Categorize 10,000 rows of a CSV with AI, using only your categories
Upload the CSV, paste your category list, and every row comes back with one category from that list in a new column. Rows are sent in blocks of 40 and each answer is matched to your list before it is written, so a 10,000-row file ends with the same labels from the first row to the last. The first rows are free without an account; the whole file, up to 20,000 rows, is $5, paid once.
- One new column, Category, filled from your list and written in your spelling
- An answer that is not on the list is asked again, then left empty and listed
- Up to 20,000 rows and 6,000,000 characters for one payment of $5
- Files of up to 100 rows are free in full
- Your other columns are copied back unchanged
Categorizing a long file: chatbot or this tool
The judgment for one row is easy for any assistant. The hard part is making it 10,000 times with the same list and writing it back into the file.
| Your situation | A chatbot | This tool |
|---|---|---|
| Thirty rows pasted into the chat | Usually categorizes them well | Does the same and returns your file |
| A 10,000-row export | The reply runs out of room and tends to stop or ask you to continue | Sends 40 rows at a time until every row has an answer |
| A list of 25 store categories | New labels and spellings creep in over a long list | Only labels from your list, matched and rewritten in your spelling |
| A row with almost no text | Often guesses a category anyway | Leaves it empty and lists the row number |
How the categories stay consistent
Your list is sent with every block of 40 rows, and the model is told to answer with exactly one value from it. Each answer is then compared with the list, ignoring capital letters, spaces and quote marks, and written back the way you spelled it: "home & garden" from the model becomes "Home & Garden" if that is your entry. Anything that matches nothing on the list is asked again once with a note, and if it still does not match, the cell stays empty.
Empty is deliberate. A guessed category in row 7,412 is hard to find later; an empty cell is one filter away. The warnings CSV lists every such row with the reason.
Writing a list that works
- Keep each category short and distinct: "Kitchen" and "Cookware" side by side invite coin flips
- Add a catch-all such as "Other" if some rows truly fit nothing
- Put rules in the instruction, such as "mugs and cups are Kitchen, not Dining"
- Name the columns to read if the file has internal codes the model does not need
How the tool works, what it costs and how your files are handled is on the AI for Every Row page; prices for every tool are on the pricing page.
Last updated
Questions and answers
Can ChatGPT categorize 10,000 rows?
It categorizes a pasted sample well, but 10,000 answers do not fit in one reply, so it tends to stop partway or switch to rough keyword rules. This tool sends the rows in blocks of 40 and checks every answer against your list.
What if I do not have a category list?
Leave the list empty and the model picks short category names itself and is told to reuse them. For a report you will compare over time, a fixed list gives steadier results.
How many rows fit in one payment?
Up to 20,000 rows and 6,000,000 characters of input for $5. Unlimited, $19 a week, covers files of up to 100,000 rows.
Which columns does the model read?
Every column except IDs, links and images, unless you name the columns to read. Long cells are shortened to their first 1,000 characters for the model; your file keeps the full text.
Is it free to try?
Yes. Files of up to 100 rows are free in full, and the first rows of a bigger file are categorized free on screen, with no account.