Yesterday at CES 2026, I had the privilege of moderating the Fireside Chat with Daniel Danker, EVP of AI Acceleration, Product, and Design at Walmart. Daniel joined Walmart around 5.5 months ago and previously served as Chief Product Officer at Instacart.
The themes of the session were the use cases, scale of opportunities, product decisions, and challenges for AI at Walmart across consumer, associate, operations, and other applications. Below, I highlight all of the questions that I prepared as moderator and a synopsis of Daniel’s comments as I recalled them after.
Consumer AI
What does a Walmart store look like in 10 years? What is the interplay between shoppers, associates, and AI?
Daniel discussed that the store of the future will be as much digital as it is physical today, i.e., the lines will blur between the digital and the physical experiences.
As a tangible example, Daniel described how in-store inventory might be customized to the shoppers for each store — what their product preferences are and when they might shop next at the store.
What does the Walmart app look like? Is it a chatbot-first interface?
Daniel discussed how the app today has distinct experiences — one based on scrolling and another based on chatting. He also spoke about how those experiences will become seamless in the future.
Third-Party Chatbots
In the OpenAI / Walmart partnership, who owns the customer journey?
Here, Daniel did not express concern on a possible loss of customer ownership. He discussed hybrid journeys where discovery might be by one party and transaction might be another party.
He also discussed how commerce journeys may begin often with non-commerce queries.
Will Walmart share its user data with OpenAI to improve OpenAI’s responses?
Daniel said that Walmart will not share its user data, particularly without user consent.
Why exclude fresh foods? Aren’t recipes a main use case for chatbots?
Daniel said that the fresh foods exclusion was simply timing. He didn’t rule it out for the future.
First-Party Chatbots
With Sparky, what metrics can you share? What are some early learnings on initial user friction and successful conversion paths?
As expected, Daniel kindly declined to share specific metrics about Walmart's AI-assisted shopping assistant. He mentioned significant growth, off of a small base.
Sparky, launched last year, is significantly behind Amazon’s AI-assisted shopping assistant Rufus, which launched two years ago. Rufus had 250 million shoppers last year.
Agentic Commerce
How large does agentic commerce become, and in what timeframe? In what year does agentic exceed search for Walmart customer journeys?
Daniel said that he thought the two would merge together and not be distinct journeys.
In agentic commerce, who decides what to show to consumers and how? What are the implications? Are small businesses at the greatest risk?
Here, Daniel said that he thought Walmart is well positioned to show up as it provides value to the consumer. He did not express a concern about small businesses being crowded out.
How does pricing work when bots are making decisions on behalf of consumers? Will AI drive lower prices for Walmart consumers?
Here, Daniel emphasized that the importance of trust. As long as consumers knew they were getting the a great price, he said, they could feel good about the interaction.
Associate and Enterprise AI
For associate agents, what is working best thus far? What metrics can you share?
Daniel described how associate agents are guiding associates to out-of-stock or floor spill situations autonomously in real time.
Walmart has announced a headcount freeze at 2.1 million for three years. Let’s look 5 years out. What is the likely net headcount impact of AI?
This is one of the contentious topics in AI. Daniel said that they spend significantly more time thinking about the growth opportunities from AI rather than the cost opportunities.
Lightning Round
How many total agents and nano agents does Walmart have today? In one year?
Daniel said too many to count.
True/False. Agentic will supercharge Walmart’s eCommerce growth.
This one came from an analyst question in Walmart’s last earnings call. Daniel said, Absolutely.
True/False. Claude-run vending machines will find their way into Walmart offices at some point.
Apparently, not too many in the audience had read about the WSJ Claude vending machine experiment late last year. Here’s the link!
True/False. Foundation models are getting commoditized, and OpenAI is right to call a Code Red.
Daniel said it was early to say how the foundation model companies evolve.
—
I left the session impressed with Daniel’s thoughtful answers and humorous undertone. It is clear that Walmart is investing heavily in AI and that the vision and the opportunities are tremendous. Retail and AI are just at the beginning.
Thank you to Daniel and Mischa @ Walmart and to Jackie and Marika @ CES for a great session! I would also like to thank and recognize Cheryl Ainoa, Somesh Bindu, Casey Carl, and Fiona Tan, among other colleagues, for their insights in my moderator prep. I greatly appreciate it!
.avif)
