Google Sheets as a data source for AI agents

Published: December 13, 2025

google-sheetsdata-sourcesautomationno-code

Feature overview

The new data source type lets you connect Google Sheets as a searchable database. Agents automatically find matching rows, show results to users, and save selected records to leads — all without writing a single line of code.

Unlike traditional API data sources, Google Sheets doesn't require endpoint development or server configuration. Simply share a spreadsheet link, specify column names — and the data becomes available for intelligent search.

Business applications

Educational platforms and corporate training

Store your course catalog, trainings, and webinars in a Google Sheet. The agent finds relevant courses for queries like "Docker courses", "network technology training", automatically enrolls employees, and sends data to your LMS.

Use case example:

Columns: Course Name, Module, Session Type, Duration, Format
Agent searches by keywords (SDN, DevOps, NoSQL)
On enrollment, saves full object: name, format, duration — all available for integrations

Product and service sales from price lists

Upload your product catalog to a Google Sheet. When a customer asks "do you have RTX 4070 laptops", the agent instantly finds matches, shows specs, prices, and SKUs, then passes data to CRM integration for order processing.

Benefits:

Update pricing through Google Sheets without code access
Data cache refreshes automatically (1-hour TTL)
Search with typos and transliteration (user types "nokiya" → finds "Nokia")

FAQ knowledge base and documentation

Instead of manually populating the Q&A knowledge base, import a ready-made spreadsheet with questions and answers. The agent searches for relevant entries and supplements responses with context from other sources.

Scenario:

Columns: Question, Answer, Category, Tags
Agent searches by semantics (substring + keyword matching)
Finds multiple close matches and synthesizes final response

Booking and scheduling

Publish service schedules, specialists, or conference rooms in a Google Sheet. The agent shows available slots, books the selected option, and records in Google Calendar or sends email confirmation.

Detailed instructions

Step 1: Prepare your Google Sheet

Create a spreadsheet in Google Sheets
Fill with data:
- First row — column names (ID, Name, Price, Category)
- From second row — actual data
Publish the spreadsheet:
- Click Share settings → Anyone with the link
- Or: File → Share → Publish to web
Copy the sharing link

Important: The spreadsheet must be public (Reader access). Don't place sensitive data in public sheets!

Step 2: Add a data source

Navigate to Dashboard → Data Sources → Add Source
Select type Google Sheets
Fill the form:

Internal name: for example, products or services

Description: "Company product and service catalog"

Sheet link: Paste the copied link or spreadsheet ID

Select sheet: Choose the needed sheet (tab) from dropdown
Configure field names (3 modes):

○ First row contains field names (recommended)
- System automatically loads the first row
- Normalizes names: "Course Name" → course_name
- Supports Cyrillic (transliteration)
- Duplicates get indices: name, name_2, name_3
○ Specify names manually
- Enter comma-separated or one per line
- Example: id, name, price, category
- System normalizes to JSON-compatible keys
○ No names
- Automatically: field_0, field_1, field_2...
Row identifier (how system distinguishes records):

Fields from table:
- id — if exists in table (auto-selected)
- email, name — any unique field
Technical values:
- [rowindex] ⚠️ Row number (may break if order changes)
- [field_0] First field (if names not defined)
Click Refresh to reload current data
Test via Test button — see table with first 20 rows

Step 3: Connect to an agent

Open your agent in Dashboard → Agents
Go to Form fields section
Add a new field:
- Name: for example, product(source) — use your source name
- Type: select type with _pick suffix (created automatically)
- Description: "Selected product from catalog"
Save the agent

Step 4: Test in chat

Open chat with the agent and try:

User: Show me laptops under $1000
Agent: Found products:
       1. HP Laptop 15 - $850
       2. ASUS X515 - $750
       ...
User: I'll take the first one
Agent: You selected "HP Laptop 15" for $850

Step 5: Configure integrations

Data from Google Sheets can automatically flow to:

CRM systems (Bitrix24, Salesforce) — full selected row object
Email — course, product, or service details
Telegram — enrollment notifications
Database via webhook

Learn more in Integrations that deliver ready leads.

Technical features

Caching and performance

Data cached for 1 hour (TTL 3600 seconds)
First request loads the sheet, subsequent — instant
LRU eviction when plan limit exceeded
"Refresh" button forces data reload

Cache limits by plan:

Free: 10 MB
Basic: 100 MB
Pro: 500 MB
Max: Unlimited

Benefits of smart search

Finds even with typos

Agent understands queries with errors:

Customer types "Nokiya" or "Nok1a" — finds Nokia products
"Laptopp" — finds laptops
"Consulttion" — finds consultation

This saves operator time who previously manually corrected typos.

Works with non-English names

Spreadsheets with Cyrillic headers work automatically:

Columns "Имя", "Телефон", "Email" — all recognized
Can mix Cyrillic and Latin names
Duplicate columns handled automatically (get numbers)

Quick updates without a programmer

Changed prices in Google Sheet? Data updates for the agent within an hour automatically. No need to:

Call a developer
Change code
Restart system

Just edit the spreadsheet — done!

Caching

Data updates every hour automatically
"Refresh" button loads changes instantly
Cache limits: from 10 MB (Free) to unlimited (Max)

How search works

System analyzes all spreadsheet columns and finds matches:

Exact matches shown first
Partial matches (word in description) follow
Considers number of matched words in query

Data storage

When selecting, all fields of the chosen row are saved:

Name, price, category, description
Available for integrations and reports
Data doesn't disappear even if spreadsheet deleted

Comparison with other sources

Feature	Google Sheets	Basic API	Google Calendar
Setup	No code	Requires API	OAuth authorization
Updates	Edit spreadsheet	API changes	Auto-sync
Search	Smart search with typos	Via API	By dates
For whom	Any user	Developers	Meeting booking

Use cases and examples

Example 1: Electronics e-commerce

Create a product spreadsheet:

SKU	Name	Price	Stock	Category
SKU-1001	HP Laptop 15	$850	5 pcs	Electronics
SKU-1002	Logitech MX Mouse	$65	12 pcs	Accessories

Scenario:

Customer asks: "Do you have laptops under $1000?"

