Elastic Path is the API-first, AI-ready commerce platform for brands with complex requirements: large catalogs, multi-tier pricing, and multiple brands and markets. Manufacturers, distributors, and retailers run account hierarchies, quoting, and negotiated pricing natively, not through bolt-on apps and workarounds.

We held the first Watson Live! The Retail AI Debates at RetailClub Presented by KIBO Commerce. Each debater spoke twice, the audience asked questions, and the room voted by holding up red or blue plates. I learned as much from the times we asked for a show of hands as I did from the speeches.

The first motion asked whether retailers should change their workflows around AI before their data is ready. Gireesh Sahukar argued for it. He asked the retailers and brands in the room to raise their hands and got about ten, and when he asked which of them had good data, only one hand stayed up.

Kelly Goetsch took the other side and said he actually believes it. He cited a post by Target’s chief architect describing a period when the company had seven systems of record for inventory, and he described finance teams arguing over whether last_updated or last_updated_1 is the real column. His view is that you can put the best model in the world on top of that and it still won’t know which number to trust.

Gireesh answered that a company’s data reflects its business model, so improving the business improves the data over time. He spent five years at one retailer reducing a mainframe until it was small enough to retire, and he doesn’t think anyone gets to skip that kind of work. When Kelly claimed Domino’s had just been sold to private equity, Kevin Reese corrected him from the floor that it was Pizza Hut. Danny Smith from Stripe added that if nobody has clean data, the retailer that starts anyway is ahead, and Gireesh won the vote by a wide margin.

I agree with the room on that one, although I suspect most people there underestimate how long the data work takes once it starts.

Casey Golden and Kevin Reese debated whether companies will regret AI layoffs. Casey said retail has never documented how most jobs get done and runs on people’s own spreadsheets, so layoffs mean buying expertise back later at a higher price. She mentioned a rehiring statistic that I’d want to see sourced before I used it, and she pointed to Dollar General scaling back self-checkout as a staffing cut that didn’t hold up even without AI involved. Kevin agreed that layoffs are brutal but said the P&L will push companies to restructure, and that the regret will come from handling them badly. Most people in that room have been on one side of a layoff or the other, and they voted for Kevin.

I argued the final motion against Danny. I said retailers should build their own agents first, because a useful agent needs inventory, delivery dates and service history, and handing that to OpenAI gives away the thing that makes a retailer worth visiting. Danny asked who in the room had researched a purchase in ChatGPT or Gemini, and most hands went up. When he asked about Rufus he counted four. He compared a branded chatbot nobody uses to a nice store on a street with no traffic. I won the vote, although Danny reminded everyone that the host tends to.

The first and last debates sit awkwardly together for me. An hour after the room agreed that almost nobody’s data is ready, I made the case for agents that depend on it, and I’m still not sure how a mid-market retailer is supposed to do both at once.