Grocery retail executives stand at a sea change moment. AI is no longer a futuristic concept, but a present-day imperative reshaping every industry.

A recent State of AI in Retail and CPG survey reveals that 69% of industry leaders have witnessed revenue growth, and 72% have experienced reduced operating costs, directly attributable to AI. These are more than statistics – they are an urgent call for strategic investment in the technologies that are redefining the future of retail.

However, Sucharita Kodali – Forrester VP and Principal Analyst – remains skeptical. She questions whether executives will “actually open their wallets and spend more on the technologies they need to differentiate themselves.”

Executives’ caution is warranted. Without the technical expertise to distinguish AI that adds real value from marketing buzz, how are executives to know which solutions will offer a return on their investment?

In January 2024, two of Afresh’s co-founders, Chief Technologist Volodymyr Kuleshov and CEO Matt Schwartz presented on this topic at the Fresh Foods Leadership Council Meeting ahead of FMI Midwinter. This blog recaps their insights on the transformative role of AI in grocery through Afresh’s own solutions, and beyond.

Afresh’s AI Use Cases in Grocery

Amidst a sea of demand forecasting solutions, Afresh’s point of difference lies in how we apply AI – including hidden Markov models and neural networks – to tackle the unique challenges of fresh inventory. 

Unlike traditional methods that rely on static inventory counts, our approach is purpose-built for the unpredictable nature of fresh, accounting for factors like spoilage and misscans. Afresh's AI has already given partners like Albertsons Companies a strategic advantage by ensuring optimal stock levels and minimizing shrink without the need for intensive manual effort. 

Let’s take a closer look at how we apply AI to inventory, demand forecasting, and store ordering:

1. Inventory Estimation

Inventory underpins everything from demand forecasting to ordering, so having an accurate inventory count is critical. However, when it comes to fresh, data is notoriously unreliable, and manual efforts to true up inventory are both tedious and costly.

Afresh's patent-pending inventory estimator (InvHMM) uses innovative techniques inspired by hidden Markov models. This approach uses a probabilistic model to distinguish between recorded data and unrecorded factors – such as shrink due to spoilage or misscans – and then prompts store associates to conduct targeted inventory checks to improve accuracy. This unique approach allows grocers to both minimize shrink and boost in-stock rates.

2. Time Series Demand Forecasting

Afresh’s demand forecaster uses a deep neural network to provide a probability distribution describing the possible demand ranges. This allows store associates placing orders to weigh the trade-offs between overstocking and underordering. For example, Afresh’s forecaster may predict that: “Over the next three days, there’s a 90% chance we’ll sell at most 50 pineapples, an 80% chance of selling at most 45, a 70% chance of selling at most 40, etc.”

Our models use a variety of neural network variables to generate order recommendations, including:

• Historical demand over the past few months
• Time of year and day of week or month to account for patterns, seasonality, and holidays
• The current price, future price, and promotion potential
• Item classification features to identify patterns between similar items

By balancing these factors, Afresh finds the optimal order quantity that aligns with the store's specific decision-making criteria, ensuring that stores not only meet consumer demand efficiently but also manage inventory in a way that minimizes waste and maximizes sales opportunities.

3. Automating Store Orders

Traditional ordering policies rest on the precarious assumption that inventory counts are always accurate. At the same time, many solution providers focus their efforts solely on refining demand forecasts. However, even the most accurate demand forecasts are futile when paired with flawed inventory data. The resultant order recommendations are inevitably skewed, mirroring the inaccuracies of the perpetual inventory counts.

Afresh confronts this challenge head-on with a fundamentally different approach. Building upon our time series demand forecasting, we provide holistic scenario testing to evaluate the full spectrum of variables affecting store orders. This includes sales data, but also perishability, delivery schedules, and unique store preferences for stock levels.  This rigorous analysis culminates in placing profit-maximizing orders.

AI Applications Beyond Fresh

There are plenty of other use cases for AI in grocery beyond fresh departments. Here are three of the most promising:

1. AI for Operations & Scheduling

On the operational side, AI is revolutionizing workforce management by enabling dynamic staff planning and scheduling. Systems like Logile accurately forecast labor needs based on sales volume and store characteristics to ensure optimal staffing, improve operational efficiency, and enhance the employee experience. This AI-driven approach enhances customer service by aligning workforce allocation with real-time operational demands.

2. AI for Personalization

As seen in solutions like EagleAI and GK Engage, AI enables retailers to harness customer data and current inventory to generate personalized promotions, creating a tailored shopping journey. This technology ensures that customers receive relevant recommendations and offers across various touchpoints, from mobile apps to checkout kiosks, enhancing their experience and satisfaction while aligning with retail goals.

3. AI-Powered Autonomous Checkout

Companies like Kwikkart use AI to fuel automated fulfillment technology for delivery drivers, increasing the efficiency and accuracy of deliveries while reducing shrink. Meanwhile, Standard AI’s computer vision tracks customers and the items they place in their carts to enable “checkout-free” shopping. These solutions not only enhance the shopping experience but also increase sales.

Beyond AI Innovation: Shaping A New Era of Smart Grocery

Afresh is setting a new standard for operational excellence in fresh. Our AI equips grocery retailers with the tools to reduce shrink, ensure optimal stock levels, and facilitate smarter, data-driven ordering. In this way, our leading-edge AI not only elevates operational efficiency but also enhances customer satisfaction by fueling an exceptional in-store experience.

Join us at the Annual Meat Conference on March 19 for CEO Matt Schwartz's presentation: “The Use and Application of Artificial Intelligence – How to Make this Advanced Technology Work for Your Business.” In this talk, Matt will explore AI's capacity to streamline the supply chain and optimize retail operations.

‍

‍

‍