Give teams and customers a conversational way to look up records and structured data stored in Airtable bases.
6 min read
Updated October 2026
Many teams use Airtable to track everything from inventory and event registrations to customer onboarding checklists. The problem is that only the people who built the base know how to find things in it. Everyone else ends up asking someone or struggling with filters.
The Airtable integration lets your AI agent query records, search across tables, and inspect the base schema. Anyone on the team (or visiting your website) can ask a natural-language question and get an immediate answer pulled from your Airtable data.
Available actions depend on your agent type and plan. Customer-facing actions serve customers. Internal team actions are for authenticated team use. AI CoPilot is Enterprise-only. Plus enables up to 5 actions per agent, Pro enables 40, and Enterprise has no enabled-action limit.
| Action | Description | Audience |
|---|---|---|
| Get Base Schema | Get schema for an Airtable base | Internal team |
| Query Records | Query records from Airtable | Internal team |
| Search Records | Search records in Airtable | Internal team |
Before connecting Airtable, make sure you have:
data.records:read and schema.bases:read| Field | Description | Example |
|---|---|---|
| API Key | Your Airtable personal access token | pat1234567.abcdef... |
After setup, open your AI Agent widget and try these realistic queries:
More test phrases:
Your personal access token may not have access to the base being queried. Edit the token at airtable.com/create/tokens and add the missing base.
Check that the table name and field names match exactly. Airtable is case-sensitive for field names. The AI uses Get Base Schema to learn your field names, so make sure that scope is enabled.
Airtable limits API requests to 5 per second per base. If you see 429 errors during heavy usage, the AI will retry automatically after a short delay.