A garage caller changing a collection arrangement needs a different response from a new service enquiry. Evaluate Zanus AI around the existing booking, vehicle and service adviser so that a convenient phone conversation does not become an unsupported promise about repair completion.
Start with the existing service reference
Capture the customer’s name, a suitable vehicle or booking reference and the reason for calling. Repeat important identifiers clearly. If a family member is calling, establish what the office needs to verify before discussing details or changing arrangements.
A household can have several cars and bookings. For an illustrative trial, create two sample vehicles for one contact, then change the drop-off arrangement for only one. Inspect whether the summary points to the intended booking.
Keep arrival arrangements distinct from workshop progress
A drop-off time describes when the customer intends to bring the vehicle. It does not establish when the technician will start or finish. Give the agent only the arrival information the business is prepared to confirm.
| Question | Suitable office action |
|---|---|
| Can I arrive later? | Review the booking and notify the adviser |
| Is the repair finished? | Consult the authoritative job update |
| Can someone else collect? | Follow the garage’s verification process |
| Can you approve extra work? | Route to the authorised service discussion |
Handle collection changes without inventing readiness
If the customer asks to collect at a different time, record the request and the responsible adviser. Use the actual workshop status before telling anyone the vehicle is ready. A previous estimate of completion is not a fresh confirmation.
Ask the supplier which record the agent reads, when it was updated and what happens if that source is unavailable. A useful fallback states that the team needs to confirm the answer and creates a recoverable callback request.
Use outbound calls for a specific confirmed purpose
An outbound reminder can refer to a booked drop-off or ask the customer to contact the adviser. Keep the approved purpose separate from additional sales offers. Review recipient selection, calling preferences and the handling of a requested change before enabling a live list.
Test voicemail and a conversation with a different household member. The result should identify what actually happened. An unanswered reminder should not count as the customer agreeing to a new arrangement.
Check the adviser’s working queue
After the controlled calls, ask whether the adviser can identify the vehicle, current arrangement, requested change and promised next response. Review incorrect matches and repeated clarification calls. Success means the garage and customer agree on the next step, rather than simply recording that an agent answered the phone.
Official documentation and related reading
- Zanus AI official product guide
- Zanus AI official supporting documentation
- Zanus AI CRM for HVAC: Turn Call Notes Into Useful Job Enquiries
- GoHighLevel for Trades: Compare Voice AI and Webchat Costs
- Zanus AI for home services: the broader evaluation
Explore the relevant Sysevo product, Voice Agents and CRM. Compare Sysevo plans and pricing or book a demo using the same requirements.
Reviewed 3 November 2026. Independent Sysevo editorial guide using the linked official documentation. Scenarios are proposed evaluation steps, not reported customer results or a hands-on vendor benchmark. Sysevo sells its own business software; availability and deployment requirements need confirmation.
Evaluating How an Agent Handles the Same Vehicle Twice in One Day
A garage that runs a high volume of daily bookings will encounter situations where a single vehicle generates more than one call within a short window. A customer might phone to push back a morning drop-off, and then a family member calls an hour later asking whether the car is ready to collect. Both conversations touch the same booking, but they carry different responsibilities and different risks of miscommunication.
When evaluating an AI phone agent for this kind of environment, a useful exercise is to script exactly this scenario before any live deployment. Create a fictional booking for a vehicle under a household contact. Run a first call in which the registered keeper moves the drop-off from an early slot to a mid-morning slot. Then, without updating any fictional workshop record, run a second call from a different caller at the same household asking about collection readiness. The question to observe is whether the agent treats these as two independent requests or correctly associates both with the same booking while keeping the answers appropriately separate. If the agent tells the second caller that the vehicle is ready based solely on a previously logged estimated completion time, that is the failure mode the business needs to prevent before going live.
This kind of structured rehearsal is not a claim about how any particular system behaves by default. It is a repeatable acceptance exercise that gives the evaluating business observable evidence: what did the agent say, to whom, and what did it log? Those three questions should have clear, auditable answers before a workshop relies on the tool for real customer conversations.
Distinguishing Who May Authorise a Change
A garage has obligations to the registered keeper of a vehicle that differ from its obligations to whoever happens to telephone. This distinction matters most when someone asks to change a collection arrangement rather than simply asking a factual question. A drop-off time change is a lower-stakes request; it affects scheduling rather than ownership or payment. A collection change, particularly one where a person other than the registered keeper intends to take the vehicle away, carries a different level of responsibility.
Before deploying any AI agent to handle collection calls, the business should write down its own policy in plain language: who is allowed to authorise a third-party collection, what verification the office requires, and what the agent should do when it cannot complete that verification. The agent's behaviour should then be evaluated against that written policy, not the reverse. It is a common evaluation mistake to accept whatever the agent does by default and then describe that behaviour as policy. The business owns the policy; the agent should reflect it.
In practical terms, this means giving the agent a defined outcome for the scenario where the caller cannot provide the required information. A useful outcome is a callback request logged against the booking, with the adviser notified. An unhelpful outcome is a vague statement that the team will be in touch, with nothing written to the booking record. When evaluating the tool, run the third-party collection scenario deliberately and inspect what the adviser sees afterward. If the adviser cannot tell that a collection change was requested, or by whom, the handover is incomplete regardless of how politely the call ended.
Mapping the Journey of a Single Fictional Drop-Off Change
A structured walkthrough of one fictional enquiry from receipt to resolution helps a business identify where information could be lost. The following is an illustrative scenario, not a reported result.
