A conversational AI voice agent that takes pizza orders, handles customizations, and processes payments via natural phone calls.
Client Overview
A growing pizza chain with 12 locations across two cities was losing a significant volume of revenue every week due to unanswered or abandoned phone calls during peak hours. Dinner rushes, Friday evenings, and weekends routinely overwhelmed the front-of-house staff who were simultaneously managing in-store customers, coordinating with kitchen teams, and attempting to answer incoming calls. Call hold times exceeded five minutes on average during busy periods, and the chain's own internal data showed that nearly one in three inbound calls during peak hours ended without an order being placed. The problem was compounded by human error during order-taking. Staff distracted by in-store activity frequently misheard pizza sizes, topping combinations, or delivery addresses, leading to remakes, refunds, and negative reviews. The chain's management had attempted to address this by adding a dedicated phone line and rotating staff to phone duty, but this created its own scheduling and cost pressures without fully solving the call abandonment problem. With no cost-effective way to staff the phones to the level required during peaks, management recognised that the solution needed to be technological. They wanted a system that could answer every call immediately, take complete and accurate orders, handle customisations and substitutions conversationally, and push confirmed orders directly into the kitchen management system without any human involvement.
Our Approach
Zentric Solutions built a conversational AI voice agent using Vapi AI and OpenAI GPT-4 that answered every inbound call to the pizza chain's phone lines within seconds, regardless of how many calls arrived simultaneously. The agent was trained on the full menu across all 12 locations, including size options, crust types, topping combinations, combo deals, and seasonal specials, allowing it to handle virtually any order conversation without escalation. The voice was designed to be warm and natural, with a conversational flow that guided callers through menu selection, customisation, and order confirmation without feeling scripted or robotic. The agent integrated directly with Twilio for call routing and the chain's existing POS system via REST APIs, ensuring that confirmed orders were sent directly to the kitchen management display in real time with no manual re-entry required. Payment processing was handled via Stripe API for phone payment capture, while callers who preferred in-person payment at pickup were accommodated through a simple confirmation flow. The agent also captured delivery addresses and validated them against delivery zone boundaries, automatically quoting accurate delivery time estimates based on current kitchen load. Order accuracy improved dramatically because the AI repeated every item back to the caller before confirming, giving customers the opportunity to make corrections before the order was transmitted. Upselling logic was built into the conversation flow — the agent suggested drinks, sides, or promotional deals where appropriate without being aggressive. The result was a measurable increase in average order value alongside a near-elimination of missed calls during peak periods.
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.
