Very Group hands pricing of 200,000 products to agentic AI: when the algorithm sets what you pay

By Drapers · July 15, 2026.

The Very Group, the British online retail group (owner of brands such as Very and Littlewoods), has announced a three-year partnership with UiPath, a platform specializing in agentic AI, to redesign its pricing strategy. According to Sam Wright, the company's chief commercial and customer officer, the catalogue exceeds 200,000 products, and the goal is for AI agents to analyze data and make pricing and merchandising decisions faster and, according to the company, more transparently. Drapers, the outlet reporting the news, has contacted Very Group for more details and has yet to receive a response, which leaves the announcement, for now, in the realm of a joint press release between customer and vendor.

This should be read with the usual caution warranted by such announcements: there are no figures on cost savings, margin impact or implementation timelines, and the primary source is a corporate statement, not an independent audit. But the fact itself is significant beyond the detail: setting prices across a catalogue of 200,000 SKUs has, until now, been the work of pricing analysts, category managers and merchandising teams who cross-reference data on demand, competition, stock and margins. Delegating that process to autonomous agents that decide and execute price changes in real time is exactly the kind of data- and rules-intensive task—repetitive in its logic though complex in volume—that agentic AI attacks first, because the return is immediate and measurable in euros of margin.

This fits with something we have already been flagging about the impact of AI on business administration and retail: the profile that disappears is not the one who thinks up the pricing strategy, but the one who executes it manually, row by row, in a spreadsheet, or reviews competitor alerts one by one. What survives—and probably gains weight—is the analyst who supervises the agent, defines the limits of the pricing policy (by brand, by minimum margin, by brand perception) and corrects course when the system gets it wrong. The promise of 'more transparent decisions' the company mentions is, at bottom, an implicit acknowledgment that today's dynamic-pricing systems are already opaque even to those who operate them; that a company should flag this as a goal says as much about the problem as about the solution.

In the short term, the impact on retail-sector employment is the one we already know from similar sectors: less need for large teams dedicated to routine pricing tasks, more concentration on profiles that combine commercial judgment with oversight of AI systems. It is a real adjustment and should not be downplayed just because the announcement sounds like minor operational efficiency. In the medium term, however, this kind of automation is precisely the brick with which the abundance we champion as our underlying thesis is built: prices that genuinely respond to supply and demand benefit consumers with better deals and less friction, and they free human talent from mechanical tasks toward product design, customer relationships or brand strategy, which is where human judgment remains irreplaceable.

The Very Group case is not a revolution in itself—it is a medium-to-large retailer announcing a multi-year technology deal, something we will see repeated dozens of times this year across different sectors—but it serves as a barometer of where retail is heading: from AI as an analytics tool to AI as a direct executor of commercial decisions. When that happens at scale across the whole industry, the relevant question will no longer be whether prices are set by an AI, but what guardrails (auditing, ethical limits, human oversight) surround that AI so that efficiency does not translate into opacity for the consumer.