Case Study

The Client

A services business with a high volume of inbound calls and a small front desk. Calls outside office hours went unanswered, and most of them were the same handful of questions: opening hours, availability, booking or moving an appointment.

Goals

No call goes unanswered; repeat questions get answered automatically; and appointments land straight in the calendar instead of being keyed in afterwards.

Challenges

Natural speech — with names, addresses and dates — had to be recognised reliably. The agent also had to know when it was out of its depth and hand over to a person rather than keep trying.

Our Solution

Speech recognition: The agent handles natural speech, reads names and dates back to the caller for confirmation, and asks for clarification when it is not sure.

Booking into the calendar: It checks live availability, offers times, books the appointment and sends a confirmation — with no second data entry.

Answers to common questions: Hours, location, services and procedures are answered from content the business controls, not from free-form generation.

Hand-off to a person: When the request is complex, or the caller asks, the call is transferred along with a summary of the conversation so far.

Implementation

We catalogued the most common reasons people call and scripted each one.

We connected the agent to the phone system and the appointment calendar.

We ran the agent out-of-hours only at first, so it could be judged without risking daytime calls.

We expanded into peak hours gradually, with human oversight and a weekly review of recordings.

Results

Next Steps

Connecting to the ERP so the agent can also answer order-status questions.

Adding a second language for international customers.