AI is ready to transform category management. The real question is your data.

VST founder and CEO Nick Theodore joined an IGD panel on 23 June 2026 to discuss how AI is reshaping category management - what has changed, what is holding teams back, and where the value now lies. 

‍

"100% yes." That was Nick Theodore's verdict at the IGD panel: AI is ready for the real work of category management - not in a pilot, not someday, but now.

‍

VST, the retail-data AI platform, works in planogram and transactional data and in testing merchandising concepts - a view from both sides of the shelf. A couple of months ago its data and infrastructure met the latest AI across billions of rows of in-store, transactional and customer data; sophisticated category strategies emerged in moments. "Suddenly,"* Nick says, "we were the best category managers in the world."

‍

The panel's question was a plain one - how, week to week, is technology actually changing category management for the people who do it? Plain to ask, less so to answer. The tools have advanced more in the past year than in the previous ten, and most teams are still catching up with what they already have.

‍

The coming wave

‍

None of this came from a single breakthrough. In barely eighteen months, three things converged. An open standard, MCP, gave AI models a common way to plug into the systems where data actually lives. A layer of "skills" equipped them with packaged, on-demand expertise. And the base models took a clear jump in capability (Opus 4.5 in November '25), crossing the line from impressive to dependable. Each advance compounded the last.

‍

The result, for the first time, simply works - against real data, on real category problems. "That,"* said Nick, "is the world we are working toward - shared data, powered by AI." The cost keeps falling, and the advantage will accrue to whoever moves first.

‍

The spreadsheet's days are numbered

‍

What makes this real rather than a demo is conversational analytics - putting questions to your data in plain language and getting answers that trace back to source. At VST it runs across the team - category visions, store performance, new-product opportunities, space and flow.

‍

The deeper change is in the daily work. "We are close to the point where no one in a category team needs to live in a spreadsheet again," Nick says - "frightening or exciting, depending where you sit." Freed from data-wrangling, teams recover their hours for the commercial, creative work that moves a category.

‍

One condition governs all of it - the data underneath. Organised well, the system works today; organised badly, it does not work at all. "The model,"

as Nick puts it, "is only ever as good as the foundation it sits on."

‍

Edison thought he'd built a dictation machine

‍

Thomas Edison was certain the phonograph he invented was a dictation machine. He ranked music near the bottom of its uses and dismissed those who played songs on it - only to concede, twenty years on, that amusement had been the point. Inventors rarely work out what their inventions are for.

‍

That is the opportunity here. The labs built the capability; what it is best used for, in the particulars of a real category, is still unsettled - and it will be settled by the people doing the work, not the researchers. Hence Nick's advice - talk to your technology partners, early. The best ideas surface between those who know the data and those who know the category.

‍

Where adoption stalls

‍

Three things separate the teams capturing value from those that stall. The first is security: Nick is in far more conversations with IT than a year ago - a healthy sign - and the fix is to pair engineers with the client's own staff so less technical colleagues can work safely. The second is cost: pricing is still settling as the model shifts back towards licences, though it should fall as providers compete; choose tools on the data and the results, not the logo on the box. The third is experience: "how optimistic someone is about AI is proportional to how much they've used it." Hands-on beats theory - the people who got their hands on Edison's phonograph understood it better than its inventor ever did.

‍

The frontier

‍

Ask Nick what the category leader of 2030 looks like and the answer is reassuring - much like today's - commercially curious, still experimenting, still human - but supported by a team of agents "that outperform any one of us working alone."

‍

What sets them apart will be where they dig. Two frontiers stand out. The first runs beneath everything else - data infrastructure - the foundation the rest depends on, and the clearest line between the teams pulling ahead and those left behind. It is why VST built its platform, and the place to begin. The second is in-store retail media, where vast sums flow in while measurement and inventory stay primitive. Close that gap and you set the standard others measure against. "That," says Nick, "is where I'd place my bet."

‍

The tools are ready and the ground is open. The next chapter of category management will be written by whoever starts now.

‍