Quick Service Restaurants and Robots: From Autocado to AI

Several weeks ago, I shared my observation on how for many industries, automation may actually increase employment before significantly decreasing it. 

The reality, however, is that the outcome will be very industry-dependent. 

An industry I’ve been following closely is QSR (quick service restaurants). 

One of this industry’s main features is that it tries to be fast and cost-efficient while maintaining a high volume of interaction. In operations, one can be responsive or cost-efficient, but it’s hard to be both —albeit not impossible. So QSR will always experiment with reducing costs while increasing responsiveness. 

As the QSR industry strives to meet evolving customer expectations and operational challenges, integrating robots into food preparation and service operations is gaining momentum. 

Enter the Autocado. 

Google keeps correcting me: 

No Google... I mean Autocado:

Chipotle’s recent introduction of robots like the Autocado for preparing guacamole and “Augmented Makeline” for assembling bowls and salads, marks an interesting shift toward automation in the fast-casual dining space. 

Today’s article explores the current state of robotics in QSRs, the cost-benefit analysis behind adopting such technologies, and the role of AI in transforming operations.

The Robotics Revolution in QSRs

Robots like Chipotle’s Autocado are aiming to revolutionize labor-intensive tasks. The Autocado can halve, core, and peel an avocado in just 26 seconds, significantly reducing the 50 minutes it typically takes employees to prepare a fresh batch of guacamole. 

Given that Chipotle processes approximately 98,667 avocados per restaurant annually, the time saved per location is substantial. This frees up time so employees can focus on customer-facing tasks or food preparation that requires a human touch, thereby enhancing operational efficiency.

Similarly, Chipotle’s “Augmented Makeline,” which assembles bowls and salads, automates repetitive tasks while maintaining the quality and consistency that customers expect. This system is designed to dispense the exact amount of ingredients needed, ensuring accuracy, especially for digital orders, which constitute about 65% of Chipotle’s orders. Such innovations could reduce complaints related to portion size and improve customer satisfaction —a growing concern for the chain.

Autocado: Detailed Cost Analysis

Those who follow my newsletter know that one of my all-time favorite books is The Goal. For those not familiar with the book, a highlight in the story is when the plant manager meets his college professor and brags about how the new robots they installed have improved things. This is met with skepticism from the professor who asks the following questions: 

1. Did you sell more?

2. Did you reduce the amount of inventory? 

3. Did you fire anyone?

The young manager gives a negative response to all three questions and the professor concludes that the robots didn’t improve anything. 

So let’s use this example on Chipotle. 

Starting with point (3), Will the recently implemented automation help Chipotle cut its labor costs? 

In QSR operations, labor costs are a significant portion of the variable costs per order. If automations like the Autocado can reduce the per-order labor costs, it directly improves operational profitability and efficiency. For any technological investment to be valuable, there must be a reduction in the per-order cost by a sufficient margin, to cover the technology’s capital cost over time.

Labor Cost Per Order: Based on Chipotle’s 2023 financials, their labor cost per order is approximately $4.21 (considering the total labor cost of $2.44 billion spread across the number of orders processed).

Labor Cost Reduction Target: We can calculate potential savings by assuming that the Autocado allows Chipotle to reduce its workforce by one employee per shift (specifically, for handling avocados). Let’s assume each employee earns $20 per hour.

If a restaurant processes around 29 orders per hour (based on 167,009 orders per year), the cost per order attributable to that one employee is 20/29≈0.69. So, by reducing the need for one employee, Chipotle can save approximately $0.69 per order.

Cost of Technology: Let’s assume the Autocado costs $500,000 (this is just an estimate so we can put something on this back-of-the-envelope math), and its cost is amortized over 5 years. With 835,045 orders processed over that period, the cost per order attributable to the robot is around $0.60.

Net Savings per Order: By reducing labor costs by $0.69 per order and accounting for the $0.60 per order cost of the robot, due to automation, each order generates a net savings of $0.09.

Break-Even Analysis: For the investment in the Autocado to be worthwhile, the labor savings must exceed or match the technology’s cost. In this case, since the robot saves $0.09 per order after accounting for its own cost, the investment is financially viable over its lifespan.

