Skip to main content
If your business already runs on an Airtable base, the Airtable integration lets your bot work in it directly: look up a customer’s order when they ask, write a new booking as it’s taken, mark a job done. Connect it from Tools → Connectors.

​
What it does

Under the hood your bot gets three operations: query records (with filters), create records, and update records. That covers looking a record up, logging a new one, and changing one that already exists.
There’s no delete. A bot that can delete rows in your business’s live base on a customer’s say-so is a bad idea, so the integration doesn’t offer it. Remove records in Airtable yourself.

​
How to authenticate

Two routes:
Click Connect with Airtable. A popup asks you to sign in and choose which base to grant access to. Approve, and it closes itself.
If you grant access to several bases, Paige uses the first one. To connect a specific base, grant just that one — or use a token instead, where you name the base explicitly.
Paige checks the credentials with Airtable before saving, so you’ll know immediately if the token is wrong or can’t see the base.

​
How to prompt the coding agent to use it

Ask in plain language and the agent writes it — it looks up your real tables and fields first, so it uses your actual names rather than guessing:
“When someone gives me their order number, look it up in the Orders table and tell them the status.”
“Log each completed booking as a row in the Bookings table.”

​
Refresh schema

Paige takes a snapshot of your base’s structure — every table, every field — when you connect, and your bot code is written against it. If you rename a table, add a field, or restructure your base in Airtable, that snapshot goes stale. Your bot is still looking for a field that no longer exists, and starts failing. Click Refresh schema on the Airtable card to take a fresh snapshot.
Refreshing updates your preview bot straight away, but your live bot doesn’t get the new structure until you deploy. If you change your base and your live bot starts throwing field errors, refresh and deploy.
If the agent seems confused about a table or field — insisting something doesn’t exist when you can see it in Airtable — refresh the schema. It’s working from the snapshot, not from your base.

​
When it breaks

Airtable errors surface in your logs. The usual causes:
  • Field name doesn’t match — Airtable field names are case-sensitive, and Order Number is not order number.
  • Schema is stale — see above.
  • Token lost access — someone changed permissions in Airtable, or the token was revoked.
  • It’s running on a schedule — integrations don’t work in scheduled tasks at all.