Ocado Retail Builds Greenfield Salesforce Stack for Agentic AI CX

What happened

Ocado Retail, the UK's dedicated online grocery business, has selected Salesforce as the foundation of a new, purpose-built technology stack — a greenfield rebuild designed to position the retailer for a shift towards agentic artificial intelligence in customer service and operations.

Rather than layering new capability onto legacy infrastructure, Ocado Retail chose to start from scratch, adopting Salesforce's suite of products as its commercial and customer-engagement platform. The move signals a deliberate architectural decision: by eliminating the technical debt of inherited systems, the business aims to create the clean data environment that agentic AI — autonomous, goal-directed AI that can act across systems without constant human prompting — requires to function reliably.

The decision reflects a broader pattern emerging among digitally native retailers: as AI moves from assistive to autonomous, the quality and accessibility of underlying data infrastructure becomes the determining factor in whether AI deployments actually improve customer outcomes or simply add complexity.

Why it matters

Agentic AI represents a meaningful step-change for customer experience design. Unlike conventional chatbots or recommendation engines, agentic systems can initiate multi-step actions — resolving a delivery issue, adjusting a subscription, or proactively flagging a substitution — without waiting for a customer to prompt each move. For that to work without creating friction or error, the underlying data must be unified, current and trustworthy. Ocado Retail's greenfield approach is, in effect, a bet that CX quality in the agentic era is won or lost at the infrastructure layer, long before any customer interaction begins.

From a behavioural economics perspective, the stakes are high. Autonomous AI agents will make consequential micro-decisions on behalf of customers — what to substitute, when to communicate, how to resolve a complaint. Each of those moments is a choice architecture intervention. Getting them right builds trust and reduces cognitive load; getting them wrong, at scale and without human review, can erode loyalty rapidly and in ways that are hard to diagnose after the fact.

The Renascence take

Most commentary on agentic AI focuses on the intelligence layer — the model, the prompt, the capability. Ocado Retail's move is a reminder that the more consequential design decisions happen underneath all of that, in the unglamorous work of data architecture and platform selection.

The organisations that will lead on agentic CX are not necessarily those with the most sophisticated AI models — they are the ones that did the hard, invisible work of cleaning their data house first. Ocado Retail's greenfield commitment is strategically sound precisely because it is operationally painful; most incumbents will not do it. Customer-obsessed operators should treat this as a prompt to audit their own data foundations now, before agentic ambitions collide with legacy reality. The question to ask is not "which AI agent should we deploy?" but "would our current data environment let an autonomous agent make a decision we would be proud of?"