Apple co-founder Steve Jobs once said, “People don’t know what they want until you show it to them. Our job is to figure out what they want before they do.”
The world of retail is getting ready for its iPhone moment using breakthrough generative AI technology. Unlike traditional AI that is used to detect patterns and make decisions, generative AI models use neural networks that are trained with massive amounts of data to create new and original content — all from just a few typed questions from a person.
Companies are racing to adopt it. Walmart says it is exploring generative AI to deliver problem-solving customer chatbots; Levi Strauss plans to supplement human models with computer-generated ones to reflect a broader diversity of body types, and Shopify is helping its network of merchants generate product descriptions by entering a few keywords to target in search results and select a tone to the content, like “expert” or “sophisticated.” Startups including Fashable and Verneek are generating unique use cases of generative AI for retail.
These first movers near universally are noting the gigantic opportunity to transform their business, from operations to customer relations and retention. And there’s plenty of money at stake. Global ecommerce sales, representing 22% of all retail sales, could increase to $5.4 trillion in 2026, from $3.3 trillion today, according to Morgan Stanley.
Yet the road to generative AI efficiency must be paved with a strong AI foundation. The AI models powering generative AI are called large language models (LLMs), and they work by learning from vast amounts of data from multiple sources to produce new content.
New state-of-the-art generative AI models for text, images, high-fidelity videos and 3D assets can be trained and fine-tuned with retailers’ proprietary data, representing their specific brand and tone and with appropriate guardrails to do domain-specific tasks. For example, a retailer looking to use the technology for generating descriptions in ecommerce product catalogs would train the model on relevant product information and metatags specific to its brand and items they sell.
Here are the four areas most likely to bear fruit quickly for retailers:
1. Hyper-personalized shopping.
Retailers have a vast amount of data about customer shopping behavior and purchase history. Generative AI can dramatically improve customer experiences and drive revenue by helping shoppers find the products that most match attributes of what they’re looking for today, not just what they have purchased in the past.
Generative AI can analyze customer data, perform social media sentiment analysis and create custom journeys to offer personalized product recommendations like a stylist does and even discounts on items the AI believes a customer wants most in order to drive revenue.
For example, Verneek, a New York City-based software startup, has a generative AI application called Quin Shopping AI. The company says Quin helps shoppers make better and faster decisions throughout their online or in-store journeys — fulfilling any sophisticated shopping request — ranging from finding available product assortments or recipes that match a whole suite of health- or budget-related constraints — to general health and customer service. It leverages hyper-localized search combined with millions of product attributes to recommend products.
2. Automated catalog and product descriptions.
SKUs are the lifeblood of any retail operation, helping keep track of the thousands — even hundreds of thousands — of items in inventory. That task isn’t getting any easier, thanks to brand extensions, new colors, new flavors, new packaging, promotions, lite products, gluten-free everything, digital device releases and more. The problem becomes especially acute online, where each new SKU requires an accurate and detailed product description for shoppers.
Retailers are discovering the power of generative AI for streamlining the process of onboarding new products for ecommerce product catalogs, using AI models trained on a wealth of specific product data. Generative AI can automatically create these product catalogs, helping to give structure to unstructured data.
This produces richer, more robust product descriptions that improve SEO ranking and offer shoppers up-to-date product information, with high-resolution images that drive revenue. That helps dramatically increase productivity while driving improved sales. To backstop the generative AI descriptions, retailers can have human ecommerce or product experts review and approve them.
Product data can also be used to train AI models that generate new designs for products based on data such as customer feedback, sales data and market trends. For example, Portugal-based startup Fashable is using generative AI to let fashion houses create virtual clothing designs without the need for physical fabric, eliminating the environmental impact of the clothing design process. And because the models are trained on retailer data, the AI ensures the new designs generated are more appealing to customers and better suited to market demand.
3. Automatic price optimization.
Inflation and concerns about whether economies around the world are sliding into recession have led to increased price sensitivity among shoppers. Consumers today are continuously looking for the best prices before making a purchase decision. Dynamic pricing, the process by which prices are automatically adjusted based on market conditions or customer behavior, is key to helping retailers improve razor-thin margins.
Generative AI can search competitor prices several times a day, analyze demand patterns and market trends, analyze operational costs and provide pricing recommendations in real time to avoid losing shoppers to competitors or adjust pricing to maximize profits.
The result is improved profit margins, as retailers can better attract and retain customers with a more fluent pricing strategy. As the factors affecting the market change so quickly and so often, dynamic pricing is a job well-suited to AI and retailers are seeing success with using it to monitor and optimize pricing.
4. Assistive customer service.
Improving customer service has long been a priority for retailers large and small. Service interactions often define a customer’s shopping journey. Surveys have shown shoppers increasingly are seeking quick, accurate and personalized service — and are quick to switch to a competitor if they grow dissatisfied when their needs aren’t met.
Generative AI can assist in this avenue, too. Once trained on a retailer’s data, virtual customer assistants can create AI-generated responses and automatically translate content in multiple languages.
This AI assistance also can make traditional agents more productive by automatically generating customer service chat or email responses, thus reducing or eliminating the amount of time the human agents spend drafting an answer. The tech can check resource planning systems when drafting these answers, helping determine when orders will ship and providing shipping status or support customers in changing orders. The agents can instead focus on dealing with more complicated customer issues or driving customer retention offers.
Another use case for AI models is to determine customer sentiment. AI can analyze customer behavior on social media to determine the importance of the situation and notify customer service agents of the most urgent or serious cases. It can also be used to analyze customer reviews and provide succinct review summaries to customers researching certain products.
Generative AI is bringing a competitive advantage to those retailers that are using it, resulting in higher revenue, greater customer loyalty and higher margins. No retailer can afford to ignore this revolutionary wave of productivity.
Learn more about generative AI for enterprises and stay up to date on new generative AI breakthroughs, developments and technologies at NVIDIA.
As VP of Retail, Consumer Product Goods and Quick Service Restaurants at NVIDIA, Azita Martin leads the Retail team. She is responsible for global go-to-market strategy in the vertical. She advises senior executives on best practices of leveraging AI to improve operational efficiencies and identify new revenue streams. Prior to NVIDIA, Martin was the Chief Marketing Officer at Maana, an AI software platform used by industrial companies to accelerate building AI applications. Martin has been working with the Chief Digital Officers and heads of analytics/AI of Fortune 500 industrial companies and is intimately familiar with the top AI use cases that are demonstrating business value.