What The Very Group signed up for
The Very Group, the UK online retailer behind Very and Littlewoods, has signed a three-year partnership with UiPath to run an agentic AI pricing solution across its retail brands. Announced around 13 July and covered by Retail Times, Retail Gazette and the trade press, the deal applies to a catalog of more than 200,000 products. UiPath's platform layers campaign simulation, scenario planning and AI-generated insight on top of The Very Group's existing data-led pricing approach, with a stated emphasis on explainability. The retailer expects to optimize gross margin and stock management, improve pricing agility, and remove manual steps so pricing staff can spend more time on technical, analytical work.
The executives framed it as a data-and-automation combination rather than a hand-off to a black box. Sam Wright, Chief Customer and Commercial Officer at The Very Group, said pricing is one of the most powerful levers in retail, and that combining the retailer's expertise with UiPath's data helps Very stay ahead of shifting demand to make better decisions on pricing and merchandising. Catherine Frame, Director of Retail Solutions at UiPath, said the partnership shows how agentic AI can fundamentally reshape retail pricing, and that pairing automation with intelligent, explainable decision-making lets retailers drive both profitability and customer value. The three-year term signals this is an operating commitment, not a pilot.
Pricing is where agentic AI earns or loses its keep
Pricing is the ideal proving ground for agentic AI because the feedback loop is short and the money is real. A price change propagates to margin and conversion within hours, so a system that can simulate a campaign, plan scenarios and adjust across 200,000 SKUs faster than a human team has an obvious edge in a competitive market. Manual pricing is slow, inconsistent and expensive to staff, and it leaves margin on the table every day a price sits wrong. That is the gap The Very Group is targeting: pricing agility measured in near-real time rather than weekly review cycles, applied across a catalog too large for any team to touch comprehensively by hand.
The risk is equally concrete. An agent that moves prices autonomously can destroy margin or trigger a public relations problem just as fast as it can protect them, and it can do so at a scale no human error would reach. That is why explainability is the load-bearing word in both companies' statements. Merchandisers and finance need to audit why a price moved, and regulators and customers need confidence the logic is defensible. Agentic pricing without a clear audit trail is a liability that will not survive the first bad week. The Very Group's insistence on explainability, alongside simulation and scenario planning, reads as a governance design choice, not a marketing line.
The governance pattern every retail CTO should copy
For technology leaders, the useful part of this announcement is the shape of the deployment, not the vendor logo. The Very Group keeps humans over the agent at every step. It wraps the agent in simulation, so decisions can be tested before they hit live prices; scenario planning, so leaders can reason about outcomes; and explainability, so every move can be traced. That is a repeatable template for putting agentic AI into any margin-critical or regulated function: give the agent room to decide, but instrument it so humans retain oversight and can intervene. The pattern generalizes well beyond pricing to promotions, markdowns, inventory allocation and fulfillment routing.
It also reframes the build-versus-buy calculus. UiPath brings the agentic automation platform and the pricing tooling; The Very Group brings its proprietary data and commercial expertise, which is the part it cannot outsource without losing its edge. That split is the right one. The differentiation lives in the data and the decision policy, not in the orchestration engine, and a three-year term suggests the retailer sees this as core infrastructure rather than an experiment to be reversed. CTOs evaluating similar moves should press vendors hard on audit trails, rollback speed and how much of the decision logic they can own, because those are the features that determine whether an agentic system stays in production or gets pulled after the first costly mistake.
Why the three-year term is the real signal
Most retail AI headlines describe pilots, and pilots are cheap signals. A three-year contract to run agentic pricing across a full catalog is a different kind of statement. It means The Very Group has done the integration math, satisfied itself on governance, and is willing to build operating processes around the system rather than around it. That is the threshold agentic AI has to cross to matter to a business: moving from a sandbox that impresses a steering committee to a production dependency that finance plans around. The length of the commitment is the tell that this crossed that line.
The broader context is that pricing agility has become a survival trait for online retailers squeezed between marketplace competition and thin margins. Agentic systems that can reprice at scale, simulate before acting and explain themselves afterward are turning into table stakes, and the retailers deploying them now are setting the pace their rivals will have to match. For CTOs, the takeaway runs deeper than a vendor selection: agentic AI has reached the production tier in a core commercial function at a major retailer. The window to treat this as experimental is closing, and the operators who wire explainable agents into their margin engines first will compound the advantage.
UiPath's pivot from RPA is the subtext
UiPath built its business on robotic process automation, the scripted bots that click through legacy software to move data between systems. Agentic pricing is a different animal, and this deal is a public marker of the company's shift from deterministic automation toward AI agents that reason over data and make decisions. For buyers, that history cuts both ways. UiPath brings deep experience wiring automation into messy enterprise environments, which is exactly the integration muscle a pricing rollout across 200,000 SKUs demands. The open question is how mature its agentic decision-making is next to dedicated pricing specialists, and that is a diligence item rather than a claim to accept on faith.
The lesson for retail CTOs is to separate the orchestration layer from the pricing intelligence when they evaluate these deals. An automation platform that can reliably connect to your commerce stack, your ERP and your data warehouse is genuinely valuable and hard to replicate. The pricing logic that decides what a product should cost is a distinct capability, and you should know whether it comes from the platform, from your own models, or from a third party. The Very Group kept its proprietary data and commercial expertise in-house and rented the automation, which is the right division. Ask any vendor to draw that same line clearly before you commit to a multi-year term.