Note that my assumptions don't account for machine maintenance and the fact that an employee may still be needed to transfer the avocados to the “production line,” or the possibility that the machine costs less to acquire. In addition to labor savings, robots like the Autocado offer the potential of reduced food waste and improved efficiency, especially when handling high-demand ingredients like avocados. Chipotle alone expects to use 5.2 million cases of avocados this year. A robot’s ability to handle these quantities with greater precision and consistency will likely translate into reduced waste and higher yields, further improves ROI. But still, this will be reflected in a few cents difference. 

Regarding points 1 and 2 from the professor’s questions above, note that I don’t think the improvement in inventory or sales will be meaningful. Avocado peeling is not the bottleneck at Chipotle.

For technology like Autocado to be a valuable investment, it should result in labor cost reductions that surpass the robot’s cost per order. With current labor costs at $4.21 per order, achieving savings through automation could significantly boost profitability, provided the technology implementation stays within a cost boundary of about $0.60 per order. Not impossible, but also not a massive improvement. 

Domino’s AI-Powered Innovation

However, Chipotle is not the only one implementing automation. 

Domino’s is also trying to incorporate AI in the QSR industry with its AI-powered pizza-making systems. Domino’s has partnered with Microsoft, developing algorithms to predict customer behavior, and to automate inventory management and pizza preparation tasks. 

Notably, Domino’s is working on a system where pizzas are made before customers even order them, based on predictive AI models. This speeds up the process and ensures faster delivery times, which hopefully will lead to increased customer satisfaction.

Moreover, Domino’s AI assistant is intended to streamline back-of-house operations by optimizing inventory, ingredient ordering, and staff scheduling. The hope is that this will help store managers focus on customer experience rather than daily operational tasks. 

When applying the same principles to Domino’s, we can see how this can help reduce inventory. I don’t think it will save any real costs, but it does have a small likelihood of improving sales. Dominos thrives on fast service, and if pizzas can be made faster, with more pre-prep work and better employee scheduling, it can achieve better speed. But these improvements will be marginal at best. 

Let’s redo the math. 

Dominos has $69 Mil. Revenues were 1.1B in June 2024. 

Inventory Cost Reduction Target

Assuming that Domino’s implements an AI-based inventory management system, leveraging machine learning models and API calls, the goal is to reduce inventory costs. With their current inventory at $69 million and an annual revenue of $1.1 billion, a 10% reduction in inventory costs would save them approximately $6.9 million.

Currently, the inventory cost per dollar of revenue is $0.063. Reducing inventory by 10% lowers this cost to approximately $0.0567 per dollar of revenue, representing potential savings.

Cost of Technology: API Calls and AI Model Training

Let’s assume the AI-driven system incurs costs primarily from API calls and training AI models. For example:

API Call Costs: If the system makes frequent API calls to process data, and each call costs $0.005 (just an estimate), assuming the system makes 10 million API calls per year, the total cost would amount to $50,000 annually.

AI Model Training Costs: Training and maintaining AI models might cost $1 million per year, given the complexity of the models and the necessary infrastructure.

Over a five-year period, the total cost for API calls and AI model training would be:

API costs: $50,000 × 5 = $250,000 and AI training costs: $1 million × 5 = $5 million, totalling approximately $5.25 million.

With projected revenue over five years of $5.5 billion, the cost per dollar of revenue for the AI system adds approximately $0.00095 per dollar of revenue.

Net Savings per Dollar of Revenue

By reducing inventory costs by $0.0063 per dollar of revenue (a 10% reduction) and accounting for the $0.00095 per dollar of revenue in technology costs, the net savings per dollar of revenue would be:

0.0063−0.00095=0.00535

Therefore, Domino’s would save approximately $0.00535 per dollar of revenue after accounting for the AI-driven system.

Not a massive improvement. 

Again, AI sounds great. But when you start crunching the numbers, the savings are not significant.  

Challenges of AI in McDonald’s Operations

So the benefits are visible, but at this point, they are not all that significant. 

However, there are several other reasons robots have made slow inroads into QSRs. The complexity of food preparation is a challenge. While robots excel at repetitive tasks, they often struggle with the nuance required for more complex meals or customized orders.

