Sixty-eight percent of UK and Irish shoppers have already used an AI agent to help them shop. Another 18% are open to trying one. Between them, that's 86% of your customer base who may soon let ChatGPT, Perplexity, Google AI Mode, or Amazon's Rufus compare brands, check prices, and decide what lands in the basket without ever browsing your website (CI&T research via Retail Technology Innovation Hub).
This is agentic commerce, and for UK mid-market CPG and FMCG brands it's no longer a future-gazing trend piece. It's a distribution channel you don't control yet, run by algorithms that either can or can't find your products — and most mid-market brands have no idea which.
What "agentic commerce" actually means
Agentic commerce is AI agents autonomously researching, comparing, and increasingly completing purchases on a shopper's behalf. McKinsey estimates AI agents could mediate $3–5 trillion of global consumer commerce by 2030 (McKinsey QuantumBlack). Deloitte's CPG practice has coined the sharper term for what this means at brand level: the algorithmic shelf (Deloitte, "Agentic Commerce in CPG: Winning the Algorithmic Shelf").
The algorithmic shelf works differently to a supermarket aisle or a search results page. When a human shopper overlooks your product, you get warning signs falling click-through, weak conversion, a bad review. When an AI agent overlooks your product because it can't parse or verify your data, you get nothing. No warning, no signal in your usual KPIs just silent exclusion from the recommendation set. Deloitte reports that 55% of consumers now start their shopping journey inside a large language model, not a search engine or retailer site. If your product data can't be read and trusted by that model, you're not losing a sale you're not even in the running.
CI&T's UK data shows where this bites hardest right now: 45% of AI-agent shoppers use them to compare brands, 42% to find the lowest price, and 39% to locate where a specific item is available and groceries are among the fastest-growing categories for this behaviour. That's squarely inside FMCG and CPG territory.
This connects directly to the adoption stats we've covered before see AI in Supply Chain Is No Longer Optional except this time the AI adoption curve isn't inside your operation. It's inside your customer's shopping journey, and you don't get a vote on whether it happens.
What "agent-ready" means in practice
"Agent-ready" isn't a marketing slogan it's a data and infrastructure standard. Deloitte's framework for CPG brands boils down to five moves:
Machine-readable product data
structured, consistent data across every SKU, not PDFs and inconsistent spreadsheets, so agents can interpret and compare products reliably.
Targeted pilots
testing agentic use cases like replenishment ordering, subscriptions, and B2B ordering rather than trying to "solve" agentic commerce all at once.
Ecosystem partnerships
working with the platforms (hyperscalers, AI-native retail tools) actually mediating these purchases.
Modernised commerce infrastructure
real-time APIs for pricing, inventory, and availability, because agents discount data that's stale or inconsistent with what's actually in stock.
Ongoing monitoring
tracking how agents are actually representing your brand, since this is a moving target, not a one-off fix.\
The technical backbone for point one already exists and most UK brands are only half using it. GS1's Web Vocabulary and structured data standards the same standards behind supermarket barcodes are what let both traditional search engines and AI agents parse product attributes consistently. GS1 UK's own case studies show measurable gains in search visibility simply from cleaning this up (GS1 UK: Boost search visibility and data accuracy with GS1 Web Vocabulary). This is the same "fix the foundations before you buy new tools" principle we've written about in the context of internal AI projects see Why Data Engineering Is the Real AI Bottleneck — it turns out your outward-facing product data has the identical problem, just with revenue attached to it directly.
Where UK mid-market brands are exposed
Three gaps show up repeatedly when we look at mid-market CPG and FMCG brands against this checklist:
- Retail media and marketplace listings built for human eyeballs — hero images, persuasive copy with the underlying structured data (allergens, dietary attributes, pack sizes, certifications) incomplete or inconsistent across retailers.
- DTC and brand websites that render product detail through JavaScript-heavy front ends agents can't reliably crawl, with no structured data layer underneath the visual design.
- Pricing and availability feeds that update daily or weekly instead of in real time exactly the staleness Deloitte flags as a trust-killer for agents making comparison decisions.
None of these are visible in a normal marketing dashboard. They only show up when you actually query the agents yourself and see whether your brand appears and how accurately.
A 90-day agent-readiness sprint
This is a scoped, sequenced problem, not a platform purchase consistent with how we've approached every other AI initiative on this blog, from 90-Day AI Pilots in Food & Beverage to why deployments that try to do everything at once take 3–6 months instead of 30 days. The same discipline applies here:
- Weeks 1–2 - Audit: Query ChatGPT, Perplexity, Google AI Mode, and Amazon Rufus with the comparison questions real shoppers ask about your category. Document what's accurate, what's missing, and what's simply wrong.
- Weeks 3–6 - Fix the data foundation: Standardise product data to GS1/schema.org structure across your top revenue-driving SKUs first not the whole catalogue. This mirrors the scoping approach in When to Scope AI Data Work Narrow vs Build for Reuse.
- Weeks 7–10 - Connect real-time feeds: Pricing, stock, and availability data flowing to retailer and marketplace platforms without the lag that erodes agent trust.
- Weeks 11–13 - Re-test and monitor: Re-run the audit queries, measure what changed, and put a lightweight ongoing monitoring process in place this doesn't stay fixed on its own as agents and platforms keep evolving.
Framing it for the board
Don't pitch this internally as "AI strategy." Pitch it the way we've argued every AI initiative should be pitched see How to Present AI ROI to Your Board and How to Turn Board Pressure for AI ROI Into Your Strategic Advantage as revenue at risk. If 68% of UK shoppers are already using AI agents and your product data isn't structured for them to find, you have a quantifiable share of the algorithmic shelf you're losing today, silently, with no line item to point to until a competitor's category share moves and yours doesn't.
That's the business case: not "we should explore agentic commerce," but "here is the percentage of our category's shopping journeys now starting inside an AI agent, and here is what it costs us to be invisible to it."
Where to start
If you're not sure whether your brand currently shows up accurately to AI shopping agents, that's the first thing to find out — before any data or infrastructure investment. It's a bounded, well-defined problem, which is exactly the kind of engagement a fractional Chief AI Officer is built to scope and run without committing to a permanent hire or a platform contract up front.
Sources: Deloitte — Agentic Commerce in CPG: Winning the Algorithmic Shelf · McKinsey — The Automation Curve in Agentic Commerce · CI&T UK/Ireland shopper research via Retail Technology Innovation Hub · GS1 UK — Web Vocabulary case study