Keep the next step clear

Better coordination for Melbourne dealerships and service businesses

You may already be trying AI across sales, service or administration and still be unsure what good looks like for the business. We help assess those experiments and choose a useful next move with your team. The service example below is one possibility, not a prerequisite for a conversation.

A customer should be able to understand what is happening with their vehicle, what needs their decision and when they will hear from you again. Your team should be able to answer those questions without reconstructing the job across several screens, notes and conversations.

Our starting point is to begin with that coordination work. AI may help prepare a status summary, organise an enquiry or draft a clear update. The useful result is a more dependable handover between customers, advisers, workshop staff and administration. Technical assessment and decisions about vehicle safety remain with the appropriately qualified people and established procedures.

Trial one operational journey through Understand, Try, Repeat, Operate, Improve. Establish whether the change clarifies the next step and reduces administration.

Follow the job between the departments

The service journey contains several different kinds of information. A customer describes a concern. A technician records findings. Parts staff confirm availability. An adviser explains the proposed work and asks for a decision. Accounts prepare the final record. Each person needs enough context to act without inheriting an unsupported assumption.

Friction can appear at any boundary. A promised callback has no owner. An arrival date is copied from an old message. An estimate is treated as a confirmed total. A customer's question reaches the workshop without the original context. A loan-vehicle arrangement lives in a conversation rather than the job record.

Sales operations have similar boundaries: an enquiry, an appointment, a vehicle's actual availability, a trade-in discussion and a delivery checklist. Different tasks need different controls. A drafting tool should not invent stock availability, finance suitability or a commercial promise.

Trace one journey before selecting a tool. An explicit status field and callback owner may solve the problem; an assistant may help assemble information.

A fictional workflow: the afternoon service update

This fictional example reports no client results.

Imagine a service department preparing afternoon updates for customers whose vehicles remain on site. The trial begins with synthetic job records. Each record separates technician findings, customer-authorised work, pending decisions, confirmed parts information and the next communication time.

An assistant prepares a short adviser packet for each job. Every factual line points to a dated source record. Missing information becomes a question. If parts availability has not been confirmed, the packet says confirmation is needed. If an additional repair is proposed, it remains pending rather than appearing in the completed-work summary.

The adviser checks the packet against the job record and technician's notes. They resolve discrepancies with the responsible person and prepare the customer update. The update explains what is known, what remains uncertain and the next step. It avoids technical conclusions that the technician has not supplied.

When a decision is needed, the adviser explains the proposed work and estimate through the business's authorised process. The decision and its scope are recorded in the system of record. A draft message never becomes evidence of customer authorisation by itself. The workshop acts on the authorised job record.

Consumer Affairs Victoria's guidance recommends written estimates describing work, parts, costs and timing, and advises customers to require contact before unauthorised repairs begin. That makes a clear distinction between information, estimates and permission especially useful in this workflow. Car services and repairs.

The adviser sends the checked update and records the next callback owner. The packet can also be prepared manually. The trial measures preparation, checking and correction.

What would make this worth continuing?

The following is a decision aid for the fictional trial, not an observed result. Agree the limits with the people responsible before testing.

• Accepted result: A checked adviser packet separating confirmed status, proposed work and pending customer decisions.
• Who judges it: The service adviser, with the workshop and callback owner.
• Hold and return to the manual route when: An unconfirmed arrival, repair approval or price is presented as settled, or an unsupported promise reaches a customer.
• Evidence for another step: Compare preparation, checking and clarification for comparable service jobs; include callback commitments and duplicated contacts.

In a fictional packet, “Parts received: confirmed in the dated stock record” belongs under confirmed status. “Completion tomorrow: awaiting technician confirmation” belongs under pending questions. Neither the draft nor its wording authorises additional work.

Use the one-workflow worksheet to record the agreed question, evidence and next decision.

Understand: find the recurring missed handover

We map the selected journey with the people who perform it. Where does information enter? Which record is authoritative? Who may confirm availability, approve an estimate, speak to the customer and move the job forward? We examine recent examples and identify a recurring delay or ambiguity.

The aim is a shared picture of the work. Continue when there is a workflow map, named owners, clear status definitions and a baseline sample. We also identify information that should stay outside the trial. Customer details and workshop records need an approved handling arrangement before live use.

Try: prepare a checked packet for one job type

We start with a limited task, such as preparing an adviser update for an existing service booking. Synthetic examples include missing dates, conflicting notes and additional work awaiting a customer decision. The assistant must preserve those distinctions and surface questions.

