Case study
Voice AI That Takes the Order, Start to Finish
A conversational voice assistant that lets customers of one of the largest pizza delivery chains call in and place a complete order by simply talking — naturally, accurately, and around the clock.
Trusted by
- CLIENT
- Pizza Delivery
- COMPLETED
- 2026
- LOCATION
- United States
The brief
What They Needed from Averlynx
Phone orders for one of the largest pizza delivery chains are handled by a dedicated call-center team rather than by in-store staff, and that team carries a heavy, uneven load. At peak times more customers call than the team can answer at once, callers wait on hold, and some hang up and order elsewhere.
They needed an assistant that could handle routine ordering calls end to end, working alongside the call-center team — answering right away, taking the order accurately, and staying available at any hour without putting the customer on hold.
Our involvement
- Voice AI
- Backend Engineering
- RAG & Search
Technologies
- VAPI
- Azure OpenAI
- Google Gemini
- Qdrant
- Redis
- FastAPI
- Twilio
- Docker
Why us
Why Did They Come To Averlynx
Phone ordering was the bottleneck. Demand spiked around meal times and weekends, staffing for the busiest moments was costly and hard to scale, and every mistaken order carried a real cost.
High volume, long waits
At peak times more customers call than the team can answer at once. Callers wait on hold, and some hang up and order elsewhere.
Peaks are hard to staff
Demand spikes around meal times and weekends. Staffing the call center for the busiest moments is costly, and hard to scale up and down.
Mistakes are expensive
A wrong size, an unavailable topping combination, or a missed deal leads to a remake, a refund, and a customer who may not call back.
Uneven coverage
Service quality varies with how busy the team is and who picks up, and calls outside staffed hours can go unanswered.
What we built
How We Helped — a Voice Assistant Built on VAPI
- 01
Capture the details
Collect the customer’s name, phone number, and whether the order is for delivery or pickup, right at the start of the call.
- 02
Validate the address
For delivery, check the address against the Google Maps API so it is a real, deliverable location before the order goes through.
- 03
Take the order
Understand items in plain language — sizes, crusts, toppings, half-and-half pizzas, quantities, and changes mid-conversation.
- 04
Know the menu
Look up what’s available, match the request to the real menu, and offer alternatives when something isn’t possible.
- 05
Apply deals & upsell
Recognize when a deal fits the order, walk the customer through it, and suggest sides naturally.
- 06
Take secure payment
Accept credit and debit card payments — encrypted and handled in line with US payment-security requirements.
Key results
What changed
The assistant only confirms an item once it has been checked against the live menu and pricing, and hands off to an agent the moment a caller asks for one or a request falls outside the routine path.
No waiting
Calls are answered right away, at any hour, instead of queueing behind a busy call-center team.
Less routine load
Agents are freed to focus on the calls that genuinely need a person.
Right first time
Every confirmation is menu-checked and correctly priced before the order reaches the store.
Before and After
Before
- Callers wait on hold at peak times, and some hang up and order elsewhere.
- Staffing the call center for the busiest moments is costly and hard to scale.
- Wrong sizes and unavailable topping combinations lead to remakes and refunds.
- Calls outside staffed hours can go unanswered.
After
- Routine calls are answered immediately, at any hour, and scale with call volume.
- Agents handle only the calls that genuinely need a person.
- Every item is checked against the live menu and pricing before it is confirmed.
- A request to speak with someone transfers straight to a call-center agent.
In closing
A voice assistant that handles routine ordering calls for one of the largest pizza delivery chains accurately and consistently, alongside the call-center team, at any hour.
