The Knowledge Base
The AI shifts that genuinely change how stores get built and run — separated from the ones still living in a demo. Each entry links to the original source and carries a short read on what it means for an operator.
How this page works. A hand-picked reading list, not an automated feed. Each entry links straight to the original publisher — primary sourcing includes The Verge AI alongside vendor and analyst research. Every item carries a short read on what it actually means for an operator. Last reviewed September 2026.
Morgan Stanley projects that nearly half of online shoppers will use AI shopping agents by 2030. Agentic commerce — where an AI completes the purchase autonomously — has moved from conference slide to roadmap item, and it changes who your storefront is actually talking to.
Research puts AI-assisted shoppers at 12.3% conversion against 3.1% without, completing purchases 47% faster, with returning customers spending around 25% more when assisted.
Visual search, voice commerce and LLM-powered recommendations are dissolving the boundary between digital and physical retail, and reshaping what on-site search is expected to do.
Over half of US consumers say they have used ChatGPT or Gemini to browse or buy online, and the large majority of ecommerce brands deploying AI report a strong return.
Adobe's analysis shows AI-driven referrals to retail sites growing by orders of magnitude, changing how brands need to think about discoverability and attribution.
AI-driven demand forecasting is credited with letting IKEA deliver goods at prices well below competitors. Custom forecasting is now within reach for mid-sized brands.
Vendor guidance converges on a short list: merchandising, service deflection, content production at catalogue scale, and forecasting — rather than the customer-facing novelty features.
Apply it
Three moves that hold up regardless of which vendor wins. In this order.
01
Complete structured data, accurate specs, real stock and price signals. Agents and answer engines can only recommend what they can parse — this is the cheapest work with the longest shelf life.
02
Pick a high-consideration category, set a single success metric before you start, and give it a fixed window. A pilot without a kill criterion becomes a permanent cost.
03
Tag AI-referred sessions and assisted journeys separately in GA4. Without a clean baseline you cannot tell whether the tool worked or the season did.
Next step
I'll look at your catalogue, your data layer and your margins, then tell you which of these is worth your budget this quarter — and which isn't.