How Retailers Are Personalizing the Shopping Experience with AI

Ahead of this year's NRF Big Show, the retail industry is looking for ways to apply LLM and Generative AI across customer experience and operations. AI21 believes that chatbots handling vast sales, inventory, and purchasing data in natural language can create personalized recommendations and messages, and reduce inventory loss through faster decision-making.

While existing chatbots were limited to standardized answers, LLM chatbots better understand context and expression, allowing for more flexible responses. Use cases include the following.

• Customer Support: Handles questions about returns, pricing, inventory, and sizing in natural language.
• Private Shopping Assistant: Provides personalized recommendations based on purchase history and preferences.
• Back Office: Generates product descriptions in bulk and varies copy to match buyer personas.

For companies that find it difficult to train general-purpose models from scratch, AI21 proposes Task Specific Models (TSMs). The Semantic Search TSM searches based on user intent rather than the exact words entered, and the Contextual Answers TSM answers based only on trustworthy data that the company has input. AI21 explains that this structure—focused on a single task, with built-in customization and verification mechanisms—can achieve faster deployment, lower latency, and lower error rates. Ultimately, AI21 presents TSMs as a practical solution for the era of conversational commerce, claiming that a single developer can start personalized commerce and self-service with just one line of code.