October 21, 2021  

Artificial Intelligence at PepsiCo

In its quest to create the perfect cheeto, steer sales teams to the most efficient routes, and much more, PepsiCo is finding innovative ways to use AI every day.

If you’ve ever asked Siri or Alexa a question, had a virtual chat with customer service or logged on to social media, you’ve used artificial intelligence. AI has become infused in daily life in small ways, like autocorrecting a text message, and large, like self-driving cars.

“AI has changed how we interact with almost every company,” says Athina Kanioura, PepsiCo Chief Strategy and Transformation Officer. “And now businesses have systems that have intelligence behind them that have transformed the way we solve problems, engage with consumers and make products.”

At PepsiCo, AI has become a vital tool for business and employees alike. It’s a fundamental part of PepsiCo’s digital transformation as the company builds on its digital resources: Starting with the best time to plant potatoes all the way to predicting how many bags of Lay’s should go on store shelves, AI is reshaping how the company plans, makes, moves, sells and delivers products. With all the potential uses on the horizon for AI, integrating the emerging technology into every facet of how the company operates has become important part of taking PepsiCo into the future.

Learn more about the surprising and helpful ways AI is impacting the world — and how PepsiCo is using this technology every day. 




If it seems like technology is advancing at a whiplash pace, you are correct. “The cycles of development have been reduced from years to months,” Kanioura says. But the concepts that AI is built upon remain relevant, and can help you understand what it’s all about: 

1. AI is a way of organizing data.

“Only five percent of the world we live in is structured data. The other 95 percent is unstructured data,” says Magesh Bagavathi, PepsiCo SVP and Chief Technology Officer. AI takes disorganized information — the other 95% — and turns it into formulas. Bagavathi shares one example: AI can look at a face and measure all its curves and angles, then detect patterns and learn how those curves and angles will move. With that data, it’s possible to construct a simulation of the face and recreate its expressions.

PepsiCo did just this with Lay’s Messi Messages, a world-first innovation that allowed fans to create personalized messages straight from the soccer star’s mouth. After filming Lionel Messi, his features were mapped so AI could predict how he would look and sound. With this technology, more than 4 million unique messages were sent in 10 languages — all based on five minutes of recorded footage. 

2. AI can help teams predict the future.

Since AI is about organizing massive amounts of information, experts can analyze all that data and spot future trends. “PepsiCo has always been using technology to engage with our consumers,” Kanioura says. “Now, because of AI we can understand what consumers want and how they want to be engaged. We have millions of data sets to tailor our products and create experiences that are highly customized.” 

AI can dramatically decrease product innovation cycles so PepsiCo teams can respond to consumer demand, in shortened timelines. Insights revealed people were discussing, searching for and ordering seaweed products online thanks to an AI tool that analyzed millions of social posts, recipes and menus. That led R&D to develop Off The Eaten Path seaweed snacks in less than 12 months. Similarly, AI insights showed that consumers were interested in immunity. Six months later, Propel with immunity-boosting ingredients was on store shelves, ready to sip. 

3. AI is critical to sustainability.

It’s one of the most significant opportunities, Kanioura says. “As a company, we are making significant commitments in the sustainability space,” she notes. “But with this ambition we need the right enablers, and AI becomes an enabler of how we protect the environment.”

Kanioura sees farming communities as one of the biggest areas of impact. And PepsiCo has found ways to help farmers use AI in the field. Growers in North America, Latin America and Europe have collected more than one million key data points about the potatoes they plant – everything from the seed they use to how deep they plant and how much they water. With machine learning, farmers can discover improvements they can make to increase productivity and optimize yields.

These smarter farming practices not only make growers more profitable, but also mean the crops are grown in a more sustainable way – requiring less water and fewer pesticides while also reducing greenhouse gas emissions. The more data available, the more sustainable PepsiCo’s farming practices become. 

4. AI is designed to enhance human capabilities.

One common misconception is that AI will replace the need for humans, but Kainoura explains this not the case. “AI technologies go hand-and-hand with human ingenuity,” she says.

