Retail CIO Connect 2026 wrapped last week at Ashridge House, and it confirmed something we have been seeing across every retail engagement this year: the gap between AI ambition and AI in production is almost never about the model.
GSPANN was on-site both days. We held a number of meetings with senior retail IT leaders about the roadblocks slowing their AI and modernization plans. Here is what stood out.
The Conversations Pointed to Some Real Challenges
Retail CIO Connect is invitation-only. The agenda is closed-door roundtables and pre-scheduled 1:1s with people who control the budget and own the roadmap.
What we heard across those conversations was consistent. Most retailers in the room already have AI pilots running. Very few have agents in production. We see the same pattern across the industry: a model gets built, a demo lands, leadership signs off, and the initiative stalls the moment it meets the organization's actual data.
Our Think Tank Session: “The Real Bottleneck Isn't AI. It's Data.”
Praveen Chandra, GSPANN's Business Head for Data, Analytics, Digital Marketing, ERP, and Data Science, led a closed-door session built around where AI initiatives get stuck.
The discussion centered on three problems that come up in nearly every retail data conversation we have:
Ownership. Data quality work rarely has a clear owner. It sits between IT, data engineering, and whichever business function feels the pain first. As a result, nobody has the authority to prioritize fixing it over the next AI initiative.
Access. Data that should inform a model is often locked inside a legacy system nobody wants to touch. Even when teams can reach it, the data is cut off from the business context that would make it useful.
Sequencing. Most AI roadmaps are built model-first, with data readiness treated as an implementation detail. By the time the gap surfaces, the project has already been staffed and scheduled around the wrong assumption.
None of this is a surprising diagnosis on its own. What made the session useful was the room agreeing on it openly and then working through what it takes to reverse the sequence: funding the data foundation as its own workstream with a dedicated budget, outside the AI program.
Where This Connects to What We Build
We went to Retail CIO Connect because we have been solving this problem in production, and the conversations in the room matched exactly what we see on client engagements.
Our data quality accelerator, BEAT, exists for this reason. BEAT automates data testing and continuous auditing across your pipelines, so quality problems surface before they reach production. We built it after watching the same stall point repeat across retail clients: nobody trusted the data enough to let the model act on it.
That is also why our application modernization work treats legacy re-platforming and AI-readiness as one project. You cannot build a durable AI roadmap on top of a data estate that nobody has made a decision about.
What's Next
If your team was at Retail CIO Connect and the think tank's themes sound familiar, or if you weren't there and they still do, we would like to continue the conversation.
Our experts are available to talk through where your data foundation stands today and what moving it forward would take. Reach them at marketing.team@gspann.com.
Our Experts
Ashish Ranjan
Vice President and Head - EMEA
