How AI reads your notes

Donor notes hold things people told you in confidence. This page describes exactly what happens to them, in plain terms, and links to the terms that govern it.

Identifiers are removed first

Known identifiers, the donor names on file, emails, phone numbers and addresses, are removed before your notes are read by AI. What travels is the note with placeholders where the people were.

The lookup that turns a placeholder back into a name stays inside your workspace. It is held for the length of one request and then discarded. It is never stored and never logged.

So the thing that would identify a donor and the thing that gets read are never in the same place.

Processing happens in Canada

Reading a note happens in Canada, the same country your records are stored in.

The one exception is the research feature described below, which is off until you turn it on, and which never sends notes.

Your notes are not training data

Your notes are not used to train models. The services Relae uses operate under commercial terms that prohibit training on customer content by default.

The features that are off until you turn them on

Two capabilities touch individual donor data, and both are off by default and controlled by an admin under data and privacy settings.

Research on individuals

What research sends, and what it never sends

Sent: the prospect’s name, entity type, city, education, and known employer, board or company affiliations, plus web search queries. This is how the right person is identified rather than someone with the same name. These requests go to services in the United States.

Never sent: giving history, amounts, ratings, prospect types, tags, and notes. These are removed before the request is made.

This happens only where your organization has opted in, and exactly what is sent is shown again at the point you turn it on. See the AI and Automated Processing Terms.

The assistant and individual donors

The assistant answers from aggregate figures by default.

Turning on live donor data lets it look up an individual, their giving history and pipeline standing, so it can answer questions like “when did Margaret last give?”. Names, emails, phone numbers and addresses are replaced with placeholders before anything reaches the model, real names are reattached only inside your workspace, and every request is processed in Canada.

It starts off. Turning it on is your organization’s instruction as the controller of that data, and you can turn it off again at any time.

The one thing that changes a record on its own

Almost nothing in Relae writes to a record without a person confirming it.

There is one exception, and it is off by default: where an admin turns on duplicate auto-merge, records the sweep is certain are the same person can be merged automatically. In the words of the terms, it is “off by default, per-organization, reversible, and limited to merging records of the same person”. Every automatic merge is listed and can be undone from the record.

With it off, every proposed merge waits for you.

What this means in practice

• Write your notes normally. The mechanism handles the identifiers.
• If your board asks whether donor names go to AI, the answer is that they do not.
• If they ask whether anything leaves Canada, the answer is that research does, it is off unless you turned it on, and it never carries the private record.

The full terms are in the AI and Automated Processing Terms, the Privacy Policy and the Data Processing Agreement.

Next

• Where your data lives
• Writing notes in plain sentences