Back
Troubleshooting and good practices for AI Calls
Discover essential troubleshooting techniques and best practices for optimizing AI call performance and user experience.
Troubleshooting
| The call doesn’t trigger |
The AI call is added to the stage, the candidate is in the right stage, and a phone number exists. |
| The candidate doesn’t answer |
Allowed calling hours and Retries when no answer in the AI Agent tab. |
| The call ends too early |
Your required (knockout) facts and the early-stop switch in Call settings. |
| The evaluation doesn’t match your criteria |
Competencies, Key facts, and Signals in the Role context. |
| No one moves to the next stage |
Lower the minimum match label, or check whether requirements are too demanding. |
| Too many candidates move forward |
Raise the minimum match label, or make required criteria more specific. |
Launch checklist
- Role context opened from Job settings.
- Competencies added with Title, Description, and relevance.
- Verifiable Key facts added with a Condition.
- Positive and negative Signals reviewed.
- New stage ready as the entry stage.
- Interview stage created as the next stage.
- AI call added to the first stage.
- Things to validate has at least one active category.
- Call settings reviewed and Run test listened to.
- AI movement enabled with an initial threshold.
Privacy and good use
- The AI assists, it never decides alone. Human review always remains.
- Be transparent with candidates: explain there will be a screening call.
- Don’t use prohibited criteria: avoid sensitive or discriminatory information.
- Candidate data is used only to evaluate that specific role.