McKinsey’s The Agentic Commerce Opportunity report was published in October. It describes what they call a seismic shift in retail. AI agents will stop being chatbots and become autonomous purchasing proxies. They project this “agentic commerce” system will control $3 trillion to $5 trillion in global revenue by 2030.
The vision is compelling. Frictionless. Intent-driven. But the structural reality for a retail chain reveals something else entirely. Significant execution hurdles. Existential risk to your current business model. This is not an opportunity to improve operations. This is a question of survival.
The Evolution of the Consumer Journey
The premise is what McKinsey calls “de-verticalisation”. Your customers currently shop within defined channels. They come to your stores or app when they want what you sell. But AI agents promise to collapse these boundaries. One interface. Intent-based flow. The customer tells their agent what they want. The agent sources it from wherever.
McKinsey identifies three interaction models.
Agent to Site: The agent interacts directly with your existing platforms to find and purchase products.
Agent to Agent (A2A): The customer’s personal agent negotiates directly with your vendor agent for bespoke deals.
Brokered Agent to Site: Intermediary systems facilitate interactions across multiple platforms.
The customer is no longer your customer. The algorithm is. Traditional SEO becomes obsolete. You need “agent discoverability”. Machine-readable product data. Your entire digital infrastructure was built for human eyes. Now you rebuild it for machines that shop on behalf of humans you never see.
Technical Infrastructure: What Actually Has to Work
The report says agentic commerce is “riding on the rails” of previous digital revolutions. It will scale faster than mobile or web because infrastructure already exists. New protocols standardise the ecosystem.
Model Context Protocol (MCP): Allows agents to share context and memory across different models.
Agent Payments Protocol (AP2): An open standard backed by Google and major financial players for verifiable, autonomous purchases.
Agent-to-Agent Protocol (A2A): Lets agents from different companies talk to each other and negotiate deals.
These protocols solve the “human-in-the-loop” problem. Delegated authorisation. Programmable spending policies. Autonomous transactions.
Your core commerce platform was built fifteen years ago. Legacy architecture. Fragmented systems. Integrating these new protocols requires complete re-engineering. Not a plugin. A rebuild. The cost is not trivial. The timeline is not short. Your competitors are already agent-ready.
Business Model Disruption and Monetisation
McKinsey frames this as an “existential choice”. They identify six domains requiring action: customer engagement, loyalty, core platforms, payments, in-store point of service, and fulfilment. Innovate or renovate.
The most immediate threat is advertising revenue. Your retail media network generates significant margin. High-margin digital advertising layered on top of physical retail. But agents bypass search results. They bypass ad-heavy interfaces. The entire retail media model faces obsolescence.
McKinsey proposes alternative monetisation strategies.
Multibrand Bundling: Agents coordinate purchases across providers. You charge a coordination fee.
Real-time Negotiation Fees: You charge for successful upgrades or discount negotiations.
Interagent Protocol Fees: Transactional fees for interoperability between platforms.
These models are theoretically sound. But they require cooperation between competitors. In the UAE retail market, that cooperation does not exist. You compete on margin, on exclusivity, on customer loyalty. Sharing transaction fees with competitors who sell the same products contradicts everything you have built. The alternative revenue models assume a level of industry coordination that may never materialise.
Critical Analysis: The Trust and Risk Paradox
The report’s most sobering section addresses trust and systemic risk. In agentic commerce, trust is no longer between customer and brand. It becomes a “multilayered phenomenon”. “Know Your Agent” protocols.
Several vulnerabilities emerge that merit attention.
Systemic Cascade Failures: A single faulty prompt triggers exponential impact. Mass over-ordering of inventory. Incorrect bookings across your entire network. One error. Catastrophic scale.
The Liability Vacuum: When an agent fails, who is responsible? The model developer? Your company? The customer? There is no global consensus. No legal framework. You are operating in grey space.
Geopolitical Data Sovereignty: Your operations span the UAE, Saudi Arabia, and broader GCC markets. Agents process data across borders. EU data via US-based APIs. This clashes with increasing national demands for data localisation. Regulatory compliance becomes a moving target.
Regional adoption will not be uniform. In markets with low institutional trust or preference for manual control, agentic commerce may stall. Germany. Japan. Parts of the Middle East where personal relationships and direct negotiation remain culturally embedded in commerce. The technology may be ready. The market may not be.
The End Game
McKinsey outlines a future where AI agents become the new gatekeepers of commerce. For your business, the message is unambiguous. The cost of hesitation is displacement. Complete displacement.
Success depends less on agent intelligence and more on establishing liability when systems fail, navigating regulatory landscape, and discovering revenue when advertising dies. You must decide now. Build experiences for people, or for algorithms shopping on their behalf.
Immediate actions:
A full audit of technical infrastructure for agent-readiness. Legal review of liability frameworks across operating markets. Financial modelling of alternative revenue if retail media collapses. Partnership discussions with competitors on protocol standardisation.
You have perhaps eighteen months before early movers gain insurmountable advantage. That timeline is not speculation. That is market reality.
The question is not whether agentic commerce arrives. The question is whether you are still standing when it does.