Airtable Canvas and your admin settings
Last updated: November 12, 2026
Plan availability | In open beta for all paid plan types |
Permissions |
|
Platform(s) | Web/Browser |
This article focuses on what Enterprise Scale and Business admins must do to enable Canvas for their organization(s). On Team plans, workspace owners must ensure that the "Turn on Airtable AI for all bases within this workspace" workspace setting is enabled.
What an admin needs to turn on
Before anyone in your organization can use Canvas, Enable Airtable AI must be turned on by an admin. If your organization restricts Airtable AI to an allowlist of workspaces, Canvas only works on bases in those workspaces — the allowlist is a sub-setting of Enable Airtable AI.
Additionally, an admin must give two consents in admin panel, under AI settings:
Allow at least one AI model that Canvas supports. Either by allowing all model developers, or, if they're restricted, by allowing models directly. Admin panel → AI settings → Restrict AI platforms & model developers.
Turn Canvas on. Admin panel → AI settings → Canvas.
Until all of these settings are enabled, Canvas will be unavailable to members of your organization.
Organizations that can't use Airtable Canvas yet
Some organizations can't use Canvas even with every consent above in place:
Organizations that are HIPAA compliant
Organizations whose data is pinned to a particular region.
Organizations that use enterprise key management (EKM)
We're working to bring Canvas to these organizations in the future.
Why a base isn't listed in the Canvas base picker
Canvas only lists bases it can build in, so an ineligible base simply doesn't appear in the picker. A base is ineligible when any of the following is true, no matter what else your organization has turned on:
The base is in a free-plan workspace. Canvas is rolling out to free plans gradually, so a free-plan workspace may not have it yet.
The base has an active sandbox.
The base is a package development base, meaning the source base a package is published from. See Managed apps - Airtable admin panel.
A workspace owner has disabled Airtable AI for the workspace. An admin may have enabled AI for the broader organization, but a workspace owner can still fully disable Airtable AI at the workspace level, which prevents Canvas from being available.
If a base becomes ineligible after someone has already built on it, the canvas stops opening and shows "This project is no longer available."
FAQs
Does Canvas use Airtable AI credits?
Yes. Generating and customizing an Airtable Canvas project consumes Airtable AI credits, drawn from the same workspace or organization balance. Viewing and using a published Airtable Canvas project doesn't consume AI credits, so a workspace with no AI credits left can still open and use existing Airtable Canvas projects, but can't generate or change any projects until more AI credits are added to your balance. See the Airtable AI billing article for more information.
I'm an admin and I've turned everything on. Why still can't someone use Canvas?
Check the workspace first. Canvas is only gradually being rolled out to free plans and may not be available if you’re a free user. If you restrict Airtable AI to an allowlist of workspaces, the base has to be in one of them. After that, check whether your organization is HIPAA compliant, pins its data to a particular region, or uses enterprise key management.
Why can't I find my base in the Canvas base picker?
The picker only lists bases Canvas can build in, so a base that's missing is usually one it can't use. The most common reasons are that the workspace is on a free plan ( Canvas is gradually rolling out to free users), Canvas hasn't been turned on for your organization, or that your organization's AI settings don't allow Canvas’s supported AI models. Your admin can tell you which applies.
Does blocking MCP access turn off Canvas?
No. Canvas uses Airtable’s MCP internally, so the "Block MCP access" setting doesn’t affect it. That setting only blocks third-party MCP clients. You can read more about our MCP here.