The AI Control Center is a proactive layer on top of the AI assistant: a permanent "AI staff member" that automatically analyses your organization every day and prepares concrete recommendations. For each recommendation you see what was detected, who is affected and what impact an action has — and you execute the proposed action with one click after reviewing it.
Where to find it
Sidebar → the AI block → AI Control Center (or via the global search bar). The feature is only visible to administrators (not teachers) and only if the AI assistant is enabled for your organization.
How it works
An analysis runs automatically every morning. You also get new or important recommendations as a bell notification. Don't want to wait? Click Analyze now at the top for a fresh analysis (available once every few minutes).
Reading a recommendation
Each recommendation card shows three things:
- What was detected — with exact numbers from your own data (the AI does not invent numbers).
- The impact — for example "23 members → push + email".
- The proposed action(s) — with Review & execute and Dismiss buttons.
Review, edit and execute
Click Review & execute to open the action. If it's a message (reminder or announcement), you see the AI-written text that you can edit per language (NL/TR/EN/AR). You also see who is affected. Adjust the text where needed and click Execute. You decide what gets sent.
What it can detect
- Engagement: members who haven't logged in for a long time, and memberships expiring soon.
- Communication: no recent announcement, and events with low enrollment.
- Finance: overdue payments, unusual cashbook transactions and looming budget overrun.
- Education: chronically absent students, overdue homework, urgent dossiers, pending registrations and open questions.
- Operations: members who are unreachable (no email or push), and a weekly briefing for the board.
Which signals are active depends on which modules are enabled for your organization.
Messages in each member's language
Member reminders are sent across multiple channels — in-app, push and email — and each member receives the message in their own language. The AI drafts the text in four languages; you can always edit it before it is sent.
Autopilot (optional)
By default every action requires your approval. Per action type you can optionally turn on autopilot: the AI then executes that action automatically during the nightly analysis. This is only possible for safe, reversible actions — member reminders and the weekly briefing. All other actions (such as publishing an announcement or financial actions) always stay manual. A maximum number of recipients per automatic run also applies.
Autopilot log
The Autopilot log tab shows exactly what the AI did: which action, by autopilot or manually, with how many recipients per channel and the exact text sent. This gives you full visibility of every autonomous action. These actions also appear in your organization's Activity log.
Settings
On the Settings tab you turn each recommendation type on or off, adjust thresholds (e.g. "inactive after 60 days") and enable autopilot for the safe types. Changing settings requires the manage settings permission.
Permissions and safety
When executing an action, the system checks whether you have the matching permission (e.g. manage members for reminders or billing/finance for payment reminders). An action is never executed twice, and the AI runs on our own EU infrastructure.
Tips
- Start with approval only: watch which recommendations come in for a few days before enabling autopilot.
- Enable autopilot first for the weekly briefing — it's informational and low-risk.
- Resolve a data hygiene recommendation quickly: members without email or push otherwise receive no reminders.
Frequently asked questions
Does the AI take actions without my permission? By default no. Only if you enable autopilot, and then only for safe, reversible actions. Everything is in the log.
Can I dismiss a recommendation? Yes, with Dismiss. The same one won't return for a while.
Are the numbers reliable? Yes — they come straight from your own data; the AI only puts them into words.