Wendy's
drive-thru order-taking voice agent with upselling
Wendy's FreshAI, a drive-thru order-taking voice agent developed with Google Cloud, equipped more than 160 restaurants in early 2025 for a target of more than 500, processing tens of thousands of orders per day.
Objective
Take drive-thru orders with a voice AI built with Google Cloud to smooth service, reduce errors, and push upsell suggestions that raise the basket, while freeing up crew members.
The deployment
FreshAI is Wendy's drive-thru order-taking voice AI, developed with Google Cloud and launched in test in 2023. It understands natural language, handles millions of order and personalization combinations, and proposes complementary items based on the order in progress. The system started as a pilot in two states in 2024, then a franchise program. In early 2025 it equipped more than 160 restaurants, with a target of more than 500 restaurants over the year, processing tens of thousands of orders per day. Wendy's has since added Spanish.
The case in action
Official videoWendy's FreshAI au drive, avec Google Cloud · voir sur YouTube
Results Proof C
Established restaurant press and official brand communications naming Wendy's, the number of restaurants, and the partner Google Cloud, with a comment from CEO Kirk Tanner on earnings about the digital mix. FreshAI's own metrics (accuracy, wait time) are not published in figures, hence C rather than A.
How it works
Documented architectureThe stack in detail
- plateforme Google Cloud Technology partner: FreshAI's large language models and voice components run on Google Cloud.
- outil FreshAI Wendy's custom voice agent on Google Cloud: order understanding, millions of personalization combinations, suggestion of complementary items.
- infra Chaine vocale ASR / TTS Voice recognition and synthesis at the drive-thru microphone, with human takeover when the model stumbles (accent, noise, off-menu).
- infra Menus digitaux et POS Wendy's Display of the order for customer confirmation and transmission to the kitchen without re-entry.
How it runs, concretely
For ops teams-
1Customer speaks Customer
The customer orders at the drive-thru microphone.
-
2Understanding and suggestion AI
FreshAI understands the order, handles personalization, and proposes complementary items to raise the basket.
-
3Display and confirmation Digital menu
The order appears on the digital menu for checking.
-
4Human takeover if needed Crew member
A crew member takes over the conversation if it gets stuck.
-
5Preparation Crew member
The validated order goes to the kitchen.
Voice understanding accuracy and human takeover rate. The model must handle millions of personalization combinations; if it stumbles on an accent, noise, or an off-menu request, a crew member takes over.
How your customers perceive this type of use
Sourced studiesLes consommateurs n'acceptent pas les chatbots par defaut : 64% prefereraient que les entreprises n'utilisent pas d'IA dans leur service client (Gartner, 2024) et pres d'un utilisateur sur cinq du service client par IA n'en retire aucun benefice (Qualtrics, 2025). L'acceptation se construit sur trois conditions mesurees par Salesforce : savoir qu'on parle a une IA, pouvoir escalader vers un humain, comprendre la logique de l'agent.
Acceptance conditions
- Etre informe qu'on parle a une IA et non a un humain (pres de 75% le demandent, Salesforce 2024)
- Un chemin d'escalade clair vers un agent humain (45% plus enclins a utiliser l'agent IA, Salesforce 2024)
- Une logique de l'agent clairement expliquee (44% plus enclins, Salesforce 2024)
Red lines
- Rendre l'humain injoignable : c'est la premiere inquietude des consommateurs sur l'IA dans le service client (Gartner 2024) et 50% craignent que l'IA les coupe du contact humain (Qualtrics 2025)
- Remplacer le service client par l'IA sans alternative : 53% envisageraient de partir chez un concurrent (Gartner 2024)
Sources: Salesforce 2024 · Gartner 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- structured menu with personalization rules
- audio recordings to train and evaluate the ASR
- catalog of complementary items for upselling
Org prerequisites
- integration with the POS and digital menus
- human takeover process
- franchises on board
Possible stack
- LLM and voice AI (Google Cloud, other vendors)
- connected digital menus
- integrated POS
The plan, step by step
- Step 1Structure the menu and personalization rules into machine-usable data.Deliverable: Structured menu with valid combinations and suggestion rules.
- Step 2Integrate the voice chain (microphone, ASR, LLM, TTS) into a lab restaurant, human takeover by default.Deliverable: Prototype taking real orders under supervision.
- Step 3Extend the pilot to a handful of restaurants and measure.Deliverable: Dashboard of order accuracy / human takeover rate / service time.
- Step 4Integrate POS and digital menus, activate upselling.Deliverable: Order sent to the kitchen without re-entry, basket tracked.
- Step 5Open the rollout to franchises with training and support.Deliverable: Multi-restaurant deployment with central metrics tracking.
First step: Test on a handful of restaurants with human takeover by default, measuring accuracy and service time before opening to franchises.
The same pattern, proven in other industries
Amazon
genAI shopping assistant
Booking.com
genAI trip planning assistant
Burberry
genAI copilot for consultants (search, image recognition, complete-the-look)
Sources
- S1 Transforming the Ordering Experience: Wendy's FreshAI Update Interested party archive pending
- S2 Wendy's to deploy drive-thru AI to over 500 restaurants this year Established press archive pending
- S3 Wendy's Plans Major Tech Investments to Improve Restaurant Order Accuracy and Speed Established press archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.