AI agents
Performance
A weekly report card for your assistant. It grades how well the agent is doing and proposes specific changes — which only take effect if you approve them.
The health score
A single number from 0 to 100 summarising recent performance, with a plain label:
- 80–100 · Excellent
- The agent is answering well and students are responding positively.
- 60–79 · Healthy
- Solid overall, with a few areas worth a look.
- 40–59 · Fair
- Noticeable weak spots — worth spending time on the Knowledge Base.
- Below 40 · Needs attention
- Something is off. Review the suggestions and recent conversations closely.
You can view the score over the last 7, 30 or 90 days.
Improvement suggestions
The platform reviews how conversations have gone and proposes small, specific adjustments to how your agent behaves — for example, being firmer about collecting a phone number early, or explaining a fee more clearly. Each suggestion is yours to act on:
Read the suggestion
It explains what it noticed and what it would change.Approve or dismiss
Approve & apply folds the change into your agent. Dismiss discards it.
Nothing changes without your approval
The learned playbook
Approved improvements collect into a playbook — a short list of learnings that guide the agent. You can edit it directly (one learning per line) or delete it entirely to start over. Think of it as the accumulated wisdom your agent has picked up from working your enquiries.
A good routine
- Check the score once a week.
- Clear the suggestion list — approve the good ones, dismiss the rest.
- If the score drops, cross-check with Knowledge Gaps and recent feedback.