Deloitte 2026: Why Consumer Goods Companies Must Redesign Retail Execution Around AI Agents

AI adoption is accelerating across the consumer industry, but most companies are still struggling to translate experimentation into scalable business impact.

Deloitte’s State of AI in the Consumer Industry 2026, based on responses from more than 950 consumer-industry business and IT leaders, shows that 73% of consumer companies plan to deploy agentic AI within two years. Yet only 24% currently report at least moderate adoption, and just 20% have a mature governance model for autonomous agents.

For Consumer Goods companies, this gap between ambition and operational readiness is particularly relevant to Retail Execution.

The pilot-to-production gap

According to Deloitte, only 23% of consumer companies have moved at least 40% of their AI experiments into production. Although 52% expect to reach that level within the next three to six months, scaling AI remains fundamentally different from running a successful pilot.

A pilot can operate within one market, banner or category using prepared data and a limited group of users. Enterprise Retail Execution must work across thousands of stores, multiple channels, product categories, market conditions and field teams.

It also needs to integrate with existing Retail Execution systems, operate reliably with real-world shelf data and generate recommendations quickly enough to influence execution while the merchandiser is still in the store.

This is where many promising AI projects stall. Production requires reliable data, enterprise integration, security controls, ongoing monitoring and a clear operating model—not simply an accurate AI model.

Access to AI is not the same as transformation

Deloitte reports that consumer companies have doubled workforce access to sanctioned AI tools in one year, from fewer than 30% to approximately 60%. However, wider access has not necessarily changed how work is performed.

Thirty-seven percent of consumer companies still use AI at a surface level, with little or no change to their underlying business processes. More importantly, 82% have not redesigned jobs around AI capabilities.

This distinction is critical. Giving a merchandiser an AI tool may accelerate an individual task, but it does not transform Retail Execution if the overall workflow remains unchanged.

In the traditional model, the field representative visits the store, collects data and submits a report. A supervisor reviews the information later, identifies an issue and sends a new task back to the field. By that point, the shopper may have already encountered the out-of-stock, missing price tag or incorrect display—and the sales opportunity has been lost.

Redesigning the store visit around AI Agents

Agentic Retail Execution changes the division of work between technology and field teams.

AI Agents can perform virtually every non-physical component of the store visit:

• consolidating store-level requirements and priorities;
• analyzing shelf conditions;
• identifying availability, placement and compliance gaps;
• generating the next-best actions;
• guiding the field representative step by step;
• collecting and structuring execution data;
• validating task completion;
• reporting verified results to supervisors and headquarters.

The field representative remains responsible for the physical actions: replenishing products, correcting placements, installing promotional materials and executing other hands-on merchandising activities.

This human-AI collaboration transforms the store visit from a data-collection exercise into an execution workflow. Shelf conditions are analyzed, actions are generated, physical corrections are completed and results are validated during the same visit.

Skills are only one part of the challenge

Deloitte identifies insufficient workforce skills as the biggest barrier to integrating AI into existing workflows. Consumer companies are responding primarily by educating employees and improving general AI fluency.

Training is necessary, but it does not solve the full problem. Organizations must also redesign workflows so that field employees do not need to interpret complex instructions, study multiple planograms or decide independently which issue to address first.

The AI Agent should bring the required intelligence directly into the workflow: understanding the store context, determining priorities and guiding execution. This reduces the dependency on individual experience while helping teams adopt new activities faster and execute them more consistently.

The role of supervisors must also evolve. Instead of manually reviewing every report and image, they can focus on exceptions, strategic decisions and performance improvement while AI validates routine execution.

Governance must scale with autonomy

As AI Agents take on more responsibility, governance becomes essential. Deloitte found that only one in five consumer companies currently has a mature governance model for autonomous agents.

For Retail Execution, governance should include clear requirements for data access, decision boundaries, validation accuracy, human oversight, system monitoring and accountability. Companies must know which actions an agent can perform autonomously, which require human confirmation and how results are verified.

This is particularly important when AI Agents operate across markets, customers and thousands of daily store visits. Scaling without governance creates risk; governance without integration prevents value. Both must be designed into the operating model from the beginning.

From AI pilots to Agentic Retail Execution

Deloitte’s findings show that Consumer Goods companies are ready to invest in AI Agents, but most have not yet redesigned the processes, roles and governance required to capture their full value.

In Retail Execution, the opportunity extends far beyond automating individual tasks. AI Agents can connect shelf analysis, decision-making, field guidance, physical merchandising and validation within an Instant Closed Loop.

The competitive advantage will belong to companies that move beyond adding AI to the existing process and instead redesign the store visit around human–AI collaboration. That is the point at which AI stops being another field tool and becomes a new operating model for Retail Execution.