An enquiry arrives. A draft reply is waiting in the mailbox, written from your own pricing, documents and customer records.
Somebody reads it, edits what they want, and sends. The agent has no ability to send anything itself.
In build for an events company handling 200 or more enquiries a day.
An email creator or a generic email ai writer produces fluent text from a prompt. Fluent is easy. Correct is the problem, because a customer question usually has one right answer sitting in a document somebody has to find.
Ai email personalization tools solve a different job again: many recipients, one message. This is one customer, one question, one answer.
That is enforced by permission, not by policy. The application holds read and write access to mail for drafting, and no send permission, so the boundary holds even if something goes wrong.
Prices come from your current price list, traced to the row they came from.
Your people stay in Outlook or Gmail.
Microsoft 365 and Outlook, Gmail, SharePoint, Google Drive, and your customer or booking system.
No. It holds no send permission.
Your own documents and records, cited back to the source.
It says so and flags the enquiry for a person.
No. The drafts appear in the mailbox everyone already uses.