A caller identifies herself as the registered keeper of a vehicle booked in for a brake inspection. She explains that her plans have changed and she would like to arrive two hours later than the original booking time. The agent confirms her name, asks for a reference or the vehicle registration to locate the booking, and reads back the booking details including the original time and the service type. She confirms this is the correct booking. The agent logs the requested new arrival time and flags the booking for adviser review, because a later arrival may affect whether the technician can complete the inspection within the same working day.
At this point, three things should be observable in the system: the original booking time, the requested change, and the name of the person who requested it. The adviser should be able to look at the booking and understand immediately that no confirmation has yet been given to the customer about whether the new time is workable. The agent should not have told the customer that the new time is confirmed unless the business has explicitly authorised the agent to make that commitment autonomously.
The follow-up from the adviser involves checking the day's schedule and calling or messaging the customer to confirm or propose an alternative. If the adviser makes that call and the customer raises a question about the brake inspection itself, that question should be handled by the adviser rather than routed back through the agent. Workshop progress conversations benefit from human continuity because the adviser can refer to inspection notes, parts availability and technician judgement in real time.
When evaluating the tool, run this scenario and then ask: how many steps required a human decision, and were those steps clearly flagged? A well-configured agent creates clean handover points rather than attempting to resolve everything within the call.
Reviewing What the Adviser's Queue Actually Shows
The measure of a useful phone agent in a garage context is not call volume or answer rate. It is whether the adviser's working information is accurate and actionable after each call. A practical evaluation exercise involves deliberately generating a mix of call types across a controlled test period: a straightforward drop-off time change, a collection enquiry where the vehicle is not ready, a call from a household member who cannot provide verification, and an outbound reminder that reaches voicemail.
After each test call, the evaluating business should sit with the adviser and ask four questions. First, can the adviser identify which vehicle and booking the call concerned? Second, does the adviser know what the caller asked for? Third, does the adviser know what, if anything, the agent committed to? Fourth, does the adviser know what action is expected from them and by when?
If any of those questions produces uncertainty or requires the adviser to listen to a call recording rather than reading a structured note, the configuration needs adjustment. Recording availability is useful for dispute resolution but should not be the primary tool for daily workflow management. The structured note is the primary tool, and its quality is what the evaluation should judge.
Setting Useful Limits on Outbound Call Content
Outbound reminder calls for drop-off arrangements serve a clear and limited purpose: they confirm to the customer what the garage has recorded, and they give the customer an easy route to notify the office if that record is wrong. That is a useful function. Difficulties arise when the scope of an outbound call is extended without equally extending the agent's ability to handle the responses that follow.
If an outbound call tells a customer their vehicle will be ready for collection at a certain time, the agent must be prepared to handle a response in which the customer says that time does not work. That response requires the same capabilities as an inbound collection change call. If those capabilities are not fully configured, the outbound call has created an expectation that the inbound process cannot reliably meet.
A conservative approach during evaluation is to test outbound calls with a narrowly defined purpose and confirm that every possible customer response routes cleanly to a logged outcome or a human decision point. Only after that test is clean should the business consider expanding outbound call content. The order of evaluation matters because expanding content before confirming response handling produces a backlog of unresolved customer expectations that falls on the adviser to clear.
Voicemail handling deserves particular attention. When an outbound call reaches voicemail, the message left should not state anything the customer could interpret as confirmed. A message that says a vehicle will be ready at a certain time becomes a commitment even if the garage intended it only as an estimate. The evaluating business should decide in advance what the voicemail message may say, write that language out, and test it before enabling outbound calls on a live list.
Exception Cases That Reveal Configuration Gaps
Several edge cases consistently reveal whether an agent configuration is robust enough for garage use. The first is a caller who provides a name that matches the booking but cannot provide any secondary identifier. The second is a caller who insists the vehicle is ready because someone at the garage told them it would be done by a particular time. The third is a caller who wants to authorise additional repair work over the phone without speaking to an adviser.
Each of these cases has a correct administrative outcome that does not require the agent to make a substantive decision. For the first case, the outcome is a logged callback request with the booking flagged for adviser verification. For the second, the outcome is a note that the caller expects the vehicle to be ready, with the adviser notified to check the actual workshop status and contact the customer with an accurate update. For the third, the outcome is a transfer or callback to the adviser responsible for that job, because authorising additional work involves cost agreement and technical judgement that fall outside administrative scheduling.
When these scenarios are run during evaluation and the agent produces the correct administrative outcome each time, that is meaningful positive evidence. When the agent attempts to resolve any of them without creating a human decision point, the configuration needs revision. The evaluation is not complete until all three scenarios have been run and their outcomes inspected.
Connecting the Booking Record to the Right Adviser
In a garage with multiple service advisers, a vehicle booking is typically associated with one named person who is responsible for communication with the customer. When a call results in a logged change or a pending query, that record should reach the responsible adviser rather than a general inbox that anyone might read or no one might check promptly.
Before deployment, the evaluating business should confirm how the tool routes logged items and what happens when the responsible adviser is unavailable. A reasonable fallback is a defined deputy arrangement: if the responsible adviser is not accessible within a given period, a named colleague receives the item. This fallback should be agreed in advance and visible in the adviser's queue so that both the primary and the deputy can see ownership clearly.
Testing this routing before live deployment involves deliberately logging a change against a booking assigned to an adviser who is marked as unavailable, then observing what happens to that item. If it sits unrouted, the business has identified a gap that would otherwise appear only when a customer calls back to ask why nothing has changed. Identifying it in a controlled test is substantially less disruptive than discovering it through a dissatisfied customer.
A garage that treats the adviser's queue as the true output of its phone agent configuration will generally make better deployment decisions than one that treats call completion as the measure of success. Calls are inputs; the adviser's ability to act on accurate, timely, correctly attributed booking information is the output that actually serves the customer and the business.