Customer perception also plays a significant role in the slow adoption of robotics. Many customers still value human interaction during their dining experience, and robots are often seen as impersonal. Moreover, maintaining robots can be costly, and any downtime could disrupt operations in fast-paced environments.

McDonald’s has made significant strides in incorporating AI into its operations, particularly with AI-powered drive-thru systems. However, the company’s recent decision to halt its automated order-taking (AOT) system, developed in partnership with IBM, highlights some key challenges. The drive-thru AI was designed to take customer orders more efficiently, but technical issues—such as misinterpreting customer requests or adding incorrect items—caused delays and dissatisfaction. For example, McDonald’s AI system was notorious for making odd mistakes, like adding nine sweet teas to an order when only one was requested.

Moreover, the system needed help understanding varied accents, background noise, and overlapping conversations, which are common in drive-thru environments. These technical challenges underscored the need for further development before such AI solutions can be deployed on a broader scale. Despite ending the IBM partnership, McDonald’s still sees potential in AI for improving drive-thru efficiency and plans to explore new voice-ordering solutions.

A Framework for the Success of Automation in Quick Service Restaurants 

Can we generalize these cases?

To assess the likelihood of successful automation in QSRs, it’s essential to map out the core processes within these businesses.

Key Processes in QSRs:

1. Order Taking

2. Front-of-House Food Preparation

3. Back-of-House Food Preparation

4. Delivery & Fulfillment

When assessing the impact of automation and technology in quick-service restaurants (QSRs), five critical criteria are essential:

Maturity of Technology: How developed and reliable is the available technology for each process?

Cost of Technology: The investment required includes capital expenditures and ongoing maintenance.

Cost Savings: The labor and operational costs are reduced due to technology implementation.

Inventory Reduction: How automation can minimize waste and improve resource-use efficiency.

Sales Impact: The potential for increased revenue through improved speed, accuracy, and customer satisfaction.

Using these criteria, we can assess key processes in QSRs:

Order-taking is highly likely to be disrupted by technology, as AI and self-service kiosks are already mature and relatively affordable. Automating order-taking reduces labor costs, while the increased speed and efficiency can lead to higher sales. These systems also minimize inventory errors caused by order miscommunication. While McDonald’s decided to abandon them, I think we’re getting closer to the order-taking process being replaced by AI. 

Front-of-house food preparation is moderately automated, with technologies like robotic assembly systems. Although these systems are costly, they save on labor and improve portion control, reducing inventory waste. The enhanced efficiency can lead to higher customer satisfaction and, in turn, increased sales. The Infinitive Kitchen at Sweetgreen is an example of the future:

We are still many years away from having a robot make more complex food, but we are getting closer for most QSRs. The cost of the tech will have to be much lower, and the labor cost will have to increase, but we are getting there. 

Back-of-house food preparation involves complex tasks like cooking, where automation is less mature. While advanced robotics for repetitive tasks are available, they are expensive and still require human oversight. Despite the high cost, these systems offer labor savings and better inventory control by reducing waste, indirectly supporting higher sales during busy periods.

Delivery and fulfillment are seeing progress with AI for route optimization, though autonomous delivery systems are still developing. While these technologies are expensive, they reduce labor and operational costs in the long term. Improved delivery times can boost customer satisfaction and loyalty, increasing overall sales.

In summary, automation in QSRs offers some cost savings, better inventory management, and increased sales potential, though the extent of their impact varies based on the maturity and cost of the technology applied in each area.

Conclusion: A Gradual Shift Toward Automation

The success of automation in QSRs will depend on balancing technological advancements with the complexity of tasks and human interaction. 

Tasks like order-taking and basic food preparation already benefit from automation, while more complex and variable tasks, like back-of-house cooking and delivery, are further away from full disruption. Domino’s experimentation with AI-driven prediction and Chipotle’s efforts with food preparation robots offer insights into how larger chains can scale automation, provided the technology can deliver substantial cost savings and operational efficiency.

There is no question that robots and AI-powered systems represent the future of QSRs. As technology improves, the combination of robotics and AI will likely enable QSRs to meet growing customer expectations for speed, customization, and quality while optimizing costs and operational efficiency.

But, while robots won’t be making Michelin-star meals anytime soon, at least they’ll keep our avocados perfectly peeled—and maybe our jobs, for now.