Using the /ai endpoint

API Spreadsheets Team · Oct 14, 2025

Download Full Python Code

What is the AI Endpoint?

The /ai endpoint lets you perform natural-language CRUD on any spreadsheet: just describe what you want in plain English, and the API will read, create, update, or delete rows for you.

Prerequisites

• API Spreadsheets account
• Account-level API keys (access_key and secret_key)
• A spreadsheet uploaded to your account (to get a file_id and file access key and file secret key)

Where to find your account keys

Where to find your File ID and File Keys

Python Setup

# $ pip install apispreadsheets
from apispreadsheets_lib import APISpreadsheets

account_access_key = "YOUR_ACCOUNT_ACCESS_KEY"
account_secret_key = "YOUR_ACCOUNT_SECRET_KEY"

file_id         = "YOUR_FILE_ID"
file_access_key = "YOUR_FILE_ACCESS_KEY"
file_secret_key = "YOUR_FILE_SECRET_KEY"

api = APISpreadsheets(access_key=account_access_key, secret_key=account_secret_key)

Starting Spreadsheet (preview)

Here’s a trimmed view of your sheet (we’ll work with the first 5 rows and the most relevant columns):

Create: add a new restaurant row

We’ll add Mekelburg’s with a short natural-language prompt. The AI will infer columns from your data and insert a new row.

create_prompt = (
  "I found a new restaurant to add to the list. It's called Mekelburg's "
  "with a Casual Fun vibe. It's a classic Jewish deli reimagined and is "
  "moderately priced. Their must get dishes are their matzo ball soup and wings. "
  "The address is 319 Kent Ave, Brooklyn, NY 11249, United States with Latitude 40.7137358 "
  "and Longitude -73.9900738. This is the Yelp URL https://www.yelp.com/biz/mekelburgs-brooklyn-3 "
  "and I found this restaurant from the Infatuation website with article at this URL "
  "https://www.theinfatuation.com/new-york/reviews/mekelburgs-williamsburg"
)

create_ok = api.ai(
  file_id=file_id,
  prompt=create_prompt,
  access_key=file_access_key,
  secret_key=file_secret_key
)

Update: fix the restaurant URL

Now we’ll correct Mekelburg’s URL from Yelp to the official website.

update_prompt = "Change the URL for a restaurant named Mekelburg's to https://www.mekelburgs.com/."
update_ok = api.ai(
  file_id=file_id,
  prompt=update_prompt,
  access_key=file_access_key,
  secret_key=file_secret_key
)

Read: query the spreadsheet

Ask questions about your data. For example, “Show me all restaurants in the East Village.” The AI returns JSON—use it in your app or print it for debugging.

read_prompt = "Show me all restaurants in the East Village"
east_village = api.ai(
  file_id=file_id,
  prompt=read_prompt,
  access_key=file_access_key,
  secret_key=file_secret_key
)
print(east_village)  # => { "data": [...matching rows...] }

Delete: remove a row by name

Finally, remove the row we added (Mekelburg’s) by referencing its name in natural language.

delete_prompt = "Remove the restaurant named Mekelburg's from this spreadsheet."
delete_ok = api.ai(
  file_id=file_id,
  prompt=delete_prompt,
  access_key=file_access_key,
  secret_key=file_secret_key
)

Tips for reliable prompts

• Match spelling exactly for entity names you reference (e.g., Mekelburg’s vs Mekelburgs).
• Use clear column hints (e.g., include “URL”, “Address”, “Cuisine” in your prompt when relevant).
• Include only one operation per request. Make multiple requests for multiple operations
• When updating, include both an identifier (Name/Address) and the new value you want.