Agent finds and responds: "Found HP Laptop 15 for $850, 5 in stock"

Customer: "I'll take it"

Agent saves selection and sends to CRM SKU, name, price — ready order for manager.

Example 2: Clinic or beauty salon

Manage specialist schedule in Google Sheet:

Specialist	Service	Day	Time	Price
John Smith	Consultation	Monday	10:00	$50
Mary Johnson	Diagnostics	Tuesday	14:00	$40

Scenario:

Customer asks: "When can I see John Smith?"

Agent responds: "John Smith provides consultations on Mondays at 10:00, price $50"

Customer: "Book me"

Agent saves booking with specialist, service, time, and price.

Example 3: Real estate agency

Manage property database in Google Sheet:

Address	Bedrooms	Area	Price ($M)	Subway
123 Main St	2	700 sq	0.85	Central
456 Park Ave	3	950 sq	1.2	Uptown

Scenario:

Customer asks: "2-bedroom near subway under $1M"

Agent finds: "123 Main St - 2BR 700 sq ft, $850K, Central subway"

Customer: "Perfect"

Agent saves selection with address, area, price — manager gets ready lead for call.

Security and limits

Public vs private spreadsheets

⚠️ Current version works only with public spreadsheets.

Don't place in public sheets:

Personal data (names, passports, addresses)
Financial information
Trade secrets
Confidential contracts

For private data use Basic API with authorization on your server.

Planned: OAuth authorization for private sheets (roadmap Q1 2025).

Google Sheets API quotas

Google Sheets API has quotas:

300 requests per minute per user
100 requests per minute per project

Our system is optimized:

Caching reduces Google API requests
Load entire sheet in one request
1-hour TTL = maximum 24 requests per day per sheet

Plan limits

Cache volume management in Dashboard → Settings → Plan:

Free: 10 MB cache (≈5000 rows)
Basic: 100 MB (≈50000 rows)
Pro: 500 MB (≈250000 rows)
Max: Unlimited

When limit exceeded, LRU eviction applies — oldest (least recently used) caches are removed.

Troubleshooting

Error "Spreadsheet not found or not public"

Solution:

Verify spreadsheet is published (see Step 1)
Ensure link is correct
Try opening link in browser incognito mode

Error "Method doesn't allow unregistered callers"

Cause: Server technical settings

Solution: Contact your system administrator or support

Search doesn't find data

Solution:

Check "Field names" settings — are columns defined correctly
Click Refresh button to reload data
Try Test — ensure data loaded
Use more specific queries

Wrong course/item saved

Cause: Agent may confuse search results

Solution:

Improve row descriptions in spreadsheet (add unique details)
Use specific queries ("Docker containers" instead of "containerization")
When selecting, state exact name from results

Best practices

Spreadsheet structure

✅ Good:

First row = column headers
Unique column names
All important fields filled
Single data type per column

❌ Bad:

Empty rows at the beginning
Merged cells
Formulas in headers
Duplicate column names (without indexing)

Naming conventions

For better performance, name columns clearly:

id — unique identifier
name/title — title
category/type — category
price — price
description — description

System auto-normalizes, but clear names simplify integrations.

Search optimization

Fast search:

Short queries (1-5 words): "SDN", "Docker", "Cassandra"
Substring match gives instant results

Precise search:

Long queries: "Evaluate SDN/NFV application in networks"
Keyword extraction finds rows with maximum matches

Fuzzy search:

Typos handled: "Dokcer" → "Docker"
Transliteration: "telefon" → "телефон"

Integration with other features

Combined with Q&A base

Use Google Sheets for structured data + Q&A base for text answers. Agent searches for course in spreadsheet, then supplements description from knowledge base.

Linked with Google Calendar

Store service catalog in Google Sheets, available slots in Google Calendar. Agent first shows services from spreadsheet, then offers to select time from calendar.

Multi-source agents

One agent can use multiple sources:

courses(source) — Google Sheets with courses
teachers(source) — Google Sheets with instructors
schedule(source) — Google Calendar with slots

Agent automatically calls the right source depending on dialog context.

Google Sheets as a data source for AI agents

Feature overview

Business applications

Educational platforms and corporate training

Product and service sales from price lists

FAQ knowledge base and documentation

Booking and scheduling

Detailed instructions

Step 1: Prepare your Google Sheet

Step 2: Add a data source

Step 3: Connect to an agent

Step 4: Test in chat

Step 5: Configure integrations

Technical features

Caching and performance

Benefits of smart search

Finds even with typos

Works with non-English names

Quick updates without a programmer

Caching

How search works

Data storage

Comparison with other sources

Use cases and examples

Example 1: Electronics e-commerce

Example 2: Clinic or beauty salon

Example 3: Real estate agency

Security and limits

Public vs private spreadsheets

Google Sheets API quotas

Plan limits

Troubleshooting

Error "Spreadsheet not found or not public"

Error "Method doesn't allow unregistered callers"

Search doesn't find data

Wrong course/item saved

Best practices

Spreadsheet structure

Naming conventions

Search optimization

Integration with other features

Combined with Q&A base

Linked with Google Calendar

Multi-source agents

Related articles and resources

Internal resources

External resources