The aim is a visible result advisers can judge. Continue when there is a completed packet, its source links, reviewer corrections and preparation time. A fluently written message with an unsupported promise is a failed example. We test the difficult cases deliberately.

Repeat: make the handover usable across shifts

We repeat the method with comparable records and different advisers. A shift change is a useful test: can another authorised person understand the job, identify the pending decision and continue the conversation? The method should not depend on the original author's memory.

The aim is a repeatable handover. Continue when there is a simple template, accepted and rejected examples, and a record of exceptions. Separate new delays from ones already present in the process. If the assistant repeatedly misreads a status, fix the input or narrow the task before adding volume.

Operate: connect carefully to everyday systems

After the trial shows value, we help define how the workflow fits the dealer or workshop management system. Begin with the least access needed. Make review and sending responsibilities explicit. Record who changes job status and who owns the customer's next update.

The aim is reliable coordination during normal operations. Continue when there is a tested procedure, approved access, cost visibility and a demonstrated manual fallback. Integration availability depends on the actual system and permissions; a demonstration does not establish a production connection.

The OAIC's guidance recommends due diligence when selecting AI products and highlights human oversight, information access and privacy risks. We use that as a starting point for checking the proposed information flow with the business's responsible advisers. It is not a certification of any particular tool or dealership. OAIC commercial AI guidance.

Improve: fix what the evidence reveals

Review the exceptions alongside the routine work. Perhaps the packets save preparation effort but advisers still spend too long chasing parts information. Perhaps a shared status definition removes more friction than the assistant. Perhaps customers need fewer, clearer updates rather than more messages.

The aim is a focused next investment. Continue when there is a decision to continue, revise or stop, supported by the trial record. Extend the method to sales appointments or delivery preparation only after assessing their distinct sources, commitments and responsibilities.

A proposed 30-day trial

This proposed evaluation is not a commercial commitment or promised result.

During days 1–5, choose one department and one handover. Establish status definitions, responsibilities and the baseline. During days 6–12, test synthetic packets with advisers and include conflicting or incomplete records. During days 13–21, repeat the method; use limited live records only after the appropriate access and information-handling approvals. During days 22–30, review quality, staff effort and operating readiness, then document the decision.

Fit the trial to the team's checking capacity. If live use is not approved, use synthetic records and state what remains unproven.

Measure clarity, effort and promises kept

Before starting, define a comparable sample by job type and stage. Measure preparation and checking time, missing required fields, duplicated contacts, callback commitments completed within the agreed window and handovers returned for clarification. Record the count and denominator for every rate.

Track material errors separately: an incorrect status, unsupported completion date, wrong amount or confusion between proposed and authorised work. Those are more consequential than a minor wording correction. Pause the trial if such errors reach customers or the workflow changes authorisation without the required decision.

The baseline should include manual work and customer follow-up, not just message drafting. Compare routine and exception jobs separately. Parts delays, staffing changes and unusual workshop demand can affect the comparison. A short observation period may show a useful direction without establishing a causal business result.

Calculate cost per accepted packet from software use, preparation, checking and rework. Ask advisers whether the process makes the next action easier to identify. Ask customers a simple clarity question through an approved process if appropriate. Do not treat more messages or higher upsell value as proof of better service.

How we help

Alongside People combines advice with implementation. We can map the journey, improve the handover, assess a tool against the actual task and build a checked workflow around your existing systems. We help establish the baseline, review exceptions and prepare a method your people can operate.

Paul Volpato is our founder and CEO. Start with one recurring handover that makes your team chase information or leaves a customer unsure of the next step.

Sources and scope

Understand → Try → Repeat → Operate → Improve is Alongside People's original discussion framework. It is not a certification, validated maturity assessment or official standard.

The five stages and trial are our proposed approach. The example is fictional. This paper concerns dealership and service administration; it makes no diagnostic, roadworthiness or safety-performance claim. Specific legal and information-handling obligations need assessment for the business and its proposed use.

• Car services and repairs, Consumer Affairs Victoria; current guidance checked 3 October 2026.
• Guidance on privacy and the use of commercially available AI products, OAIC; published 21 October 2024, updated 17 January 2025, checked 3 October 2026. Refresh before live use.

If you are already trying AI and are unsure what deserves further investment, tell us what you have tried and what you want to improve. We can help find the starting point together.