment (bulk fill a column with AI-derived values) is available. Requires the master AI switch.
AI Automation Actions	On	Whether automation rules can include AI action types (AI fill column, AI summary digest, AI anomaly alert). Requires the master AI switch.
Enable Public Sharing Links	Off	Whether sheet owners can generate a public read-only link to a sheet.
Max Rows Per Sheet	5,000	The maximum number of data rows allowed in a single sheet. Accepts values from 100 to 50,000.

6. More help

• AI Sheets FAQ — specific setup and operating questions.
• Ask AI — the assistant answers questions about AI Sheets from these articles.