Sathish:
“Retail has adapted many initiatives over the years. Why is agentic AI different, and what kind of impact will it have?”
Brandon:
“I have always believed retail is a blend of the arts and sciences. If you veer too far into the sciences and lose the art, it is a death sentence for the business and the customer experience. Traditional AI is more about being predictive. It provides recommendations, but it requires human intervention to execute.
Agentic AI is about business outcomes. It continuously monitors signals, looks for triggers and opportunities, makes contextual decisions autonomously, and coordinates across systems and functions in ways that were not possible a couple of years ago. But you have to set the guardrails, set the expectations, and ensure it is continuously improving. It is still in the very early stages. Agentic AI is an enabler. It is a tool. It is there to help drive better decision-making with humans in the loop, navigating and driving the decisions ultimately.”
Sathish:
“What is the biggest pain point agentic AI can address?”
Brandon:
“In my entire career as a merchant planner, the biggest challenge was sourcing one version of the truth. Having integrated data sources that can actually drive better merchandising decisions across the entire retail value chain. What I see now is agentic AI enabling seamless integration between all systems, from sourcing and product development to merchandising, assortment planning, inventory management, replenishment, and supply chain fulfillment down to the store level. It has to be driven toward meeting the customer’s needs and empowering frontline workers to serve customers seamlessly across channels, whether it is social commerce, e-commerce, or in-store. There has to be one consistent experience.”
Sathish:
“Should retailers do a complete transformation to adopt agentic AI or take a different approach?”
Brandon:
“The days of large-scale ERP transformations are over. Spending two or three years changing your backend systems, your OMS, merchandising, point of sale, supply chain. Retailers do not have the luxury of doing that anymore. Agentic AI autonomous agents can sit on top of existing integrations and drive faster decision-making across your systems. But the challenge is whether your tech stack is integrated or it is a spaghetti tech debt situation where you are not sure where your data sources are. I always recommend taking a step back. Look at your data strategy and data governance. Is your data clean? Is it integrated? Is it one version of the truth? The old saying goes, garbage in, garbage out. If you are not feeding the LLMs and agentic systems with the right data, you should pause and get your data strategy in place first. Once you achieve that, then identify business outcomes through a cross-functional executive workshop and prioritize where the biggest pain points are.”
Sathish:
“What if a retailer is still using on-prem databases and the data is not in the cloud?”
Brandon:
“That is a challenge. You need data stewards and change champions. If you do not have one source of truth that you have trust and confidence in, it is very difficult to think about agentic AI. You can start with a limited subset of data and pilot it in a lab or a few stores. But you are introducing significant risk if you do an enterprise rollout with data that is not accurate, not timely, not integrated, and not governed. I suffered for a decade in retail, trying to pull all kinds of data together from Excel, from downloads and uploads and VLOOKUPs. Here we are in 2026. There are still many organizations that operate in Excel.”
Sathish:
“Where has agentic AI already made a real impact in retail?”
Brandon:
“Louis Vuitton and Dior are empowering store associates with digital platforms and real-time autonomous styling. They have reduced overstock by 20% through AI-driven demand sensing. Conversion rates are up to 15% better using clienteling tools. Louis Vuitton is using AI assistants that now handle 60 to 70% of routine customer inquiries, while human advisors handle the more complex challenges. Burberry is leveraging agentic AI and digital twin capabilities to authenticate products and remove $100 million of counterfeit listings in a single year, leading to a 10% lift in sales and much greater trust in the brand. 80 to 85% of sales still happen in physical stores, but when customers can see online with 90% accuracy that the inventory exists, there is a lot more confidence to visit the store on a Saturday.”
Sathish:
“Where are the risks in implementing agentic AI?”
Brandon:
“The quality of data is probably the biggest risk. The second is hallucinations, being led down a path you did not expect. You have to check the quality of the results. Then there is cybersecurity. Customer data, credit card information, and personal information. We have seen hacks and hijacks. Retailers need security and resiliency with guardrails that give customers confidence that their information is protected. There is also the bright and shiny object pressure. Everyone sees things at NRF and ShopTalk and feels pressure to jump on agentic AI, but there are so many foundational things that need to be achieved before you can do that.”
Sathish:
“If a mid-sized company is looking to implement agentic AI, where do they start?”
Brandon:
“Start with core principles and what you represent in the market. Do not take a big bang approach. Take a purpose-led approach. What are the business outcomes you want to achieve? What KPIs do you want to drive: revenue growth, EBITDA improvement, cost savings, better customer experiences? Then identify a pilot. You need board approval in many cases, and unless you have a cost savings play or an ROI case in a measurable timeframe, these AI conversations go to a halt. Set expectations: if we optimize the supply chain with this agent, we should expect X results in one year, two years, five years. The execution strategy has to account for company culture, key sponsors, change agents, how it impacts roles, what training and organizational change management you need.”
Sathish:
“Where have you seen agentic AI implementations fail?”
Brandon:
“I have seen cases where AI has been embedded without the right sponsorship from the business teams that own the processes. They did not do the due diligence to ensure alignment from the start. The teams did not own it and never embraced it. When a director or VP is engaged, aligned, and sees the value, it extends throughout the organization. Any transformation will fail unless you have the sponsorship, co-creation, and collaboration needed to drive the KPIs. I have seen it fail time and time again, even before agentic AI became a thing. As humans, we resist change unless we fully understand the why, the purpose, the how, and the impact on our role and our team. If those questions are not answered, it fails.”
Sathish:
“What metrics should companies look at for agentic AI?”
Brandon:
“If you look at revenue only, you are shortsighted because it does not tie back to profitability. EBITDA and gross margin are critical. From an inventory perspective, look at conversion rates, in-stock rate, and inventory turns. Look at pricing and promotional strategies and the profitability of price changes. You can see up to 10% improvement in some categories, maybe 5% in others, depending on the retailer and the segment. But you need to be realistic. It has to be a collaborative co-creation with the executives to build those models together and align with the company’s North Star goals.”
