Skip to content

ConsentAlly works with AI assistants. A merchant can have their assistant set up, sign in to or check their ConsentAlly account from the chat: the assistant needs only the merchant's email address, and the merchant approves it on one ConsentAlly page with a code. Nothing is created or connected before that approval. How it works: https://consentally.com/auth.md

Docs pages

Confirmation and undo

Before an AI assistant makes a change visitors would see, ConsentAlly answers with a preview and a confirmation token instead of making it. The assistant shows the person the preview, calls again with the token only after they say yes, and the account owner then gets an email with a link that undoes the change. Everything else an assistant changes, such as the banner's draft, stays out of visitors' sight until it is published.

Which changes ask first

The tools that ask the person first
ToolWhat it is forAsks the person first
publishPublish the bannerAlways
rollbackGo back to an earlier versionAlways
classify_scriptSet a script's categoryWhen the change lets a script run before the visitor chooses
Every other tool only reads, changes what visitors do not see yet (the banner's draft and its privacy rules, a new website), or makes a script wait for consent, which applies at once without asking.

How the confirmation works

  1. The assistant calls the tool. Nothing changes: the answer is a preview, in plain words, and a confirmation_token.
  2. The assistant shows the person the preview word for word and asks.
  3. After a yes, the assistant calls the same tool again with the same arguments and the token.

The token works once, for 10 minutes, and only for that tool, connection, website, arguments and preview. For publishing and rolling back it is also tied to the version that was live, so it never confirms a change over a version someone published meanwhile. A refusal never carries a new token: the assistant calls again without one for a fresh preview.

When a yes in the chat is not enough

ConsentAlly cannot see the chat, so a yes there is the assistant's word. An organisation admin who wants more turns on Require approval on ConsentAlly under Connected assistants: each of these changes then waits until an admin approves it on ConsentAlly, from a link the assistant hands over, with two-step verification, and the change is made on the assistant's next call. The approval link expires with the confirmation token.

Undo

After a confirmed change the account owner gets an email about it, with a link that undoes it for 7 days. The link opens a ConsentAlly page that says what changed and what undoing restores, and undoes it with one button, once. If the banner or the script changed again since, the page says so and changes nothing. A first publish has no undo link, since undoing it would take the banner down.

An assistant can also make an earlier version live again with rollback, which asks first like publishing; the person can do the same in Publish and history. What ConsentAlly itself does not do is on Limitations.