Canned Message Learning

Canned Message Learning suggests better wording for your canned responses. A person reviews every suggestion. Nothing changes until you approve it.

What it learns from

Velaro looks at your canned responses and finds the ones agents rate low (average under 3.5 stars from at least 3 ratings) and the ones nobody uses. For each one it proposes a clearer version and tells you what it learned and why it changed the text.

Where to find it

Open Canned Message Learning in the Messaging menu, next to Canned Responses. It requires the Intelligence Suite on your plan.

Review a suggestion

1. Select Find improvements. Velaro checks up to 5 messages per run, one run per day.

2. Read the current text beside the proposed text and the reason for the change.

3. Edit the proposed text if you want, then select Approve, or select Reject to leave the message as is.

Approving updates the live canned response right away. Rejected suggestions are kept in the Rejected tab.

Learning from edits

When an agent inserts a canned response and changes it before sending, Velaro records the final wording (emails, phone numbers and long numbers masked) and uses it as extra evidence. A message edited three or more times with little left of the original becomes a suggestion candidate. Edit records are kept for 90 days, capped per site, and never used outside your own site.

Weekly suggestions

Velaro also generates suggestions once a week for every site with the Intelligence Suite, using the same daily limit and 5 message cap as the Generate button. Nothing changes until you approve.

Undo an approval

Open the Approved tab and select Roll back. The original text is restored exactly.

Privacy

Emails, phone numbers and long account numbers are masked before text is sent to the AI model. Suggestions only ever use your own site's canned responses.

Also available through

The velaro CLI (velaro canned-learning list|generate|approve|reject|rollback) and the Velaro MCP tools canned_learning_suggestions, canned_learning_generate, canned_learning_approve, canned_learning_reject and canned_learning_rollback. To see which messages agents edit most: velaro canned-learning edits or canned_learning_edit_summary.