Agentic commerce: the new economy is already here (and it's not a smarter chatbot)

From responding to acting

The market shift underway isn't about smarter chatbots, faster replies, or scripts that sound more natural. It's about AI agents that take autonomous actions across the entire customer lifecycle: discovery, purchase, service, and repurchase.

The difference is simple but structural. A chatbot answers a question and waits. An agent responds, recommends a personalized product, closes the sale, processes the return, manages the reorder, and follows up the next week. It's not an incremental improvement to customer service: it's a reconfiguration of how a business operates.

And the market is moving fast. Some figures from the report:

• US$3–5 trillion in global economic impact from agentic commerce by 2030, according to McKinsey.
• 55% of consumers are shifting product discovery from traditional search engines to AI platforms.
• Gartner projects that by 2028 more than 60% of enterprise customer service interactions will be handled end-to-end by agentic AI (today it's around 20%).
• More than a million businesses already use AI agents every week on Messenger and WhatsApp.

The 75/15 problem: the model isn't your bottleneck

If three-quarters of companies have already adopted agentic AI, why do so few see results? The report is categorical: the constraint isn't the model, it's the infrastructure underneath the model.

40% of companies get stuck in pilots that never scale: agents that generate reports nobody acts on, recommendations a human has to execute manually in another system. That gap between recommendation and execution is exactly where ROI is built… or erodes. Only 12% of CEOs report that AI has delivered both cost and revenue benefits.

The lesson for any business — large or small — is that the value isn't in “having a bot,” but in connecting the agent to real workflows: inventory, orders, payments, scheduling, CRM.

The five layers of the agentic economy

According to Meta, moving from chatbots that respond to agents that act requires five capabilities working together:

1. Identity. Verified identity at scale is the foundation for any commitment an agent makes on behalf of your business.
2. Relationships. Agents need trusted directories to find other agents and businesses.
3. Messaging. With over a billion daily conversations between people and businesses on WhatsApp, Messenger, and Instagram, the best interface for agents is the one people already use — but it only works if it's two-way conversation.
4. Commerce. Catalogs, marketplaces, and checkout within the conversation, with no redirects.
5. Models and protocols. What's missing is an ecosystem of tools and specialized agents to rely on.

Those who already moved

The report documents production cases on the Meta Business Agent Platform, several of them in Latin America:

• Trendyol (retail, EMEA): more than 123,000 customers reached via WhatsApp in a single week; 13% of conversations led to product recommendations presented as shoppable carousels with one-tap checkout.
• Sem Parar (financial services, Brazil): 70% completion rate for PIX payments initiated within the conversation, and 13% conversion from inquiry to payment.
• Movida (automotive, LATAM and EMEA): a 54% increase in conversion rate versus its best historical benchmark (14.9% vs. 9.7%) and 85% of conversations fully resolved by the agent.

The common pattern: the agent doesn't “assist,” it transacts — and it does so within the channel where the customer already lives, which in our region is overwhelmingly WhatsApp.

Where to start

1. Scope and connect. Identify the highest-impact use case and define the path: build the agent natively or integrate an existing one.
2. Configure and integrate. Ground the agent in business knowledge, define tone of voice and escalation rules, and connect it to your systems.
3. Launch and measure. Start with the highest-volume conversations and measure cost per outcome, average ticket, and customer satisfaction.

The advantage of moving early

Every conversation your agent holds generates context that makes the next one more relevant — a personalization advantage a competitor can't buy later. As Meta's Chief Data Officer Alex Schultz puts it: “agents are at their dumbest today”; whoever builds for what agents do today is building for the past.

Based on Meta's report “Beyond Chatbots: The Agentic Economy Is Here” (2026), citing data from Forrester, Gartner, McKinsey, Deloitte, and PwC.