Five ways retailers are getting real results with generative AI

A look at top uses cases where retail disruption is happening now

Generative AI is doing something the retail industry has never seen before. It has dramatically compressed the gap between the initial phase of euphoria and elevated expectations that accompany technological innovation and the hard work of real-world adoption and value creation.

More typically this process takes years or even decades. The internet and e-commerce emerged nearly 30 years ago, but online channel profitability remains challenging for many retailers. Cloud computing has been available for decades, but some retailers have yet to transition from on-premise computing. In contrast, generative AI has moved quickly from abstract possibilities to real-world value creation in a shockingly short amount of time. This rapid evolution was the focus of the recent webinar, “Five ways retailers are getting real results with generative AI.”

Retail experts from SymphonyAI detailed how retailers are using generative AI today to drive tangible business outcomes. The presentation, led by John Lin, Jonathan Tye-Walker, and Julian Miller, focused on five critical use cases demonstrating the practical application of generative AI in retail environments, including:

#1 Optimizing promotions using AI

ChatGPT brought generative AI into the mainstream when it debuted in November 2022; however it is what’s known as a general-purpose AI model, and it has no specialization for retail-specific tasks. To prove the point, the panelists asked ChatGPT to provide a general promotion optimization plan. But that ChatGPT-generated plan would take 10-20 weeks to implement. In contrast, SymphonyAI’s CINDE Gen AI Copilot, trained on retail data and best practices, performed complex promotional optimizations in seconds.

The Copilot combines the strengths of generative AI with retail-specific AI models. It has secure access to company data and understands retail-specific concepts, eliminating the need for extensive context in prompts. The system can analyze promotional performance week by week and provide recommended actions for numerous product groups simultaneously.

#2 Computer vision for store intelligence

Advances in AI computer vision now provide highly accurate product recognition and stock levels on store shelves, which means retailers can determine accurate on-shelf availability, whether planograms are compliant, and whether prices and promotions are accurate.

In one remarkable example shared on the webinar, a computer vision-based product discerned between packaged pineapple and mango chunks on a store shelf. Beyond straightforward product recognition, this technology can detect pricing errors, missing price labels, and even assess freshness of select products such as bananas. This opens up possibilities for automating processes like markdown pricing based on item freshness, potentially reducing waste, and optimizing sales.

#3 Streamlining category management

The webinar showcased how AI can transform the weekly category management process. Traditionally, category managers spend significant time pulling data from various systems to prepare for weekly category performance meetings. The new AI-powered approach aggregates data from disparate sources, including pricing, assortment, promotion, and supply chain systems, to provide rapid insights into category performance.

This AI-driven process allows category managers to quickly understand exactly what happened in their category, why it happened, and what actions to take next to meet business goals. The system can prescriptively identify the causes of performance changes, saving hours or even days of analysis time. This frees up category managers to focus more on strategic initiatives rather than data compilation and basic analysis.

#4 Improving product master data

The presenters addressed a common retail challenge of incomplete or inaccurate product attribute data. High quality descriptive data is crucial for effective space planning, assortment optimization, and understanding product transferability. Generative AI was shown to be a powerful tool for enhancing this data.

Using a Diet Coke bottle as an example, the presenters demonstrated how generative AI could quickly provide detailed product information, including product type, brand, flavor, size, and nutritional information. This capability can significantly improve the quality of master data, enabling more effective analytical solutions across various retail functions.

#5 Realizing the connected retail store

The fifth and arguably most significant point of the webinar involved the concept of the “connected retail store,” where AI enables seamless data integration across all retail operations. This approach allows for holistic optimization of the retail value chain, extending from the supply chain to in-store execution.

Connected retail uses AI to break down silos between different departments, thus providing a truly cohesive view of operations. It allows retailers to understand how actions in one area, such as promotions, impact other areas like inventory management and in-store execution. This holistic, interconnected view can help retailers pinpoint exact causes of performance issues and take targeted action.

Throughout the presentation, the speakers emphasized that while generative AI offers exciting possibilities, its true value in retail comes from combining it with industry-specific expertise and secure, tailored AI models. They noted that SymphonyAI’s models are specifically built, trained, and tested in a grocery retail environment that accounts for the sector’s unique nature and fast pace.

Looking to the future, the presenters suggested that many regular, repeatable retail tasks have the potential to be done autonomously, freeing workers to focus on higher-value strategic work. While AI promises to enhance productivity and decision-making, it also raises questions about the changing nature of retail roles that open up new possibilities to evolve organizational structures.

Generative AI is already providing real value to retailers, noted all of the webinar speakers. However, they also acknowledge that the content shared during the webinar only scratched the surface of what is possible today and the future impact of the quickly evolving technology.

For a deeper understanding of how generative AI can be applied to a specific retailer’s operation, the speakers pointed to SymphonyAI’s new initiative, the Enterprise AI Bootcamp. These bootcamp sessions, typically conducted in-person over a half or full day, cover industry research on AI’s impact on retail in the next one to two years, provide in-depth and hands-on use of advanced solutions. and help companies build a roadmap for AI implementation and optimal impact on their organization.