What can an AI receptionist take off the front desk?
Phone work that follows clear rules suits it best: calls after hours, calls that arrive while the receptionist is serving someone at the counter, questions about opening hours, parking and directions, simple bookings and message taking. Each of these is a call your receptionist either misses today or answers by interrupting the person standing in front of them. An AI receptionist does not greet visitors, take payments at the desk, sort mail or calm an upset person in the waiting room. Those tasks stay with your staff.
Which calls should stay with a person?
Keep the calls that need judgement, discretion or an existing relationship. Complaints, distressed callers, billing disputes, long-standing clients who expect a familiar voice and anything involving a sensitive personal matter belong with staff during business hours. Your receptionist knows which callers fall into these groups better than any written rule, so ask them to draft the list. The AI receptionist can still answer these calls after hours, take the details and tell the caller that a person will follow up.
How does the receptionist's job change?
The role moves from answering every ring to managing the calls the AI handled. That includes reviewing call summaries each morning, making callbacks, fixing out-of-date answers, keeping hours, services and prices current in the receptionist's knowledge base, and flagging calls that went badly. These tasks need skill and time, so put them in the roster. If the person who used to answer the phone has no time set aside for this work, the AI's mistakes will go unnoticed and callers will be the ones who find them. Give the receptionist what the new work needs:
- A fixed time each morning for the call review
- Access to edit the receptionist's business information
- A short written process for correcting a wrong answer
- A direct contact for the provider's support team
What do awards require before a major change?
The Clerks Private Sector Award 2020 covers clerical work, which it defines to include receiving and answering calls and attending a reception desk. Under its consultation clause, once an employer makes a definite decision to introduce major changes in technology that are likely to have significant effects on employees, the employer must give notice, discuss the changes, their likely effects and ways to reduce adverse effects, and start those discussions as soon as practicable. Significant effects include altered hours of work, the need for retraining and job restructuring. Check the consultation clause in the award or agreement that covers your staff.
The employer must also give affected employees, in writing, all relevant information about the changes, including their nature and expected effect, and must promptly consider matters they raise. A separate clause requires consultation before changing an employee's regular roster or ordinary hours. This is administrative guidance only; for a specific decision, contact the Fair Work Ombudsman or an employment adviser.
How should you talk to staff about it?
Start before you choose a provider. Show the front desk team the call data that prompted the idea, such as missed calls during busy periods or after hours, and ask which calls they would hand over first. Involve them in writing the greeting and the rules, and in making test calls. Be clear about what is and is not changing in their hours and duties. The Australian Government's Guidance for AI adoption recommends engaging stakeholders early, especially staff and customers, and keeping a register of the AI systems a business uses.
How do you measure whether the split works?
Agree a few measures before launch and compare a month before with a month after. Useful ones are missed calls, callbacks completed the same day, interruptions while serving visitors and complaints about phone handling. Ask the receptionist for their view as well: how long the morning review takes and which AI answers they keep correcting. Set a review date and change the split if the numbers or the team say it is not working.