An AI voice agent that handles inbound taxi booking calls, captures pickup/drop-off details, and dispatches rides automatically without human operators.
Client Overview
A privately owned taxi company operating a fleet of 80 vehicles was haemorrhaging bookings due to its inability to handle call volumes during morning and evening rush hours. The company employed a small team of dispatch operators who managed both inbound bookings and driver coordination simultaneously. During peak windows, callers were placed on hold for four to seven minutes on average, and the company estimated that between 25 and 35 percent of peak-hour callers hung up before completing a booking, many of whom switched to competitor apps or ride-hailing platforms permanently. The dispatch operation was also prone to booking errors. Operators taking rapid-fire calls during busy periods frequently misrecorded pickup addresses, confused similar-sounding street names, or failed to note passenger-specific requirements such as wheelchair accessibility or large luggage. These errors resulted in driver no-shows, passenger complaints, and repeated refund requests that strained both the operations team and customer relationships. The company's management had explored hiring additional operators but found the cost unsustainable at the margin, particularly for the relatively short daily peak windows that drove the majority of call volume. They needed a solution that could scale call-handling capacity instantly without proportional staffing costs, while maintaining the personal and accurate service their regular passenger base expected.
Our Approach
Zentric Solutions deployed an AI voice booking agent built on Vapi AI and OpenAI GPT-4 that answered every inbound call to the company's booking lines immediately and conducted a natural, conversational booking flow. The agent captured the caller's name, pickup address, destination, requested pickup time, and any special requirements — such as accessibility needs or child seats — through a guided dialogue that felt relaxed and unhurried regardless of how many simultaneous calls were in progress. Address accuracy was validated in real time using Google Maps API, which confirmed that the captured address was a genuine, geocoded location before the booking was committed. Once a booking was confirmed, the agent transmitted the full booking details — including validated coordinates, passenger name, contact number, and special requirements — directly to the company's dispatch system API. The dispatch system then automatically identified and assigned the nearest available appropriate vehicle, sent the driver the booking details, and triggered an SMS confirmation to the passenger with the driver's name, vehicle registration, and estimated arrival time. The entire process from call answer to driver assignment happened without any human operator involvement. For complex requests that fell outside the agent's configured scope — such as account billing queries, corporate account management, or multi-stop journeys with unusual requirements — the call was cleanly escalated to a human operator with a context summary, ensuring that staff time was preserved for cases that genuinely required human judgment. Operators were freed from routine booking capture entirely, allowing them to focus on fleet coordination, driver management, and exception handling. Call abandonment during peak hours dropped to near zero within the first month of deployment.
Tech Stack
Project Tags
Everything you need to know about this project and our approach.
Zentric Solutions delivers cutting-edge digital products that streamline operations, enhance engagement, and drive lasting growth.
