Home Depot Is Rebuilding Its Retail Operation With AI

Home Depot's CIO has a simple rule for AI investments. If it solves a problem the business has already identified, it gets built. 

Angie Brown, a 27-year company veteran who has led technology strategy since May 2025, said she has no intention of limiting how many use cases can use AI, according to Fortune. 

The constraint is not volume. It is whether the investment links to one of three core priorities, merchandising in physical stores, an interconnected retail ecosystem, or growing the professional contractor customer base.

The company's technology leadership was also recently rebuilt. Franziska Bell became CTO in April 2026, bringing experience as Chief Data, AI, and Analytics Officer at Ford Motor. 

Brown, a 27-year Home Depot veteran, has led the CIO function since approximately May 2025. Jordan Broggi, who became EVP of Customer Experience and President of Online in June 2024, oversees the company's $25 billion e-commerce business and is the executive most directly responsible for Magic Apron.

What Each AI System Actually Does

Magic Apron, which debuted in March 2025, answers shopper questions and summarizes product reviews using generative AI trained on Home Depot's own product data. 

The launch immediately revealed a user segmentation problem Broggi described. "The consumers loved it and the pros hated it,” he said to Fortune.

Home Depot learned that the web-based version was asking professional customers questions that were too simplistic for their level of expertise. The pro version was pulled offline and is being fine-tuned with LLMs better suited to that audience before it is relaunched. 

The employee-facing version of Magic Apron is being rolled out to smartphones, with a multilingual upgrade planned as a subsequent development, according to the company.

The customer service AI system built with Google Cloud was tested in 50 stores. During the pilot, voice agents could identify what a customer was calling about in 10 seconds, a specific, verified performance metric that Home Depot cited as evidence of the system's readiness for broader deployment.

The order intelligence system addresses a different problem entirely. The gap between a delivery being scheduled and a delivery actually going smoothly. 

The system looks backward at millions of data points from Home Depot's past deliveries and generates a risk score for each new order, factoring in variables such as whether a property requires a gate code or sits on a narrow access road where a 22-foot delivery truck is more appropriate than a 36-foot truck. 

It proactively contacts customers with potential delivery complications and more accurate arrival windows. Broggi noted that customers do not know or care that generative AI is working in the background. 

"They just want their stuff delivered on time, complete, undamaged, and with clear communication," he said.

Internally, Microsoft Copilot has been made available to office workers, Anthropic's Claude is helping speed software development, and machine learning algorithms are guiding more efficient workflows for store associates.

The GEO Question

Home Depot is also developing a Generative Engine Optimization strategy as consumers increasingly shop through AI platforms including Google Gemini, ChatGPT, and Claude. 

The company allows shoppers to browse its goods directly on ChatGPT and supports Google's Universal Commerce Protocol, which advocates for a common technical language to support agentic commerce.

Broggi was measured about how far along the agentic commerce strategy actually is. "They've got to try to figure out how they want to go to market," he said of the AI companies building the shopping platforms, noting that priorities among those platforms have shifted several times.