The Watson Weekly is sponsored by Avalara — the agentic AI platform automating global tax and compliance for leading eCommerce brands with simple plans, transparent pricing, and complete coverage across Shopify, Stripe, WooCommerce, BigCommerce, and more.

I've been in retail and eCommerce long enough to get skeptical when a retailer tells me they're AI-ready. So I asked Abhi Sachdeva, co-founder and CTO of EKYAM, to come on the Watson Weekly and explain what that should actually mean. Abhi spent 18 years leading technology inside retailers including QVC and Tory Burch, so he has been in the rooms where these projects get approved and funded.

I took three things from the conversation.

The first is that a data lake doesn't make you ready. Most retailers I talk to point at the lake as proof the work is done. Abhi's point is that the lake holds rows and nobody has agreed on what the rows mean. If you ask different teams for customer count or gross margin you'll get different numbers, and today that gets resolved by whoever built the dashboard and knows which system to trust. An agent has nobody to ask, so it pulls the first matching row and answers with full confidence. For a merchandising decision I'd take no answer over that one.

Second, the gap is mostly a people gap. Abhi estimates that roughly 5 out of every 100 retailers he works with have product data owners in place, meaning people who sit between the business and the technology team, write the definitions and decide who gets access to what. In his model the operators in merchandising and finance own what the words mean, and technology owns enforcement. I think 5 in 100 is generous. It's rare that I see anyone in a retail company whose job is defining what the data means, and I told him as much on the show.

Third is buy vs. build, which is one of my pet topics. Abhi would buy the control and governance layer, the gateways and monitoring that track who accessed which data and what it cost, because the models underneath change every few months and nobody budgets for keeping up with that. Definitions and workflows are what retailers should own. He gives the big consulting firms credit for pushing master data cleanup and operating model work. His objection is to the 18-month custom agent data layer that only its builders understand and that the retailer has to maintain after the engagement ends.

All three connect to the belief Abhi is betting his company against, which is that data readiness is a project you can scope and close out. Retailers made the same assumption about eCommerce. They built a checkout, considered it finished, and reopened the project when Apple Pay and then Google Pay arrived.

What I don't have a good answer to yet is the budget. If readiness is ongoing, it needs an owner and a line item after the consultants leave, and in most retailers I work with nobody has claimed either. Who has claimed it at your company?