Once your Browse AI data is in Google Sheets, you can analyze and categorize it right there using Google's built-in =AI() function — no code, no API keys, and no extra tools. This is often the missing "last step": you've scraped the data, it's sitting in a spreadsheet, and you want to turn it into something useful like categories, sentiment labels, or short summaries. The =AI() function does exactly that from a plain-English prompt.

📖 The AI function is a Google Workspace feature, not a Browse AI feature. It requires an eligible Google Workspace or Google AI plan with Gemini turned on. If you'd rather not depend on a Google plan, Browse AI's built-in Formula AI handles many of the same tasks inside your table, and our Claude enrichment guide covers automated, large-scale enrichment.

What you can do

The =AI() function works on the data already in your sheet. Common uses:

Use case	Example
Categorization	Tag each scraped item with a type or label (e.g., New, Potential, Active, Old)
Sentiment analysis	Classify scraped reviews or comments as Positive, Neutral, or Negative
Summarization	Condense a long scraped description into a one-line summary
Entity extraction	Pull names, locations, prices, or dates out of a free-text field
Standardization	Normalize messy values into a consistent format
Flagging	Mark rows that match a condition (e.g., "Does this mention a discount? Yes/No")

Before you start

Your Browse AI data needs to be in a Google Sheet. See How to sync scraped data with Google Sheets.
You need an eligible Google Workspace or Google AI plan with Gemini features enabled. If you don't see results when you use the function, your account may not have access yet.

Step 1: Get your Browse AI data into Google Sheets

Connect your robot to Google Sheets so each scraped row lands in your spreadsheet automatically. Follow our Google Sheets sync guide. Once data is flowing in, you'll have columns like Title, Description, or Price to work from.

Step 2: Add a column and write your first AI formula

Add a new column next to your data — this is where the AI output will go. In the first data row of that column, type a formula that describes what you want and points to the cell to analyze. The syntax is:

=AI("your instruction", cell_or_range)

For example, to classify a scraped review in cell D2 by sentiment:

=AI("Classify this customer review as Positive, Neutral, or Negative. Respond with one word only.", D2)

You can reference more than one cell to give the AI more context:

=AI("Based on the title and description, categorize this listing as New, Potential, Active, or Old.", B2, C2)

📖 =AI() and =GEMINI() are the same function. Use whichever you prefer.

Step 3: Generate the result

After entering the formula, the cell shows a preview of the prompt. Click Generate and insert to run it. Google sends your instruction and the referenced cells to Gemini and writes the result into the cell.

Step 4: Apply it down the whole column

To categorize every row, you don't need to retype the formula:

Select the cell with your formula and drag the fill handle (the small square at the bottom-right of the cell) down the column, or select the full range of cells you want to fill.
With the cells selected, click Generate and insert to process them as a batch.

You can generate up to 200 cells at a time. Once the first batch finishes, select the next set and repeat for larger datasets.

⚠️ AI results are static. Unlike normal spreadsheet formulas, values produced by =AI() don't recalculate when your source data changes. If Browse AI syncs new or updated rows, run Generate and insert again on those rows to categorize them. For data that refreshes on a schedule (for example, from a monitor), plan to regenerate periodically, or use an automated method like the Claude enrichment guide instead.

Writing prompts that give reliable results

The quality of your output depends on how clearly you describe the task. A few guidelines:

List the exact options. For categorization, spell out the allowed values: "Respond with exactly one of: New, Potential, Active, Old." This keeps the output consistent and easy to filter.
Ask for the format you want. "Respond with one word only" or "Return a single number from 1 to 5" prevents long-winded answers.
Reference the right cells. Point the formula at the specific columns that hold the information needed to make the decision.
Keep it to one task per formula. If you need both a category and a summary, use two columns rather than asking for both at once.

✅ Tip: For categorization, mirror the buckets you already use elsewhere (including your Browse AI columns) so the AI-generated labels line up with the rest of your data. Adding "If none apply, respond Unknown" gives the AI a safe fallback instead of guessing.

Example prompts

Goal	Formula
Categorize a product	`=AI("Categorize this product as Electronics, Clothing, Home, or Other.", B2)`
Score sentiment	`=AI("Rate the sentiment of this review from 1 (very negative) to 5 (very positive). Return only the number.", D2)`
Summarize	`=AI("Summarize this description in one sentence under 15 words.", C2)`
Extract a value	`=AI("Extract the city name from this address. If none, respond Unknown.", E2)`
Flag a condition	`=AI("Does this listing mention free shipping? Answer Yes or No.", C2)`

When to use this vs. other methods

Method	Best for	Notes
Google Sheets `=AI()` (this guide)	Quick, no-code categorization and summaries on data already in a sheet	Results are static; needs an eligible Google plan
Browse AI Formula AI	Transforming and labeling data inside your Browse AI table	Built in, no external tools — see the Formula AI guide
Claude via Zapier / API	Automated enrichment that runs every time data is scraped	Scales to thousands of rows and refreshes automatically — see the Claude enrichment guide