With the increasing application of artificial intelligence technology, it is no longer a competitive advantage of enterprises, and has become a necessity. As the world’s leading beverage and casual retail giant, PepsiCo has taken the lead in realizing this.
PepsiCo owns many brands such as Pepsi, Gatorade, Pure Fruit, Lipton, Frito-Lay, and Quaker. The products of these brands are continuously sold to more than 200 countries around the world, and earned $64.7 billion in revenue for PepsiCo last year.
The public enjoys the delicacies provided by PepsiCo, but few people know that it has used new technologies such as artificial intelligence and big data in all aspects of production and operation.
Vending Robot: Snackbot
On the campus of the University of the Pacific, there are so many robots that have six wheels and are filled with PepsiCo’s drinks and snack foods. These beverages and foods are provided by PepsiCo’s vending brand, Hello Goodness, which includes products from Frito’s SunChips and Starbucks.
These robots, called Snackbot, were developed by PepsiCo and its partner, Robby Technologies, a technology company based in the Gulf.
These robots will be on standby from 9 am to 5 pm. Students can order drinks and food on the campus via the appropriate mobile app, and the robot automatically delivers the order to one of the more than 50 delivery points on campus.
Snackbot can travel 20 kilometers on a single charge, and it uses an all-wheel drive to deal with curbs and ramps. The body is equipped with a camera and lights to work even in the evening or on rainy days.
Snackbot offers a new retail approach that provides a friendly and convenient solution for college students who want to eat while they are busy with schedules, and can help PepsiCo better understand consumer preferences.
Intelligent production control system
The application of artificial intelligence technology in the manufacturing process is also very extensive. One of the beneficiaries of the manufacturer of PepsiCo, the brand of PepsiCo.
In the process of fries production, Fitoli uses the laser to illuminate the French fries, allowing the algorithm to analyze the texture of the fries according to the reflected light to determine whether the fries processing system is functioning properly.
This project has enabled Pepsier Mirza, a senior research and development engineer at PepsiCo, to realize that machine learning can also play an active role in more aspects of factory production. So he developed a machine learning model with a computer vision system that predicts the weight of the potatoes being processed. In this way, the company does not have to spend $300,000 on each production line (Fidley has only 35 production lines in the US) for weighing. Since Shameer Mirza’s system only includes a camera and machine learning model, there is almost no additional cost, just need to add some data collection points.
In addition, Frito-Lay is developing a system to evaluate the “peeling rate” of potatoes. Based on the data analyzed by the system, Fitoli can optimize the potato peeling process. It is estimated that in the United States alone, the system can save the company more than $1 million annually.
PepsiCo will also offer a training course on machine learning and computer vision for internal R&D personnel this year to enhance the team’s ability to apply these new technologies and continue to find new ways to optimize production efficiency.
Smart Recruitment Assistant: Vera
PepsiCo’s human resources department needs to fill 250 job vacancies in the Russian factory within two months, so it began to use the robot Vera to interview sales, drivers and other positions.
Vera is a smart recruiting robot developed by Russian startup Staffory, which can interview 1,500 job seekers in 9 hours, and it takes 9 weeks for humans to complete the job.
Vera integrates Amazon, Google, Microsoft and the Russian technology company Yandex’s advanced speech recognition algorithm, which automatically calls candidates to screen candidates for vacant positions such as forklift operators, factory workers and salespeople.
Vera can scan the candidate’s resume to determine whether he has the qualifications and experience of the relevant position, use the “yes” and “no” to answer the candidate’s questions and ask follow-up questions, and send an email to the candidate. In the process of communicating with the candidate, it will be recorded throughout the process and then sent to a human resources expert for further review.
At present, most candidates are very active in recruiting robots, and the human resources professionals are hesitant. As a result, the biggest obstacle to recruiting robotic applications is “reprogramming humans” to make them more accustomed to this new technology.
Internal technology platform: Ada
Ada’s philosophy is to combine human insight and algorithms to achieve “enhanced intelligence.” PepsiCo hopes to increase its learning speed in this way.
Ada can collect data from a variety of sources to help PepsiCo better explore the value of data for all aspects of the company’s operations, including innovation, design, R&D and price decisions.
In addition, PepsiCo also uses big data for new product development, sales and marketing. For example, use social forecasting tools to combine publicly available consumer conversations with internal big data to identify new products to market. This means that PepsiCo has been able to gain insights from existing data and apply it to business practices. In the future, as PepsiCo precipitates more and more data, it can also predict changes in consumer trends and constantly adjust its marketing positioning.
Big Data Analytics Platform: Pep Worx
Pep Worx, a cloud-based big data analytics platform, can help PepsiCo provide the best inventory advice to retail stores: what products they recommend, where they are placed, and what promotional strategies to use.
For example, when PepsiCo intends to launch Quaker Overnight Oatmeal, the Big Data platform can screen 24 million accurate audiences from 110 million American households. At the same time, it can also analyze where these families tend to go shopping and launch targeted promotions to attract them.
This scenario-specific big data application drove the product to achieve 80% sales growth in the 12 months after its launch.