How Zen AIworks

Zen is the AI assistant built into Zenchronity. People write to it in plain language, in the web app or in Slack, and it answers from their organisation's leave data. The chat always makes clear that you are talking to an AI.

• request leave for you, for example “I need next Friday off”;
• cancel one of your own requests that hasn't started yet, after you confirm;
• tell you your balances and which leave types you can use;
• tell you when your next day off is and which public holidays are coming up in your country;
• show your pending and past requests;
• tell you who in your team is out, and give administrators company-wide leave statistics;
• for managers and administrators: list the requests waiting for their decision, warn when teammates are off on the same days, show the team's remaining days, and approve or reject a request when they ask for it and confirm.

Zen works with the same permissions as the person asking, so it never shows anyone more than the app itself would.

• It never decides on its own. It approves or rejects a request only when the manager or administrator allowed to decide asks for it and confirms, and the decision stays theirs. Read-only administrators can't approve through Zen.
• It never acts in bulk: one cancellation, approval or rejection per message, each confirmed separately. Bulk actions stay in the app.
• It never changes settings or balances, and never touches requests you aren't allowed to act on.
• It never evaluates employees.

Zen's answers are generated automatically and can be wrong, so check anything important in the app. A request Zen files for you goes through the usual approval, and you can cancel it the usual way or by asking Zen.

Zen's replies are generated by a model from OpenAI, which acts as our sub-processor under our Data Processing Agreement.

• Before a message leaves our servers, the names and email addresses of your colleagues are replaced with aliases, and the real names are put back only on our side.
• Leave comments and rest scores are never sent.
• OpenAI does not use this data to train its models.
• Your conversations are stored with your account in the European Union.

Anything you type is sent as you typed it, so please don't share health details or other sensitive information with Zen.

Zen is on by default. Your organisation's administrators can turn it off at any time in Settings. While it is off, Zen is hidden in the app, does not answer in Slack, and nothing is sent to the AI model provider. Existing conversations are kept and come back if Zen is turned on again.

Rest scores and burnout signals are calculated from annual leave by fixed rules: how evenly leave is used through the year, whether it is taken in real breaks, and how long it has been since the last day off. The app shows the reason for every score. No AI model is involved, and the scores are never sent to one.

They describe how time off is taken. They are not a medical or psychological assessment and do not decide anything on their own.

Write to privacy@zenchronity.com with any question about Zen and your data.