AI Control Centre
The assistant, agents, models, memory, approvals, findings, policies and scheduling.
- Use the staff AI AssistantAsk operational questions about your own provider, keep the conversations, and cancel anything you no longer need.
- Create, version and retire AI agentsAgents with a role, instructions and a personality; every save is a new version, and any version can be restored.
- Configure models and complexity routingWhich model each agent uses, what it falls back to, and how simple queries are routed to cheaper models.
- Search and manage AI memorySee what the assistant remembers, pin what matters, redact what should not have been kept, delete what must go.
- Review AI action approvalsEvery action an agent wants to take that your policies gate waits here for a person to accept or reject, with a comment.
- Review expert findingsFindings the AI raises about a learner or a session — review them, revert them, and open the learner's record from them.
- Set AI approval policiesWhich tools need a person to approve them, how risky they are, and how many approvers they need.
- Test an agent safely and compare promptsA sandbox that records nothing, and an A/B prompt test with a side-by-side comparison.
- Generate proactive learner notificationsAttendance warnings and similar nudges, generated in bulk and never duplicated in a day.
- AI analytics, insights, DevOps checks and the schedulerHow much the AI is used and what it costs, what learners are asking, whether the services are healthy, and the scheduled tasks that run.