Retail and consumer packaged goods companies are no longer debating whether to invest in AI, but how effectively to deploy and scale it. According to the 2025 Deloitte Retail AI Report, 77% of global retailers believe autonomous, decision-making AI will be the single biggest differentiator in retail performance over the next five years.
Walmart’s AI transformation stands out for its strategy anchored in a four-agent architecture to integrate retail operations, technology platforms, and global capability centres (GCCs) into a single system.
Over the past 15+ years, Walmart’s India journey has evolved from early-stage support functions to omni-channel integration, and now to enterprise-level AI and platform unification.
In an exclusive interaction with AIM, Balu Chaturvedula, SVP and country head, Walmart Global Tech, says, “We used to build systems in silos. Then we talked about building once and leveraging multiple times. Now we are talking about building for the Walmart enterprise once.”
Today, Walmart’s India centre no longer operates as a satellite or offshore hub. It functions as a fully unified extension of Walmart’s global technology organisation, playing a central role in shaping enterprise-wide AI platforms, agentic systems, and next-generation retail innovation.
Four Agents
Walmart’s AI strategy is not built around hype cycles, but focused on retail fundamentals. Chaturvedula explains that the entire AI and innovation stack still hinges on four non-negotiable pillars: assortment, price, speed, and trust.
“None of them are negotiable. Each dimension is very important, ” he adds.
AI investments are mapped directly to these pillars—whether it’s catalogue intelligence, predictive pricing, ultra-fast fulfilment, or trust-driven product accuracy, delivery reliability, and returns experience.
Then there are AI agents which Walmart has structured around four core personas, each represented by a dedicated “super-agent”—customer/member agent, associate agent, seller–supplier–advertiser agent, and developer agent. “For us, building everything revolves around these four personas—how it’s going to make them more productive,” Chaturvedula remarks.
These agents are not chatbots; rather, they are autonomous, system-level AI layers embedded into enterprise workflows.
Furthermore, in 2025, Walmart partnered with OpenAI to allow customers and members to shop on Walmart through ChatGPT using instant checkout.
Sparky And the Multimodal Future of Retail
Walmart’s customer-facing AI agent, Sparky, represents the shift from search-based shopping to decision-based shopping.
“How do you make the decision on behalf of the customer and actually complete the purchase—and if required, listen to a signal and also complete the return—that is the future of shopping,” he mentions.
Sparky is designed as a multimodal agent, not limited to text. It generates 30-second audio summaries of product reviews, allowing customers to listen instead of reading—an example of agentic automation already operating at scale. Voice, image-based recognition, and advanced inputs are next in the roadmap.
Much of Walmart’s AI infrastructure, including its chatbot and agent platforms, has been built in India.
“The original chatbot engine was built in India by a group of engineers—and that’s the same engine powering multiple agents today,” Chaturvedula adds.
Walmart’s India operations—primarily in Bengaluru and Chennai—are deeply embedded in enterprise AI systems, data platforms, and agent orchestration layers. “You cannot touch the life of a retailer today without India having a significant presence in that system,” he states.
Moreover, one of the most distinctive elements of Walmart’s strategy is treating developers themselves as an “AI persona”. Chaturvedula believes that rather than replacing developers, AI is reshaping their role.
“Human force can’t create that amount of code—but humans can control what is being produced,” he adds
Validation, governance, prompt design, and quality control are becoming core engineering functions.
Data: The Hidden Backbone of Agentic AI
Walmart began large-scale data unification years before AI became mainstream. “We started this journey four years ago. Data is under one single organisation—schema, lineage, metadata, serving, correlation, enriching—all aspects of it,” he explains.
This unified data foundation enables agentic AI systems to operate at enterprise scale, with governance and trust embedded into the architecture.
Meanwhile, it continues to invest in engineering talent in India, hiring 300+ graduates annually from IITs, IIITs, and tier-two institutions.
As agentic AI matures, Walmart’s strategy signals the future of retail AI. Not tools, not pilots, not chatbots—but enterprise AI systems, deeply integrated into data, operations, trust, and customer experience.
“We’re not building future products—we’re evolving agents into what they will become. That transformation will be phenomenally different,” he reveals.
Your reaction