Kanioura says the most innovative AI technology allows people to be more productive by removing repetitive tasks and allowing individuals to focus on more skilled functions, like interpreting data insights and driving business decisions.

At many of Frito-Lay's plants, technology monitors and tracks data on the plants’ machines to predict any mechanical failures before they happen. After one year, the plants saw zero unexpected breakdowns or interruptions. With the help of AI, mechanics were able to be more efficient with their time so they could focus on planned maintenance and be proactive before equipment failed.  


AI is even helping design a Cheeto where nothing is left to chance. “There is a lot of R&D and engineering going into making sure every time you open a bag of Cheetos, they’re perfect,” says Denise Lefebvre, PepsiCo SVP Foods R&D. 

“We study the crunch, the texture, how it melts in the mouth...even the curve of the arc is important,” she explains. There’s also a formula for the precise amount of Cheetle (the signature cheese dust) coating each puff.

Natural variations in corn, cheese and cooking temperatures can cause irregularities. Typically, these are spotted while Cheetos are being cooked and moving through the processing line. Any issue means manufacturing stops so a new batch can be created — inefficient in terms of time and ingredients.

Now PepsiCo is using AI to crack the code for the perfect puff. “We’ve trained a machine ‘brain’ to measure the Cheetos, look at them and detect when adjustments are needed,” Lefebvre says. Engineers, machine operators, product developers, food scientists and more teamed up to create an algorithm, or the “recipe” for the model Cheeto. A computer then scans each Cheeto as it comes down the line. Is the arc too straight? Does it look too dense? Not puffy enough? The computer responds in nanoseconds to make adjustments to temperature, shape and consistency, all while keeping the flow of Cheetos virtually uninterrupted.

The technology will first be tested on Cheetos in Spain, with the intent to expand globally. It could even be applied to other products like Cap’n Crunch and Doritos. “This is the ideal application for AI because we know really well what consumers love and value about Cheetos and we can train a system to understand that.” Lefebvre says.


“I think we’re just scratching the surface,” Bagavathi says. “We’re going to keep seeing more ways AI will simplify our lives and make us more productive.” Not only does AI present opportunities to remove repetitive and time-consuming tasks, it can also keep people safer on the job.

One way AI might reshape how PepsiCo sells products is by helping create the “perfect store.” Bagavathi explains that by using a camera on a smartphone, a sales rep can take a video of store shelves and create a virtual map of which products need to be stocked in real time, all while the sales rep is standing in the store. But AI even goes one step further. “You can immediately start tailoring products based on what sells in that region, based on events in that region. There are in-depth insights on seasonality and promotions. It’s highly customized and changes the way we plan.”

And now quantum computing (the ability to process even more complex problems at exponentially faster speeds than standard computers) opens the door to the next evolution of AI. Bagavathi sees opportunity in helping traveling sales teams fine-tune their strategies for reaching customers. “Planning the right route, right stores, right products, right sales is complicated due to all the variables involved,” he explains. “Multiply that by the number of routes we do on a daily basis and it becomes gargantuan.” But with a quantum computer, you can run simulations to find the most efficient routes in seconds. 

Despite all the new ways experiences will be digitalized, people remain a key aspect of the AI equation. “We have a lot of data sets available from the markets, sectors, segments and different systems, but you still need somebody to make meaning out of all these data sets,” Bagavathi says. 

Kanioura notes that driving this type of change across PepsiCo requires upskilling for every part of the organization. “We will provide a platform to train individuals to be able to make the most use of these technologies, from a frontline merchandiser to a revenue management officer,” she says. “Both of them need to understand what AI means for the specific job that they are asked to do.” 

In order to equip employees with the tools, knowledge and qualifications to work with AI, PepsiCo is creating a Digital Academy that will offer a wide range of skills trainings on topics ranging from machine learning and robotics to cloud computing and virtual reality. “We have a responsibility to employees,” Kanioura says. “The ability to upskill opens up opportunities to change faster into other functions within PepsiCo, and even career opportunities outside the company.”

There’s no denying that AI will have vast implications for generations to come. PepsiCo is perched at the forefront of these innovations. As Bagavathi puts it, “The opportunities for ways we can use AI are exponential and unlimited.”