Frnd Go to app

What's already working

Frnd is in early alpha. Here is what you can actually connect and use today – and a few honest notes on where we are headed next.

Abilities

Currently implemented

Calendar – read-only

List your calendars, look up events, and check your tasks – read-only, and only in the calendars you allow. You can also follow a public calendar feed by its ICS link.

Mail – read, search, draft

List folders, read messages, search your inbox, and mark messages read or unread – all within the permissions you set. Your AI Assistant can also write drafts – new mail or replies – into your drafts folder, for you to review and send from your own mail client. Sending mail directly is not available yet.

World clock

Look up the current time in any timezone, convert a time between zones, and work out the difference between two of them.

Future abilities

We are not pinning this to a roadmap – plans shift as we build – but here is the direction:

Next in line:

  • Action approval system (see below).
  • Sending email.
  • Writing to your calendar.

Further out:

  • Drive / file access.
  • Contacts.
  • Other services – especially open-protocol ones, or teams who want to work with us.
  • A mobile app. Everybody needs an app 🙂️ But we envision that our app will make your usage of Frnd more user friendly – so try it out when it comes.

Every action is logged

Every request that reaches Frnd is recorded – what happened, when, and through which connection. You can check the log at any time.

Content from others is fenced off

An email or a calendar invite can be written by anyone – including someone trying to trick your AI Assistant with hidden instructions. Frnd strips the hidden-character tricks and clearly marks such content as untrusted. This helps your AI Assistant to distinguish between what it reads and what you ask it to do. This is always on, for every ability.

Longer term we want to take this further: A dedicated AI that classifies incoming content – flagging spam, phishing, and prompt-injection attempts before they reach your AI Assistant. This will, of course, be an opt-in feature.

Approval for sensitive actions Coming soon

Right now, anything you have allowed runs straight through. We are building an approval step so sensitive actions wait for your sign-off before they happen, instead of only showing up afterwards in the log.

Got suggestions?

Sign up in the app and write us there – tell us which ability would be most useful to you, and it will shape what we build next. Go to app →

New to Frnd? Getting started guide →