What Claude can and can't do in your ReminderIt account

Giving an AI assistant the ability to place phone calls sounds like a lot of power to hand over. In practice, Claude's reach into ReminderIt is tightly bounded. It works only inside your own account, cannot spend beyond your plan, and cannot reach other people without their agreement. Here is the honest map of what it can and cannot do.

It acts only in your account

When you connect Claude, it is scoped to your ReminderIt account and nothing else. It can see and change your reminders, check your usage, preview voices and manage the care-recipients you have added. It has no view into anyone else's account and no back door to the wider system.

Every action it takes is one you could take yourself in the dashboard. It is an assistant working on your behalf, not an administrator with special reach.

It cannot exceed your quota

Claude cannot conjure extra calls. Your plan sets a pooled monthly allowance, 5 calls on Free and 70 on Pro, and the assistant draws from exactly that pool. Ask it to schedule more calls than you have left and it simply cannot place them beyond your limit.

You can check where you stand any time by asking "How many calls have I used this month?", which runs get_usage. There is no hidden budget for the assistant to tap.

It cannot override opt-outs

If someone has opted out, that stands. When a person replies STOP to ReminderIt's WhatsApp messaging, they are opted out, and no instruction to Claude will force a message through against that choice. The assistant respects opt-outs as a hard boundary, not a preference it can weigh up.

This matters most for care-recipients: their choices about being contacted are theirs, and the assistant cannot talk over them.

Adding another person needs their consent

Claude cannot quietly start calling a relative or friend. Adding someone as a care-recipient runs add_recipient, which triggers a consent call to that person first. Until they confirm on that call, no reminder is scheduled to them.

So even the one action that reaches beyond you is gated by the other person's explicit agreement. You can review who is set up with list_recipients and remove anyone with remove_recipient.

You stay in control, and can revoke

Access is not permanent. Each connected client, whether Claude, ChatGPT or Cursor, is granted its own separate key, and you can revoke any of them at reminderit.com/api-keys. Revoking one leaves the others untouched, so cutting off Claude does not disturb your ChatGPT connection.

The moment you revoke, that assistant can no longer act in your account. Nothing about the connection is one-way or irreversible.