Eight free working guides. Adapt the boundaries and system actions to your organization before use.
Free playbook / dental-receptionistDental AI Receptionist Call Flow
Book a routine visit within the practice’s approved scheduling rules.
Collect only what is needed
- New or returning patient
- Requested appointment type
- Preferred dates and callback number
The call flow
- Disclose the automated assistant and confirm the purpose.
- Check only the approved appointment types and calendar.
- Read back the date, time, location and name before writing.
- Confirm the booking only after the scheduling system succeeds.
Where AI should stop: Clinical questions, pain or urgency go to the practice’s approved staff pathway. Do not diagnose or promise treatment.
Test before rollout: Collision, unavailable slot, spelled name and unanswered transfer.
Open the pilot test sheet →Free playbook / restaurant-reservationRestaurant Reservation Call Flow
Confirm a reservation that the actual inventory can support.
Collect only what is needed
- Party size
- Date and time
- Accessibility or seating request
The call flow
- Check opening hours and capacity for the requested party.
- Offer available alternatives without promising exceptions.
- Read back party size, time and the cancellation policy supplied by the restaurant.
- Write once and confirm only the returned reservation.
Where AI should stop: Allergy assurances, large-event exceptions and policy disputes require staff confirmation.
Test before rollout: Last available table, changed party size and failed reservation write.
Open the pilot test sheet →Free playbook / property-maintenanceProperty Management Maintenance Intake
Create an accurate maintenance ticket and route it to the agreed responder.
Collect only what is needed
- Property and unit
- Issue in caller’s words
- Permission and callback preference
The call flow
- Confirm the location before dispatching anything.
- Classify the issue using approved categories.
- Repeat the ticket details and record access preferences.
- Provide a reference number from the ticket system.
Where AI should stop: Use the property’s reviewed urgent incident script; never improvise repair or safety advice.
Test before rollout: Ambiguous unit number, duplicate report and unavailable responder.
Open the pilot test sheet →Free playbook / real-estate-leadsReal Estate Lead Qualification
Capture a prospect’s requested service and book a human follow-up.
Collect only what is needed
- Buying, selling or renting
- Area and stated budget
- Preferred contact time
The call flow
- Ask permission to collect the minimum useful details.
- Record the prospect’s own requirements without assumptions.
- Offer actual available follow-up slots.
- Send the confirmed brief to the assigned agent.
Where AI should stop: Financing advice, protected personal attributes, contract interpretation and guarantees require an appropriate human.
Test before rollout: Budget correction, existing lead and withdrawal of contact permission.
Open the pilot test sheet →Free playbook / after-hoursAfter-hours Escalation
Give callers a predictable path when the main team is unavailable.
Collect only what is needed
- Service and location
- Callback details
- Reason for escalation
The call flow
- Use the current hours and on-call roster.
- Explain what can be completed now.
- Attempt the approved transfer with agreed context.
- If unanswered, offer the approved callback or ticket path and record the outcome.
Where AI should stop: Never claim a human is attending until the transfer is accepted. Follow reviewed urgent-call policy.
Test before rollout: Stale roster, unanswered transfer and repeated caller.
Open the pilot test sheet →Free playbook / support-handoffCustomer Support Human Handoff
Transfer ownership without making the caller repeat every detail.
Collect only what is needed
- Verified account context where allowed
- Problem summary
- Steps already attempted
The call flow
- Confirm that the caller wants a person or has reached a boundary.
- Create a short factual summary without unsupported diagnosis.
- Transfer to the correct queue with approved context.
- Track whether the destination answered; retain ownership until resolved.
Where AI should stop: Do not expose sensitive information to an unverified recipient or treat a transfer attempt as resolution.
Test before rollout: Wrong queue, inaccessible account and dropped transfer.
Open the pilot test sheet →Free playbook / voice-agent-qaVoice Agent QA Checklist
Review a representative call set before widening a rollout.
Collect only what is needed
- Approved call types
- Evaluation rubric
- Owners and rollback triggers
The call flow
- Include failures and escalations in the sample.
- Evaluate transcript accuracy separately from action correctness.
- Check every success claim against the system of record.
- Log defects, fix configuration and rerun the same cases.
Where AI should stop: A pass on scripted calls does not establish safety or reliability across all callers.
Test before rollout: Use the 15-case pilot sheet and retain configuration/version evidence.
Open the pilot test sheet →Free playbook / pilot-test-sheetPilot Test Sheet
Keep a reproducible record of a buyer-operated pilot.
Collect only what is needed
- Platform, configuration and date
- Scenario and expected result
- Observed result and evidence reference
The call flow
- Assign a reviewer and an acceptance criterion before running a call.
- Run normal and failure cases using non-sensitive test data.
- Mark Pass, Fail or Not run with notes.
- Record sign-off separately from implementation and keep a rollback owner.
Where AI should stop: No universal platform rating is produced. Your team decides the deployment boundary.
Test before rollout: Export the pilot checklist with its scenario, result, owner and evidence fields.
Open the pilot test sheet →