Skip to main content

Quickstart for spreadsheets

GPT for Work is an AI agent for bulk tasks in Google Sheets and Microsoft Excel. Use your favorite AIs to bulk-generate, translate, search, categorize, analyze, and otherwise bulk-process text at machine-powered scale and speed.

GPT for Work contains two spreadsheet add-ons:

  • GPT for Sheets

  • GPT for Excel

This guide gives you a quick walkthrough of the add-ons from installation to trying out the Agent.

Install the add-on

To use GPT for Work in Excel or Sheets, install the appropriate add-on:

For detailed installation instructions, see Install GPT for Work add-ons.

Excel is better than Sheets

If you have a choice between using Excel or Sheets, choose Excel. Excel is generally faster at processing data, provides a smoother user experience, and can reliably handle bigger spreadsheets (200,000 rows and above) than Sheets.

  • GPT for Excel can execute up to 1,000 prompts per minute and reliably process up to a million rows in one go.

  • GPT for Sheets can execute up to 360 prompts per minute and reliably process up to 200,000 rows in one go.

Open the add-on

  1. Create a new spreadsheet.

  2. In the menu bar, select Extensions > GPT for Sheets and Docs > Open.

    Open GPT for Sheets

    The GPT for Sheets sidebar opens.

    Spreadsheet with GPT for Sheets sidebar

Use the Agent

The Agent is an AI-powered chat interface that helps you accomplish spreadsheet tasks at scale: bulk content generation, bulk translation, bulk web search, bulk data preparation and analysis, and more. Just describe what you need in plain language, and the Agent analyzes your spreadsheet data and your intent, and automatically executes the task.

Let's have the Agent do some quick market research for us. We'll use one of the built-in examples for convenience.

  1. In the sidebar menu, select Agent.

  2. In the Try an example dropdown menu, select Market research.

    Selecting a built-in Agent example

    GPT for Sheets:

    1. Creates a new sheet prepopulated with example data. The data is a list of company names and a series of empty columns for company information that we want the Agent to fill in.

    2. Prefills the Agent chat field with a prompt that will accomplish the example task: Fill out this table. This simple prompt is enough for the Agent to fetch the missing company information from the web and fill the empty columns.

    Sheet with example data and a prompt for the Agent
  3. Click Send.

    The Agent:

    1. Analyzes the prompt and drafts a plan for accomplishing the requested task.

    2. Executes the plan. In this example, the Agent fetches the missing company information from the web and fills the empty columns B-E with what it found, one by one. Note that for column F, which does not require web search, the Agent uses gpt-4o (no web search).

    tip

    You can follow the Agent's thinking and workflow live in the sidebar.

The following video shows the entire process.

What's next