NVIDIA Enters the CX Arena With a Blueprint for AI Customer AssistantsNVIDIA Enters the CX Arena With a Blueprint for AI Customer Assistants

Welcome to 2025, which is already giving CX professionals noteworthy tech developments. Week one was highlighted by the Consumer Electronics Show (CES) followed by the National Retail Federation (NRF) holding its annual event in New York starting January 12^th. At NRF’s annual show, retailers can check out technology vendors’ latest CX innovations.

Ahead of the event, NVIDIA -- the chip designer which has become a powerhouse thanks to its ability to provide a complete ecosystem for any company needing massive computing power for AI -- announced a new AI blueprint for building digital assistants that act like sales associates and provide a personalized and interactive shopping experience.

AI-powered digital assistants are touted as the next game changer for retailers looking to improve their customer experience, and better AI-enabled interactions have been the primary focus of all vendors in the CX ecosystem, from web analytics to sales and marketing to contact centers..

The NVIDIA AI Blueprint for retail shopping assistants would enable retailers to build a mobile or web interface where someone could ask a question in natural language, such as, “I bought a new home on the beach, and I’m looking for some coastal furniture.” The AI agent could make recommendations based on past purchase history or other criteria by connecting to sales and inventory data from the retailer.

The blueprint is a multimodal, multi-query retrieval-augmented generation (RAG) pipeline. It leverages NVIDIA’s NIM microservices for large language models (LLMs), NVClip for image-based searches, and NVIDIA NeMo Guardrails for topic control and safe interactions. Its flexible architecture works with other models, including NVIDIA’s Llama Nemotron and custom solutions.

Using the blueprint, built on NVIDIA AI Enterprise and Omniverse platforms, developers can create AI assistants that comprehend text and images, handle many requests, and answer product questions. NIM microservices make the assistants very responsive. They provide a deeper understanding of both written and visual inputs (like text descriptions or pictures), perform complex tasks (like suggesting products that go well together), and offer detailed answers (like whether a product is waterproof or fits a specific need).

In addition to processing text and images, the AI assistants can help consumers search for multiple items at once while handling many users simultaneously without slowing down. With image-based similarity search, consumers can find comparable products to save time. The AI component keeps conversations with consumers focused on the retailer’s catalog, ensuring all interactions are relevant and brand-appropriate.

This versatility is necessary to support various retail applications, from fashion to home furnishings. If a consumer shops for furniture, an AI assistant can use the Omniverse platform to enhance the experience with realistic product visualizations. For example, it can show how a couch fits into their space. The assistant also excels at providing recommendations by combining data from product catalogs, purchase history, and customer search behavior.

“The furniture cost makes it more of a considered purchase for consumers. Decreasing any barriers to purchase is important. Yet the expectation set with e-commerce is that it should be easy and free to return, which is also very costly. So, ensuring the right product expectations are set and met is key to minimizing return rates,” said Cynthia Countouris, director of product marketing for retail, consumer packaged goods, and quick service restaurants at NVIDIA.

Several partners and early adopters, including SoftServe, Dell, and World Wide Technology (WWT), are using the new AI blueprint to develop their digital assistants. SoftServe, for instance, just launched its Gen AI Shopping Assistant, which was built using the blueprint. The YouTube video does an excellent job of highlighting how the solution works. One key feature is a virtual tool that allows consumers to try on clothing in an online chat, making the shopping experience more interactive.

The blueprint for retail is the latest addition to a portfolio of NVIDIA’s Blueprints, several of which were unveiled at CES. Among them is an updated version of the AI blueprint for video search and summarization, designed to help retailers create video analytics AI tools that can process live and archived video content.

Beyond improving customer experiences, NVIDIA is addressing operational challenges in the retail industry. Agentic AI-powered tools can improve supply chain operations in retail, such as assisting warehouse and distribution center staff by diagnosing issues, reducing downtime, and predicting maintenance. Agentic refers to AI that actively performs tasks involving reasoning, planning, and interacting with people, such as digital assistants.

“The shift in technologies to agentic AI is key. This is going to be transformative. It’s taking a step further in helping provide more tools and resources to retail workers to take on new tasks. I see it in the back office and the supply chain side,” said Countouris.

These benefits align with current trends, as NVIDIA’s State of AI in Retail and CPG report highlights. The report reveals that 89 percent of retailers are already using or testing AI technologies, with 87 percent reporting increased revenue and 94 percent experiencing reduced operating costs. AI is also gaining traction for creating personalized marketing content and better customer engagement. In next year’s report, Countouris envisions many new use cases for AI in retail.

AI agents are here to stay and will impact every aspect of customer experience. The Internet democratized access to information, and AI will similarly make expertise available to everyone. Shoppers can start a discussion with a virtual agent and often complete the task without talking to a human. If the discussion becomes more complicated, a human agent can quickly be brought in to augment the virtual agent. Retail will be among the industries that see the most significant transformation because of AI. The NVIDIA blueprint is a turnkey way of deploying the technology and can dramatically reduce the time taken to go to market.

It's important to understand that what NVIDIA announced is a blueprint to enable businesses and ISVs to build agents into their agents. NVIDIA CEO Jensen Huang has said several times that the company is a market maker, not a share taker, meaning it’s not trying to compete with application vendors; in this case, it would be retail-focused CX providers, which includes CRM, contact center, and other vendors.

 Virtual agents allow any vendor in the CX industry to expand its reach. For example, contact center software has historically been used for service and support on the back end of the customer journey. Sales and marketing tools have been used at the front end of the customer acquisition process. Both are important, but having them deployed in silos creates inconsistent experiences for customers. A virtual agent that can span the entire customer journey can bring consistency to customer interactions, but more importantly, remember customer activity and connect the dots between them. Businesses often invest heavily in acquiring customers but under-invest in service, which creates churn. Virtual agents won’t solve all problems but will create better experiences across the customer journey.